{
"meta": {
"title": "Agent governance playbook",
"edition": "Public edition",
"version": "3.3.0",
"published": "2026-11-18",
"nextReview": "2027-02-18",
"licence": "Open Government Licence v3.0",
"provenance": "Ported from an internal AI agent governance toolkit on 18 November 2026. Each chapter was rendered, classified as public or withheld, redacted where needed and rebuilt as plain semantic content. Figures the source labels illustrative stay labelled illustrative; nothing here is an official policy position.",
"classification": "docs/content-classification-2026-11-18.md"
},
"parts": [
{
"id": "start-here",
"name": "Start here",
"colour": "#1b41d8",
"chapters": [
"overview",
"raistart",
"howto",
"playbook",
"domains"
]
},
{
"id": "decide-and-prioritise",
"name": "Decide and prioritise",
"colour": "#a2451f",
"chapters": [
"evidence",
"usecases",
"raicases",
"raiassess",
"criticality",
"maturity"
]
},
{
"id": "design-and-build",
"name": "Design and build",
"colour": "#00703c",
"chapters": [
"railifecycle",
"capclasses",
"skills",
"raiagents",
"raiconnectors",
"architecture",
"knowledge",
"autonomy"
]
},
{
"id": "assure-and-govern",
"name": "Assure and govern",
"colour": "#0f766e",
"chapters": [
"raicontrols",
"raidata",
"trust",
"govmodel",
"risks",
"registers",
"atrs",
"recovery"
]
},
{
"id": "operate-and-oversee",
"name": "Operate and oversee",
"colour": "#912b88",
"chapters": [
"tower",
"raioperate",
"agentjourney",
"employment",
"cascade",
"observatory"
]
},
{
"id": "fund-and-buy",
"name": "Fund and buy",
"colour": "#003078",
"chapters": [
"value",
"finops",
"commercial",
"dependency"
]
},
{
"id": "people-and-adoption",
"name": "People and adoption",
"colour": "#17756e",
"chapters": [
"teams",
"serviceteam",
"roles",
"academy",
"capability",
"inclusion",
"journeys",
"rollout",
"community",
"operating",
"raiadoption"
]
},
{
"id": "live-pilot",
"name": "Live pilot",
"colour": "#6f2dbd",
"chapters": [
"pilotview",
"pilottest",
"pilotselect",
"pilotroles",
"pilotassurance",
"pilotjourney",
"pilotcomms",
"pilotsuccess",
"pilotdecision"
]
},
{
"id": "product-documentation",
"name": "Product documentation",
"colour": "#5b4bd6",
"chapters": [
"docprd",
"docftir",
"docsad",
"dochld",
"doclld",
"doctoc",
"docresearch",
"docresources"
]
},
{
"id": "knowledge-and-navigation",
"name": "Knowledge and navigation",
"colour": "#a4231a",
"chapters": [
"thesaurus",
"taxonomy",
"stack",
"federated"
]
},
{
"id": "deploy-for-real",
"name": "Deploy for real",
"colour": "#b8005a",
"chapters": [
"atlas",
"casegen",
"worked",
"boundary"
]
},
{
"id": "reference",
"name": "Reference",
"colour": "#712b8b",
"chapters": [
"nations",
"world",
"pace"
]
}
],
"chapters": [
{
"id": "overview",
"part": "start-here",
"title": "Overview",
"icon": "grid_view",
"eyebrow": null,
"heading": "AI Capability Governance and Implementation",
"lede": "The guidance for deploying AI agents in the public sector, sequenced into one route — and the operating capability that turns experimentation into governed services. Crown-copyright sources are the government's own; every illustrative figure is labelled.",
"body": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "icon",
"v": "home"
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Start here"
},
{
"t": "p",
"v": "Eight stages from mandate to governed operation. The last stage feeds the third: every review either re-authorises the agent, changes it, or ends it."
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "loop"
},
{
"t": "h",
"l": 3,
"v": "Stage 8 returns to stage 3 — the loop does not close"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Usage and cost** Who uses it, how often, what it spends"
},
{
"t": "p",
"v": "**Accuracy and drift** Evaluation results against the baseline"
},
{
"t": "p",
"v": "**Risk and incidents** New risks, near misses, escalations"
},
{
"t": "p",
"v": "**Value realised** Benefits claimed against benefits evidenced"
},
{
"t": "p",
"v": "**Review or retire** Re-authorise, improve, restrict or decommission"
}
],
"box": false
},
{
"t": "title",
"v": "Monitoring and human oversight are continuous duties held by the named owner, not a phase that ends at go-live.",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Where this guide sits"
},
{
"t": "p",
"v": "The hierarchy of AI-related guidance a public-sector team works inside. This guide is a proposal: it interprets and sequences the guidance above it, and it does not replace or override any of it."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "International frameworks"
},
{
"t": "p",
"v": "OECD AI Principles, EU AI Act risk tiers, ISO/IEC 42001 management standards."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "UK statute and cross-government policy"
},
{
"t": "p",
"v": "Data Protection Act 2018 and UK GDPR, Equality Act 2010, Procurement Act 2023, the AI Opportunities Action Plan."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Central guidance in force"
},
{
"t": "p",
"v": "AI Playbook for the UK Government, Service Standard, Technology Code of Practice, ATRS, Data Ethics Framework, NCSC secure-AI guidelines."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Departmental policy and standards"
},
{
"t": "p",
"v": "Organisational AI policy, data classification, security standards, delegated authority and spend controls."
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "This implementation guide"
},
{
"t": "p",
"v": "AI Capability Governance and Implementation — 69 chapters, assessment tools and register schemas that turn the guidance above into delivery."
}
]
},
{
"t": "eyebrow",
"v": "Proposal"
}
],
"box": true
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Team-level artefacts"
},
{
"t": "p",
"v": "Agent register entries, ATRS records, assurance evidence, instruction blocks and cost models."
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "groups"
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "Multidisciplinary from the outset"
},
{
"t": "p",
"v": "Every source in this app lands on the same point. The AI Playbook makes skills and expertise a principle in its own right. The Service Manual builds the multidisciplinary team into the Service Standard. NCSC wrote its secure-AI guidelines for data scientists, developers, managers, decision-makers and risk owners together. The roster is in **Teams**; the readiness barometer is in the **Observatory**."
}
]
}
],
"box": true
}
]
},
{
"id": "raistart",
"part": "start-here",
"title": "Can I use AI for this?",
"icon": "help",
"eyebrow": "Responsible AI · start here",
"heading": "What do you want to do?",
"lede": "Start from your task. This part of the playbook routes you to the proportionate tool, template, assessment or support. Governance follows what the AI can see, create and do, not which company made it.",
"body": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Quick route selector · five questions"
},
{
"t": "h",
"l": 2,
"v": "Can I use AI for this?"
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "1. Would the output decide, or directly change, someone’s rights, entitlements, safety or access to a service?",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "2. Will you use personal data or information marked OFFICIAL-SENSITIVE or above?",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "3. Can the AI take actions, run on a schedule or work through several steps without you?",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "4. Is there an approved tool in your organisation's catalogue that already does this?",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "5. Could a conventional rule, form or automation do the job reliably?",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Your route"
},
{
"t": "title",
"v": "Answer the five questions",
"big": false
},
{
"t": "p",
"v": "The route appears when every question has an answer. “Not sure” is a valid answer."
},
{
"t": "p",
"v": "A first sort, not an approval. Any route other than an approved tool leads to a registered use case."
}
]
}
],
"box": true
},
{
"t": "h",
"l": 2,
"v": "Six things to know first"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "AI concepts and limitations",
"big": false
},
{
"t": "p",
"v": "Generative AI predicts plausible output. It can be fluent and wrong at once, so outputs need checking in proportion to what depends on them."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Models and providers",
"big": false
},
{
"t": "p",
"v": "A product can use several models from several providers, sometimes chosen automatically. The model can change without the product name changing."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Enterprise grounding",
"big": false
},
{
"t": "p",
"v": "Grounding feeds organisational content to the model. Whatever the index can reach, the AI can surface, including titles, filenames and metadata."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Agents, tools and actions",
"big": false
},
{
"t": "p",
"v": "An agent plans steps and uses tools to act: send, update, create or call other agents. Acting changes the risk boundary."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Human oversight",
"big": false
},
{
"t": "p",
"v": "Oversight works only if the reviewer has the competence, the evidence and the authority to challenge, override or stop."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Deployment route",
"big": false
},
{
"t": "p",
"v": "The same model reached directly, through a cloud platform or inside another supplier’s product is a different governance object each time."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "For everyone in the organisation"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "p",
"v": "**All colleagues**Find guidance, approved tools and practical help"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "person"
},
{
"t": "p",
"v": "**Leaders and managers**Make informed decisions, manage risk and value"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "shield"
},
{
"t": "p",
"v": "**Risk, security and compliance**Assure, monitor and advise"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "settings"
},
{
"t": "p",
"v": "**IT and digital**Enable, integrate and manage"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "database"
},
{
"t": "p",
"v": "**Data and information**Ensure data is ready and well managed"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "payments"
},
{
"t": "p",
"v": "**Commercial and procurement**Manage suppliers, licences and costs"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "accessibility_new"
},
{
"t": "p",
"v": "**Accessibility and inclusion**Support all users and workplace adjustments"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "insights"
},
{
"t": "p",
"v": "**Policy, strategy and transformation**Drive adoption, value and continuous improvement"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Enabled by common services"
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Real-world outcomes · turning responsible AI into a stronger public sector"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "verified_user"
},
{
"t": "title",
"v": "Safer services for citizens",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "star"
},
{
"t": "title",
"v": "Greater value for taxpayers",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "bar_chart"
},
{
"t": "title",
"v": "More efficient working",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "diversity_3"
},
{
"t": "title",
"v": "More inclusive for all",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lightbulb"
},
{
"t": "title",
"v": "A more innovative public sector",
"big": false
}
]
}
],
"box": false
},
{
"t": "title",
"v": "Ideas → responsible implementation → trusted operation → lasting public value",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Govern the capability and behaviour first"
},
{
"t": "p",
"v": "Microsoft Copilot, Anthropic Claude, GitHub Copilot, Gemini and future tools sit as product overlays beneath one common control model. Role routes for every audience are in How to use this guide."
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Get help"
},
{
"t": "title",
"v": "Using an approved tool",
"big": false
},
{
"t": "p",
"v": "Your IT service desk, then the product’s support route."
},
{
"t": "title",
"v": "Assessing a use case",
"big": false
},
{
"t": "small",
"v": "The AI assurance team owns screening and the decision register."
},
{
"t": "title",
"v": "An incident or concern",
"big": false
},
{
"t": "p",
"v": "Report through the incident route on the Operate page. Treat data exposure as a security incident."
},
{
"t": "title",
"v": "Accessibility or adjustments",
"big": false
},
{
"t": "small",
"v": "Your line manager and workplace adjustments service."
},
{
"t": "p",
"v": "Named contacts are set by each organisation when the guidance is deployed."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Standards and policy"
},
{
"t": "title",
"v": "AI policy",
"big": false
},
{
"t": "small",
"v": "Your organisation’s AI policy and acceptable use."
},
{
"t": "title",
"v": "Government guidance",
"big": false
},
{
"t": "small",
"v": "The AI Playbook for the UK Government."
},
{
"t": "title",
"v": "AI Playbook on GOV.UK ↗",
"big": false
},
{
"t": "title",
"v": "Data protection",
"big": false
},
{
"t": "small",
"v": "UK GDPR, DPIAs and international transfers."
},
{
"t": "title",
"v": "ICO guidance on AI ↗",
"big": false
},
{
"t": "title",
"v": "Security",
"big": false
},
{
"t": "small",
"v": "Threat modelling, secure configuration and incident response."
},
{
"t": "title",
"v": "Records",
"big": false
},
{
"t": "small",
"v": "Retention and disposal of prompts, outputs, transcripts and logs."
},
{
"t": "title",
"v": "Accessibility",
"big": false
},
{
"t": "small",
"v": "WCAG 2.2 AA for AI interfaces and outputs."
},
{
"t": "title",
"v": "WCAG 2.2 ↗",
"big": false
},
{
"t": "title",
"v": "Procurement",
"big": false
},
{
"t": "small",
"v": "Supplier assurance, subprocessors, exit and service commitments."
},
{
"t": "title",
"v": "Transparency",
"big": false
},
{
"t": "small",
"v": "Algorithmic Transparency Recording Standard for public-facing tools."
},
{
"t": "title",
"v": "ATRS hub on GOV.UK ↗",
"big": false
},
{
"t": "title",
"v": "Templates",
"big": false
},
{
"t": "small",
"v": "Twelve core artefacts and seven implementation artefacts."
},
{
"t": "title",
"v": "Product guidance",
"big": false
},
{
"t": "small",
"v": "Overlays for each supplier and deployment route."
}
]
}
],
"box": false
}
]
},
{
"id": "howto",
"part": "start-here",
"title": "How to use this guide",
"icon": "explore",
"eyebrow": "Start here · orientation",
"heading": "How to use this guide",
"lede": "87 chapters in 12 parts. Nobody needs all of it. Pick a reading mode, then a route.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "menu_book"
},
{
"t": "h",
"l": 2,
"v": "Front to back"
},
{
"t": "p",
"v": "12 parts in delivery order: decide, build, assure, operate, fund, the people and pilot material, then the deployment part and the reference. Roughly a day's read."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "badge"
},
{
"t": "h",
"l": 2,
"v": "By role route"
},
{
"t": "p",
"v": "10 routes below, each an ordered set of chapters for one job. Twenty to forty minutes."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "bolt"
},
{
"t": "h",
"l": 2,
"v": "By task"
},
{
"t": "p",
"v": "Use the launchpad or search: rate readiness, tier an agent, price a job, build an instruction block, check exit readiness."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Routes by role"
},
{
"t": "p",
"v": "Each route is an ordered set of chapters. Click any step to jump there."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Data and information"
},
{
"t": "p",
"v": "Get data ready, keep metadata and indexing honest, govern connectors and treat AI outputs as records."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "All colleagues"
},
{
"t": "p",
"v": "Find an approved tool, get a route for your task, and know where to get help."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Accountable leader or SRO"
},
{
"t": "p",
"v": "Decide where AI is worth the risk and the spend, and be able to answer for it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Service owner"
},
{
"t": "small",
"v": "Adopt agents without destabilising a service people depend on."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Product or delivery manager"
},
{
"t": "small",
"v": "Frame the problem, prove the value, keep the team moving."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI engineer or technical owner"
},
{
"t": "small",
"v": "Build well-grounded agents and prove they work."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance, ethics or audit"
},
{
"t": "small",
"v": "Apply proportionate control and see the evidence behind it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Commercial, finance or FinOps"
},
{
"t": "small",
"v": "Fund adoption honestly and keep exit possible."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "People, learning and adoption"
},
{
"t": "small",
"v": "Build capability that outlasts the pilot."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Pilot participant"
},
{
"t": "small",
"v": "Use the guide during a real deployment and record what happens."
}
]
},
{
"t": "h",
"l": 2,
"v": "What each part is for"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "1 · Start here",
"big": false
},
{
"t": "p",
"v": "What the guidance says, what this guide adds, and the ten capability domains it all rests on."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "2 · Decide and prioritise",
"big": false
},
{
"t": "p",
"v": "Whether to do it at all: evidence quality, use-case selection, criticality and the assurance tier that follows."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "3 · Design and build",
"big": false
},
{
"t": "p",
"v": "Capability class — generative, agentic or declarative — then instructions, architecture, grounding and the autonomy boundary."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "4 · Assure and govern",
"big": false
},
{
"t": "p",
"v": "The trust cycle, the operating model, risks and mitigations, registers, transparency records and recovery."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "5 · Operate and oversee",
"big": false
},
{
"t": "p",
"v": "Life after approval: the portfolio view, the agent lifecycle, public-service impact and readiness tracking."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "6 · Fund and buy",
"big": false
},
{
"t": "p",
"v": "The full investment line, unit economics, supplier requirements and exit."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "7 · People and adoption",
"big": false
},
{
"t": "p",
"v": "Teams, roles, pathways, inclusion and the organisational adoption programme."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "8 · Live pilot",
"big": false
},
{
"t": "p",
"v": "How this guide is being tested in a real deployment, and what would justify a permanent toolkit."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "9 · Product documentation",
"big": false
},
{
"t": "p",
"v": "Requirements, architecture, design, theory of change and the research base for the guide itself."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "10 · Knowledge and navigation",
"big": false
},
{
"t": "p",
"v": "Glossary, thesaurus, generated knowledge graph, taxonomies, ontology, site map, guided tours and platform links."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "11 · Deploy for real",
"big": false
},
{
"t": "p",
"v": "The part that turns the guide into a deployment: fifty-four real services placed on volume and criticality, a generator that derives the whole governance set for any case, four agents carried end to end — one in each quadrant — and an explicit list of what this guide still does not cover."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "12 · Reference",
"big": false
},
{
"t": "p",
"v": "Devolved and international comparators, and every source behind the content."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "How to read the labels"
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Verified"
},
{
"t": "small",
"v": "Present in a surfaced source."
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Interpretation"
},
{
"t": "small",
"v": "A conclusion drawn across sources."
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Proposal"
},
{
"t": "small",
"v": "Recommended future state, not adopted policy."
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Illustrative"
},
{
"t": "small",
"v": "Looks like data, carries no measured source."
}
]
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "What this guide is not"
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "info"
},
{
"t": "p",
"v": "It does not override departmental policy, professional advice or formal approvals."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "info"
},
{
"t": "small",
"v": "It is not the register of record for any live deployment."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "info"
},
{
"t": "p",
"v": "It holds no personal, case or OFFICIAL-SENSITIVE data; assessments stay on your device."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "info"
},
{
"t": "small",
"v": "It cannot issue an approval — the accountable owner does that."
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "playbook",
"part": "start-here",
"title": "The playbook",
"icon": "library_books",
"eyebrow": "The playbook",
"heading": "Ten principles, one life cycle",
"lede": "The [[https://www.gov.uk/government/publications/ai-playbook-for-the-uk-government|AI Playbook for the UK Government]] (GDS, 10 February 2025) supersedes the Generative AI Framework and covers all AI: machine learning, NLP, computer vision, generative and agentic systems. The ten principles are paraphrased below; open each for its first moves.",
"body": [
{
"t": "h",
"l": 2,
"v": "The deployment life cycle"
},
{
"t": "p",
"v": "Playbook principle 5: manage the whole life cycle, including the unglamorous parts. Each stage has a gate; nothing moves right without passing it."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Stage 1"
},
{
"t": "h",
"l": 3,
"v": "Discover"
},
{
"t": "small",
"v": "Frame the problem and the user need: is AI the right tool at all?"
},
{
"t": "small",
"v": "Triage risk: personal data, decisions about people, public-facing?"
},
{
"t": "p",
"v": "Assemble the multidisciplinary team; screen for DPIA and equality impacts"
},
{
"t": "card",
"kids": [
{
"t": "small",
"v": "**Gate · **A named SRO and a written problem statement"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Stage 2"
},
{
"t": "h",
"l": 3,
"v": "Design"
},
{
"t": "small",
"v": "Decide buy vs build; source and assess the data"
},
{
"t": "small",
"v": "Threat-model with security; place the human-oversight points"
},
{
"t": "p",
"v": "Legal, ethics and commercial shape the approach, not the finished thing"
},
{
"t": "card",
"kids": [
{
"t": "small",
"v": "**Gate · **DPIA drafted; risks owned; oversight designed in"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Stage 3"
},
{
"t": "h",
"l": 3,
"v": "Build & evaluate"
},
{
"t": "small",
"v": "Develop securely; verify the model and data supply chain"
},
{
"t": "small",
"v": "Evaluate accuracy, bias and robustness against agreed baselines"
},
{
"t": "small",
"v": "Red-team it: prompt injection, misuse, edge cases"
},
{
"t": "card",
"kids": [
{
"t": "small",
"v": "**Gate · **Evaluation results reviewed by the whole team"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Stage 4"
},
{
"t": "h",
"l": 3,
"v": "Deploy"
},
{
"t": "small",
"v": "Pass assurance gates; publish the ATRS record"
},
{
"t": "small",
"v": "Switch on human review and incident response"
},
{
"t": "small",
"v": "Tell users they are interacting with AI"
},
{
"t": "card",
"kids": [
{
"t": "small",
"v": "**Gate · **Assurance sign-off; transparency record live"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Stage 5"
},
{
"t": "h",
"l": 3,
"v": "Operate & retire"
},
{
"t": "small",
"v": "Monitor drift, bias and hallucination in production"
},
{
"t": "small",
"v": "Re-evaluate on every model or data update"
},
{
"t": "small",
"v": "Plan decommissioning and knowledge handover"
},
{
"t": "card",
"kids": [
{
"t": "small",
"v": "**Gate · **Monitoring dashboards live; exit plan on file"
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "domains",
"part": "start-here",
"title": "Capability domains",
"icon": "account_tree",
"eyebrow": "The domains",
"heading": "Five lenses on every deployment",
"lede": "Cyber, data, digital, technology, and the legal-ethical frame around them. Each domain has its own authoritative guidance; a deployment is ready only when all five say so. That is why the team is multidisciplinary.",
"body": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Cyber"
},
{
"t": "h",
"l": 2,
"v": "Cyber security"
},
{
"t": "title",
"v": "NCSC secure AI guidelines",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "NCSC machine learning principles",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "Secure by Design (UK Government)",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "NCSC Guidelines for secure AI system development cover secure design, development, deployment and operation. Agreed with CISA and 21 international agencies; aimed at any system that uses AI, built or bought."
},
{
"t": "p",
"v": "Secure by Design and the Government Cyber Security Standard apply to AI services like any other, plus AI-specific threat modelling: prompt injection, data poisoning, model extraction, over-permissioned agents."
},
{
"t": "p",
"v": "NCSC's principles for the security of machine learning translate the threats into controls. Walk before you run with agentic AI."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Data"
},
{
"t": "h",
"l": 2,
"v": "Data & information"
},
{
"t": "title",
"v": "Data Ethics Framework",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "ICO: AI and data protection",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "ATRS hub",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "The Data Ethics Framework guides appropriate, responsible data use across the public sector: transparency, accountability, fairness."
},
{
"t": "p",
"v": "UK GDPR and the DPA 2018 require a DPIA before high-risk processing. The ICO's AI guidance, with a statutory AI code of practice in development, is the regulator's view of what good looks like. Cross-border processing (for example a US-processed model behind a UK product) needs a Transfer Risk Assessment as well as SCCs."
},
{
"t": "p",
"v": "The Algorithmic Transparency Recording Standard is mandatory for central departments and arms-length bodies: a published record for every in-scope algorithmic tool."
},
{
"t": "p",
"v": "Model quality is data quality: govern provenance, representativeness and retention for training and grounding data alike — retrieval systems inherit the permissions, quality and currency of the estate."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Digital"
},
{
"t": "h",
"l": 2,
"v": "Digital & service"
},
{
"t": "title",
"v": "Service Standard",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "AI Playbook for the UK Government",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "AI Knowledge Hub",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "The Service Standard and Service Manual set the operating model: start with user needs, work in multidisciplinary teams, iterate in the open. AI services are still services."
},
{
"t": "p",
"v": "The AI Playbook and the AI Knowledge Hub supply the patterns, case studies and communities; the PM's AI Exemplars programme is the test-and-learn engine for scaling what works."
},
{
"t": "p",
"v": "Accessibility is law, not polish: WCAG 2.2 AA under the Public Sector Bodies Accessibility Regulations 2018. AI interfaces (chat, voice, summaries) need their own accessibility evidence."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Technology"
},
{
"t": "h",
"l": 2,
"v": "Technology & architecture"
},
{
"t": "title",
"v": "Technology Code of Practice",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "AI Opportunities Action Plan",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "Action Plan delivery progress",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "The Technology Code of Practice governs the technology choices: cloud first, interoperability, open standards, sustainability, exit."
},
{
"t": "p",
"v": "Compute and infrastructure are national policy: AI Growth Zones and a compute roadmap from 21 ExaFLOPs (2025) towards 420 by 2030 under the AI Opportunities Action Plan."
},
{
"t": "p",
"v": "Buy well: use Crown Commercial Service AI routes, insist on evaluation access and model-change notice in contracts, and design the exit before signing the entry."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Legal & ethics"
},
{
"t": "h",
"l": 2,
"v": "Legal, ethics & society"
},
{
"t": "title",
"v": "Explaining decisions made with AI",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "Data Ethics Framework",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
},
{
"t": "title",
"v": "Equality Act 2010 guidance",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "The Equality Act 2010 and the Public Sector Equality Duty apply to automated and AI-assisted decisions. Public-law duties (fairness, rationality, giving reasons) still bind when a model is in the loop."
},
{
"t": "p",
"v": "Explainability has a manual: 'Explaining decisions made with AI', by the ICO and The Alan Turing Institute — and explanations shown to people should reference functional needs, never a diagnosis or label."
},
{
"t": "p",
"v": "Public trust is earned in the open: engage affected people early, say clearly where AI is used, and give a route to challenge and redress."
}
]
}
],
"box": true
}
]
},
{
"id": "evidence",
"part": "decide-and-prioritise",
"title": "Evidence before assertion",
"icon": "fact_check",
"eyebrow": "The evidence",
"heading": "Same product. Four UK evaluations. Four answers.",
"lede": "What varies is method, not the tool. Uncontrolled self-report gives the highest figures; comparison-group econometrics gives mid-range; counterfactual-free evaluation of a short pilot gives the lowest. Role mix is the second driver.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "26"
},
{
"t": "eyebrow",
"v": "minutes / day"
},
{
"t": "title",
"v": "Cross-government, GDS",
"big": false
},
{
"t": "small",
"v": "20,000 staff, self-reported"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "60"
},
{
"t": "eyebrow",
"v": "minutes / week"
},
{
"t": "title",
"v": "HMRC, randomised cohort",
"big": false
},
{
"t": "small",
"v": "3,000 licences"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "19"
},
{
"t": "eyebrow",
"v": "minutes / day"
},
{
"t": "title",
"v": "DWP, comparison group",
"big": false
},
{
"t": "small",
"v": "3,549 staff, econometric"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "0"
},
{
"t": "eyebrow",
"v": "robust evidence"
},
{
"t": "title",
"v": "DBT, no counterfactual",
"big": false
},
{
"t": "small",
"v": "1,000 licences"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "**Never average these numbers.** The spread is evidence about evaluation design. Sources: GDS findings report (Jun 2025); HMRC phase 3; DWP evaluation (Jan 2026); DBT evaluation (Aug 2025)."
},
{
"t": "h",
"l": 2,
"v": "Four things block the move to governed scale"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Unremediated permissions",
"big": false
},
{
"t": "p",
"v": "AI surfaces what a user may already reach. Permission drift becomes an incident on day one."
},
{
"t": "eyebrow",
"v": "Makes scaling UNSAFE"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "No agent inventory",
"big": false
},
{
"t": "p",
"v": "The transparency standard records algorithmic tools. It is not an operational register of agents."
},
{
"t": "eyebrow",
"v": "Makes scaling UNGOVERNABLE"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Unpredictable consumption cost",
"big": false
},
{
"t": "p",
"v": "Credits are billed per feature, not per conversation. A builder's design choice sets a budget holder's cost."
},
{
"t": "eyebrow",
"v": "Makes scaling UNBUDGETABLE"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Unfunded human effort",
"big": false
},
{
"t": "p",
"v": "Workflow redesign, local support and output verification happen but rarely appear as funded lines."
},
{
"t": "eyebrow",
"v": "Makes scaling UNSUSTAINABLE"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Activity is running ahead of visibility"
},
{
"t": "p",
"v": "Between deployment and disclosure sits a gap no current UK instrument closes. The ATRS is a transparency mechanism, not an inventory: it does not capture team-shared agents, orphaned agents, data-source approval, consumption cost or retirement."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "21%"
},
{
"t": "small",
"v": "of bodies responding to the NAO had an AI strategy"
},
{
"t": "eyebrow",
"v": "NAO · Mar 2024"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "74"
},
{
"t": "small",
"v": "use cases deployed, reported at the same date"
},
{
"t": "eyebrow",
"v": "NAO · Mar 2024"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "33 → 125",
"big": true
},
{
"t": "small",
"v": "transparency records: Jan 2025 → end-2025"
},
{
"t": "eyebrow",
"v": "PAC · OECD.AI"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "28%"
},
{
"t": "small",
"v": "of central government IT is legacy"
},
{
"t": "eyebrow",
"v": "PAC · Mar 2025"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Coding agents: two findings, both real"
},
{
"t": "small",
"v": "A finding about where to deploy, not whether."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "1 hr",
"big": false
},
{
"t": "small",
"v": "saved per day — GDS cross-government trial, 50 organisations"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "72%",
"big": false
},
{
"t": "small",
"v": "said good value; 58% would not return to working without it"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "19%",
"big": false
},
{
"t": "small",
"v": "slower — METR randomised trial: experienced developers, mature code"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "20%",
"big": false
},
{
"t": "small",
"v": "felt faster — the same developers, in the same trial"
}
]
},
{
"t": "p",
"v": "Benefit is largest on well-scoped, greenfield and boilerplate work; smallest or negative on mature, complex codebases. Developer self-assessment is unreliable in both directions. Join the AI Engineering Lab (OCTO, GDS) before procuring independently — the evidence you generate is reusable."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "What the enthusiastic version leaves out"
},
{
"t": "small",
"v": "Contrary evidence, recorded deliberately."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "17%",
"big": false
},
{
"t": "p",
"v": "of cross-government trial users noticed no clear time savings (GDS, Jun 2025)"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "7%",
"big": false
},
{
"t": "p",
"v": "of Australian participants reported the tool added time (DTA, Oct 2024)"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "61%",
"big": false
},
{
"t": "p",
"v": "of Australian managers could not confidently identify AI outputs (DTA, Oct 2024)"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "72%",
"big": false
},
{
"t": "p",
"v": "of councils were not ready on workforce skills and expertise (LGA, Jun 2025)"
}
],
"box": false
},
{
"t": "p",
"v": "Canada names a further risk no UK instrument addresses: skill atrophy and automation drift. Skill maintenance is a prerequisite for meaningful human oversight, not an optional extra (TBS Canada, May 2026)."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "A caution on AI-produced research"
},
{
"t": "small",
"v": "Twenty sources in one AI research draft. What checking found:"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "7"
},
{
"t": "small",
"v": "verified precisely"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "6"
},
{
"t": "small",
"v": "verified in substance"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "4"
},
{
"t": "small",
"v": "not verified as cited"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "1"
},
{
"t": "small",
"v": "could not be located"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "None of the unverifiable entries was obvious: each was correctly formatted, plausibly dated, attributed to a real institution. Treat AI-produced research as a hypothesis-generating artefact, never an evidence base — and keep the verification log with the output."
}
]
},
{
"t": "h",
"l": 2,
"v": "Evidence labels used across this app"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Verified fact** — supported by official, regulatory, audit or primary sources"
},
{
"t": "p",
"v": "**Organisation-reported** — reported by the organisation itself, not independently measured"
},
{
"t": "p",
"v": "**Supplier claim** — authoritative for product terms; interested-party evidence for value"
},
{
"t": "small",
"v": "**Interpretation** — analytical judgement drawn from cited evidence"
},
{
"t": "small",
"v": "**Recommendation** — action proposed, not established practice"
},
{
"t": "small",
"v": "**Unresolved** — evidence incomplete, conflicting or arrangement-specific"
}
],
"box": false
}
]
},
{
"id": "usecases",
"part": "decide-and-prioritise",
"title": "Use cases & evidence",
"icon": "apps",
"eyebrow": "Use cases & evidence",
"heading": "Where AI creates value across government and public services",
"lede": "Eighteen categories, from productivity tools to enterprise agents — supporting people, processes and public outcomes. Use AI and agents responsibly to enhance the work of public servants and improve outcomes for citizens.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Knowledge retrieval & insight** Find, summarise and synthesise information from trusted sources"
},
{
"t": "p",
"v": "**2. Content & correspondence** Draft, edit and refine high-quality content and communications"
},
{
"t": "p",
"v": "**3. Contact centres & helpdesks** Support staff and citizens with accurate, consistent responses"
},
{
"t": "p",
"v": "**4. Casework & service operations** Streamline case handling, triage and workflow automation"
},
{
"t": "p",
"v": "**5. Citizen services & engagement** Deliver simpler, faster and more accessible services"
},
{
"t": "p",
"v": "**6. Meetings & collaboration** Capture, summarise and turn meetings into actions and outcomes"
},
{
"t": "p",
"v": "**7. Data analysis & reporting** Analyse data, create summaries and produce management information"
},
{
"t": "p",
"v": "**8. Software engineering & IT operations** Accelerate development, testing and secure operations"
},
{
"t": "p",
"v": "**9. HR & people management** Improve people processes and the employee experience"
},
{
"t": "p",
"v": "**10. Finance & procurement** Enhance financial planning, analysis and supplier management"
},
{
"t": "p",
"v": "**11. Education, training & skills** Personalise learning and improve knowledge retention"
},
{
"t": "p",
"v": "**12. Programme & project management** Plan, track and report with intelligent support"
},
{
"t": "p",
"v": "**13. Health & care** Support clinical and administrative workflows safely and securely"
},
{
"t": "p",
"v": "**14. Policing, justice & public safety** Improve investigations, evidence handling and public protection"
},
{
"t": "p",
"v": "**15. Regulatory & compliance** Monitor compliance and produce regulatory outputs"
},
{
"t": "p",
"v": "**16. Policy & strategy development** Scenario analysis, option generation and evidence synthesis"
},
{
"t": "p",
"v": "**17. Communications & media** Create campaigns, content and prepare responses"
},
{
"t": "p",
"v": "**18. Research & horizon scanning** Scan trends, research and emerging opportunities"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cross-cutting capabilities — enablers for all use cases"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Secure access to trusted data & knowledge",
"big": false
},
{
"t": "title",
"v": "AI agents & orchestration (autonomy with oversight)",
"big": false
},
{
"t": "title",
"v": "Human oversight & accountability",
"big": false
},
{
"t": "title",
"v": "Privacy, security & data protection",
"big": false
},
{
"t": "title",
"v": "Quality assurance & model evaluation",
"big": false
},
{
"t": "title",
"v": "Observability, monitoring & auditability",
"big": false
},
{
"t": "title",
"v": "Integration with systems & workflows",
"big": false
},
{
"t": "title",
"v": "Change management & adoption",
"big": false
},
{
"t": "title",
"v": "FinOps & cost management",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Foundational enablers"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Data quality & governance",
"big": false
},
{
"t": "title",
"v": "Information management",
"big": false
},
{
"t": "title",
"v": "Access management",
"big": false
},
{
"t": "title",
"v": "Standards & interoperability",
"big": false
},
{
"t": "title",
"v": "Skills & culture",
"big": false
},
{
"t": "title",
"v": "Ethics & responsible AI",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The use-case evidence card"
},
{
"t": "p",
"v": "A consistent way to describe, evaluate and share AI use-case evidence: what the capability does, who it helps, how it works, the value it delivers, and the evidence that supports it. Evidence first. Value always. Capture consistently, share openly, decide confidently."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Basic information** Use case name · Short description · Unique ID · Date created · Last updated · Author / owner"
},
{
"t": "p",
"v": "**2. Problem & purpose** Problem / opportunity · Purpose & objectives · Success criteria"
},
{
"t": "p",
"v": "**3. Users & context** Primary user groups · Who benefits · Operating context / environment · Geography / organisation"
},
{
"t": "p",
"v": "**4. Solution & capability** AI capability / agent type · Key features / functions · How it works (high level) · Human oversight points"
},
{
"t": "p",
"v": "**5. Data & information** Key data sources · Type of data · Data sensitivity · Information governance"
},
{
"t": "p",
"v": "**6. Implementation & integration** Platform / technology · Systems / tools integrated · Deployment model · Date went live / expected · Resource requirements"
},
{
"t": "p",
"v": "**7. Outcomes & value** Quantitative outcomes (measured) · Qualitative outcomes (observed) · Value delivered (how & for whom)"
},
{
"t": "p",
"v": "**8. Costs & financials** Cost model (licence, usage, etc.) · Estimated / actual cost · Cost per user / transaction · Ongoing cost considerations · Funding source / owner"
},
{
"t": "p",
"v": "**9. Risks & assurance** Key risks · Mitigations in place · Assurance / evaluation approach"
},
{
"t": "p",
"v": "**10. Evidence & validation** Evidence summary · Evidence sources · What evidence is available (performance results, user feedback, evaluation report, audit / review, other)"
},
{
"t": "p",
"v": "**11. Limitations & lessons** Known limitations / constraints · Lessons learned / recommendations"
},
{
"t": "p",
"v": "**12. Transferability & reuse** Potential for reuse (high / medium / low) · Reusable artefacts available · Transferability notes"
},
{
"t": "p",
"v": "**13. Lifecycle & review** Lifecycle stage (ideation, pilot, live, scaled, retired) · Review frequency · Next review date · Sunset / retirement plan"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence strength"
},
{
"t": "small",
"v": "**Strong** — Independently validated"
},
{
"t": "small",
"v": "**Medium** — Reported results with some validation"
},
{
"t": "small",
"v": "**Emerging** — Pilot / early indications"
},
{
"t": "small",
"v": "**Weak** — Unverified claims"
},
{
"t": "small",
"v": "**Unknown** — Insufficient evidence"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence status"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Live / in use",
"big": false
},
{
"t": "title",
"v": "Pilot / trial",
"big": false
},
{
"t": "title",
"v": "Planned",
"big": false
},
{
"t": "title",
"v": "Retired / stopped",
"big": false
},
{
"t": "title",
"v": "Unknown",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Worked, failed and unproven"
},
{
"t": "p",
"v": "Sharing the full picture — successes, failures and emerging ideas — so we learn faster and build smarter. Transparent evidence helps the public sector avoid repeating mistakes and scale what works. The cases below are illustrative."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Worked — repeatable success"
},
{
"t": "small",
"v": "Evidence shows value, quality and user benefit."
},
{
"t": "h",
"l": 3,
"v": "Case: ACAS 'Haste' helpline knowledge assistant"
},
{
"t": "p",
"v": "Helps advisers quickly find accurate, impartial information during calls."
},
{
"t": "h",
"l": 4,
"v": "What worked"
},
{
"t": "small",
"v": "Grounded in authoritative, approved and consistent formats"
},
{
"t": "small",
"v": "Tested with 1,000 differently worded questions"
},
{
"t": "small",
"v": "High accuracy and consistency across impartiality and bias checks"
},
{
"t": "small",
"v": "Clear roles, assistant, not decision maker"
},
{
"t": "small",
"v": "Positive adviser feedback and time savings"
},
{
"t": "h",
"l": 4,
"v": "Outcomes"
},
{
"t": "small",
"v": "Faster information retrieval"
},
{
"t": "small",
"v": "Improved accuracy and consistency"
},
{
"t": "small",
"v": "Better adviser experience"
},
{
"t": "small",
"v": "Scalable to wider teams"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Failed — learned & corrected"
},
{
"t": "small",
"v": "Initial attempt did not deliver — lessons captured and applied."
},
{
"t": "h",
"l": 3,
"v": "Case: AI-assisted NHS document processing"
},
{
"t": "p",
"v": "Automated extraction from clinical documents produced high initial error rate."
},
{
"t": "h",
"l": 4,
"v": "What went wrong"
},
{
"t": "small",
"v": "Poor data quality and inconsistent document formats"
},
{
"t": "small",
"v": "Incomplete metadata and missing context"
},
{
"t": "small",
"v": "Model hallucinations and incorrect extractions"
},
{
"t": "small",
"v": "Insufficient validation and no independent review loop"
},
{
"t": "small",
"v": "Rushed from prototype to wider use"
},
{
"t": "h",
"l": 4,
"v": "What we did next"
},
{
"t": "small",
"v": "Reduced manual re-keying and error tracking"
},
{
"t": "small",
"v": "Good accuracy in structured readings"
},
{
"t": "small",
"v": "Went back with clear agendas and permissions"
},
{
"t": "small",
"v": "User trust and change impact still being assessed"
},
{
"t": "small",
"v": "Scalability and cost model not yet validated"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Unproven — promising, not yet validated"
},
{
"t": "small",
"v": "Early potential identified — evidence still limited."
},
{
"t": "h",
"l": 3,
"v": "Case: Agent automatically summarises meeting action agent"
},
{
"t": "small",
"v": "Summarises meeting decisions, actions and sends updates to owners."
},
{
"t": "h",
"l": 4,
"v": "What we know so far"
},
{
"t": "small",
"v": "Long-term testing across teams and meeting types"
},
{
"t": "small",
"v": "Long-term accuracy and retention unknown"
},
{
"t": "small",
"v": "Cost at scale and return on investment not established"
},
{
"t": "small",
"v": "Governance and data sensitivity assessments in progress"
},
{
"t": "h",
"l": 4,
"v": "Next steps"
},
{
"t": "p",
"v": "Extended pilot, independent evaluation, costing and governance review before scaling"
},
{
"t": "small",
"v": "Data quality and validation are critical"
},
{
"t": "small",
"v": "Human review and iterative testing prevent scale of error"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How we succeed together"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Share openly, including failures",
"big": false
},
{
"t": "title",
"v": "Ground in authoritative sources",
"big": false
},
{
"t": "title",
"v": "Test with real users",
"big": false
},
{
"t": "title",
"v": "Measure impact and value",
"big": false
},
{
"t": "title",
"v": "Capture lessons and publish patterns",
"big": false
},
{
"t": "title",
"v": "Reuse and continuously improve",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Four proven delivery patterns"
},
{
"t": "p",
"v": "Each pattern is a pipeline with human oversight built in — not a single tool. Use them as the starting shape for a new use case."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Meeting intelligence pipeline"
},
{
"t": "p",
"v": "Transforming conversations into decisions, actions and organisational memory"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "1. Meeting planning",
"big": false
},
{
"t": "title",
"v": "2. Consent & capture",
"big": false
},
{
"t": "title",
"v": "3. Transcript creation",
"big": false
},
{
"t": "title",
"v": "4. AI processing & summarisation",
"big": false
},
{
"t": "title",
"v": "5. Decisions & actions",
"big": false
},
{
"t": "title",
"v": "6. Risks & insights",
"big": false
},
{
"t": "title",
"v": "7. Distribution & integration",
"big": false
},
{
"t": "title",
"v": "8. Organisational memory",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Authoritative knowledge assistant"
},
{
"t": "p",
"v": "Delivering trusted, accurate answers grounded in authoritative sources, with human expertise in the loop"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "1. Define need",
"big": false
},
{
"t": "title",
"v": "2. Retrieve",
"big": false
},
{
"t": "title",
"v": "3. Evaluate & ground",
"big": false
},
{
"t": "title",
"v": "4. Generate response",
"big": false
},
{
"t": "title",
"v": "5. Human review",
"big": false
},
{
"t": "title",
"v": "6. Deliver answer",
"big": false
},
{
"t": "title",
"v": "7. Feedback & learn",
"big": false
},
{
"t": "title",
"v": "8. Maintain sources",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI-assisted engineering lifecycle"
},
{
"t": "small",
"v": "Secure. Quality. Production-ready. Human in control."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "1. Discover & plan",
"big": false
},
{
"t": "title",
"v": "2. Architect & design",
"big": false
},
{
"t": "title",
"v": "3. Build & code",
"big": false
},
{
"t": "title",
"v": "4. Test & validate",
"big": false
},
{
"t": "title",
"v": "5. Review & assure",
"big": false
},
{
"t": "title",
"v": "6. Deploy & release",
"big": false
},
{
"t": "title",
"v": "7. Operate & monitor",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI-assisted programme & portfolio management"
},
{
"t": "small",
"v": "Smarter oversight. Better decisions. Greater impact."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "1. Capture & ingest",
"big": false
},
{
"t": "title",
"v": "2. Analyse & understand",
"big": false
},
{
"t": "title",
"v": "3. Plan & optimise",
"big": false
},
{
"t": "title",
"v": "4. Monitor & track",
"big": false
},
{
"t": "title",
"v": "5. Decide & recommend",
"big": false
},
{
"t": "title",
"v": "6. Collaborate & execute",
"big": false
},
{
"t": "title",
"v": "7. Learn & improve",
"big": false
}
]
}
]
}
]
},
{
"id": "raicases",
"part": "decide-and-prioritise",
"title": "Public sector case library",
"icon": "cases",
"eyebrow": "Decide and prioritise · public sector case library",
"heading": "What central, local and devolved government has already done",
"lede": "Sourced cases from central government, councils, the Scottish, Welsh and Northern Ireland governments and health bodies. Each carries its source and an evidence grade. Supplier-published figures stay graded as reported.",
"body": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Central government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Corroborated"
}
]
},
{
"t": "h",
"l": 2,
"v": "Humphrey: Consult, Parlex, Minute, Redbox and Lex"
},
{
"t": "title",
"v": "DSIT · i.AI, now part of GDS",
"big": false
},
{
"t": "p",
"v": "A package of AI tools for civil servants: consultation analysis, parliamentary search, meeting transcription, document summarisation and legal research."
},
{
"t": "p",
"v": "**Governance lesson:** Consult was tested on past consultations before live use."
},
{
"t": "title",
"v": "Source: UKAuthority ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Central government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Extract: digitising planning documents"
},
{
"t": "title",
"v": "i.AI with MHCLG",
"big": false
},
{
"t": "p",
"v": "Extract digitises planning documents. Early testing completed a one-to-two-hour manual task in under a minute."
},
{
"t": "p",
"v": "**Governance lesson:** Check extraction against the source before it feeds planning decisions."
},
{
"t": "title",
"v": "Source: Roadmap for modern digital government ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Scotland"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Consult tested live on a Scottish Government consultation"
},
{
"t": "title",
"v": "i.AI with the Scottish Government",
"big": false
},
{
"t": "p",
"v": "The first live test of Consult used a Scottish Government consultation, followed by several other government consultations."
},
{
"t": "p",
"v": "**Governance lesson:** Compare AI theming with human analysts before relying on it."
},
{
"t": "title",
"v": "Source: Roadmap for modern digital government ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Local government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Minute trialled in frontline local services"
},
{
"t": "title",
"v": "i.AI",
"big": false
},
{
"t": "p",
"v": "Minute, a transcription and summarisation tool, was trialled in frontline local services in July 2025."
},
{
"t": "p",
"v": "**Governance lesson:** Settle consent and retention for recordings before a trial starts."
},
{
"t": "title",
"v": "Source: Roadmap for modern digital government ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Local government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Magic Notes in adult social care"
},
{
"t": "title",
"v": "Kingston Council with Beam",
"big": false
},
{
"t": "p",
"v": "Piloted AI recording and summarising of assessments to free social workers for direct care."
},
{
"t": "p",
"v": "**Governance lesson:** A DPIA examined data flows, storage and access. The privacy notice was updated and practitioners get consent to record."
},
{
"t": "title",
"v": "Source: Local Government Association ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Local government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "h",
"l": 2,
"v": "Magic Notes scaled to over 1,000 staff"
},
{
"t": "title",
"v": "Kent County Council",
"big": false
},
{
"t": "p",
"v": "After 29 practitioners piloted the tool, more than 1,000 frontline staff use it. An evaluation reported seven hours a week saved on average."
},
{
"t": "p",
"v": "**Governance lesson:** Figures are supplier-published. Validate locally before using them in a business case."
},
{
"t": "title",
"v": "Source: techUK ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Local government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Service-level privacy notice for AI transcription"
},
{
"t": "title",
"v": "Hartlepool Borough Council",
"big": false
},
{
"t": "p",
"v": "A privacy notice for Magic Notes in adult social care sets out purposes, legal basis and rights."
},
{
"t": "p",
"v": "**Governance lesson:** A published, service-level privacy notice is a reusable transparency artefact."
},
{
"t": "title",
"v": "Source: Hartlepool Borough Council ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Local government"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Corroborated"
}
]
},
{
"t": "h",
"l": 2,
"v": "AI recording in adult social care"
},
{
"t": "title",
"v": "Swindon Borough Council",
"big": false
},
{
"t": "p",
"v": "Piloted AI to improve the quality and speed of recording assessment conversations."
},
{
"t": "p",
"v": "**Governance lesson:** Evaluate quality of the record, not only time saved."
},
{
"t": "title",
"v": "Source: Socitm ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Local government"
},
{
"t": "eyebrow",
"v": "· Declarative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Data & Analytics Board for data and AI governance"
},
{
"t": "title",
"v": "Belfast City Council",
"big": false
},
{
"t": "p",
"v": "Set up a Data & Analytics Board to oversee data and AI governance, alongside a council-wide generative AI policy and staff training."
},
{
"t": "small",
"v": "**Governance lesson:** Data and AI governance sit under one board."
},
{
"t": "title",
"v": "Source: Belfast City Council ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Scotland"
},
{
"t": "eyebrow",
"v": "· Agentic"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Scotland’s AI Strategy 2026–2031"
},
{
"t": "title",
"v": "Scottish Government",
"big": false
},
{
"t": "p",
"v": "Published 20 March 2026. An eight-layer “AI Stack” and a data matchmaking pilot for trusted public-sector datasets by 2027."
},
{
"t": "p",
"v": "**Governance lesson:** Public-sector AI use is made visible through the Scottish AI Register."
},
{
"t": "title",
"v": "Source: gov.scot ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Scotland"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "AI guidance on the Digital Scotland Service Manual"
},
{
"t": "title",
"v": "Scottish Government Public Sector AI Team",
"big": false
},
{
"t": "p",
"v": "A suite of AI guidance and resources published on the Digital Scotland Service Manual in August 2026."
},
{
"t": "p",
"v": "**Governance lesson:** Guidance lives in the service manual teams already use."
},
{
"t": "title",
"v": "Source: Digital Scotland blog ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Wales"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "AI Cymru: the AI Plan for Wales"
},
{
"t": "title",
"v": "Welsh Government",
"big": false
},
{
"t": "p",
"v": "AI adoption across health, education and local government, with a commitment that AI works equally well in Welsh and English."
},
{
"t": "p",
"v": "**Governance lesson:** Public bodies assess risks to vulnerable groups and consider offline users."
},
{
"t": "title",
"v": "Source: gov.wales ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Wales"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "h",
"l": 2,
"v": "Office for AI finds duplicated AI work"
},
{
"t": "title",
"v": "Welsh Government Office for AI",
"big": false
},
{
"t": "p",
"v": "A baselining exercise found public bodies duplicating AI work. The office now coordinates adoption and says not every pilot should go live."
},
{
"t": "p",
"v": "**Governance lesson:** Duplication is a governance failure; a shared register and reuse check fix it."
},
{
"t": "title",
"v": "Source: Resultsense via UKAuthority ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Wales"
},
{
"t": "eyebrow",
"v": "· Agentic"
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "h",
"l": 2,
"v": "Senedd research on generative AI and agents"
},
{
"t": "title",
"v": "Senedd Research",
"big": false
},
{
"t": "p",
"v": "A briefing on generative AI in Welsh public services covers inaccurate outputs, discrimination, privacy and the added security and accountability challenges of AI agents."
},
{
"t": "p",
"v": "**Governance lesson:** Agent autonomy brings separate security and accountability questions."
},
{
"t": "title",
"v": "Source: Think Digital Partners ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Health"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "h",
"l": 2,
"v": "AI scribing in children’s neurodevelopmental services"
},
{
"t": "title",
"v": "Cwm Taf Morgannwg University Health Board with CDPS",
"big": false
},
{
"t": "p",
"v": "A pilot assessing whether AI scribing reduces clinicians’ administrative workload."
},
{
"t": "p",
"v": "**Governance lesson:** Pilot and evaluate before scaling clinical documentation tools."
},
{
"t": "title",
"v": "Source: BusinessIn Wales ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Devolved: Northern Ireland"
},
{
"t": "eyebrow",
"v": "· Agentic"
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "h",
"l": 2,
"v": "Draft Northern Ireland AI Strategy"
},
{
"t": "title",
"v": "The Executive Office",
"big": false
},
{
"t": "p",
"v": "Consultation from 12 August to 7 October 2026. Eight principles include human oversight, accountability and redress, and data governance."
},
{
"t": "p",
"v": "**Governance lesson:** Proposes human oversight teams, a cross-sector AI governance board and inventories of AI projects."
},
{
"t": "title",
"v": "Source: Executive Office ↗",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "International comparator"
},
{
"t": "eyebrow",
"v": "· Generative"
},
{
"t": "eyebrow",
"v": "Corroborated"
}
]
},
{
"t": "h",
"l": 2,
"v": "Guidelines for AI in the public service"
},
{
"t": "title",
"v": "Government of Ireland",
"big": false
},
{
"t": "p",
"v": "Advises against free generative AI tools because information given to them could be used for training."
},
{
"t": "p",
"v": "**Governance lesson:** A clear position on consumer tools reduces shadow AI."
},
{
"t": "title",
"v": "Source: RTÉ ↗",
"big": false
}
]
}
],
"box": false
},
{
"t": "p",
"v": "17 cases in the library. Gathered from public sources on 19 October 2026. Re-check each source before citing it."
}
]
},
{
"id": "raiassess",
"part": "decide-and-prioritise",
"title": "Classify & assess",
"icon": "rule",
"eyebrow": "Responsible AI · assess a use case",
"heading": "Classify first, then answer only what your tier needs",
"lede": "As AI can see more, do more and act more independently, the assurance required rises. Criticality & gates scores the service impact; this chapter sets the route from what the AI itself can do.",
"body": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "eyebrow",
"v": "Class"
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "1. Ask",
"big": false
},
{
"t": "small",
"v": "Answer and assist"
},
{
"t": "small",
"v": "Provides information, answers or summaries based on a prompt."
},
{
"t": "small",
"v": "**Example:** Ask about a policy or summarise a document."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "2. Create",
"big": false
},
{
"t": "small",
"v": "Generate content"
},
{
"t": "small",
"v": "Creates new content, analysis, images or code."
},
{
"t": "small",
"v": "**Example:** Draft a briefing, a presentation or code."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "3. Connect",
"big": false
},
{
"t": "small",
"v": "Use organisational information"
},
{
"t": "p",
"v": "Retrieves and works with data from organisational sources and external systems."
},
{
"t": "small",
"v": "**Example:** Search SharePoint or analyse internal data."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "4. Act",
"big": false
},
{
"t": "small",
"v": "Take actions"
},
{
"t": "small",
"v": "Makes changes in systems, files or records, or triggers workflows."
},
{
"t": "small",
"v": "**Example:** Create a record, send an email or start a process."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "5. Delegate",
"big": false
},
{
"t": "small",
"v": "Complete multi-step work"
},
{
"t": "small",
"v": "Plans and performs several steps with tools and limited supervision."
},
{
"t": "small",
"v": "**Example:** Research, analyse, write a report and share it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "6. Persist",
"big": false
},
{
"t": "small",
"v": "Operate proactively"
},
{
"t": "small",
"v": "Runs autonomously or on a schedule, monitoring and acting over time."
},
{
"t": "small",
"v": "**Example:** Monitor changes, raise tickets or act on triggers."
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Information access",
"big": false
},
{
"t": "small",
"v": "User-provided only"
},
{
"t": "small",
"v": "User-provided only"
},
{
"t": "small",
"v": "Organisational sources"
},
{
"t": "small",
"v": "Organisational and system"
},
{
"t": "small",
"v": "Multiple sources and tools"
},
{
"t": "small",
"v": "Ongoing and broad"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "System access",
"big": false
},
{
"t": "small",
"v": "None"
},
{
"t": "small",
"v": "None"
},
{
"t": "small",
"v": "Read (typically)"
},
{
"t": "small",
"v": "Read and write"
},
{
"t": "small",
"v": "Multiple systems"
},
{
"t": "small",
"v": "Multiple systems, scheduled"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Autonomy",
"big": false
},
{
"t": "small",
"v": "User led"
},
{
"t": "small",
"v": "User led"
},
{
"t": "small",
"v": "User led"
},
{
"t": "small",
"v": "User led, with action"
},
{
"t": "small",
"v": "Partial autonomy"
},
{
"t": "small",
"v": "High autonomy"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Persistence",
"big": false
},
{
"t": "small",
"v": "Single interaction"
},
{
"t": "small",
"v": "Single interaction"
},
{
"t": "small",
"v": "Single or repeated use"
},
{
"t": "small",
"v": "Repeated use"
},
{
"t": "small",
"v": "Multi-step tasks"
},
{
"t": "small",
"v": "Continuous or scheduled"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Human oversight",
"big": false
},
{
"t": "small",
"v": "Review output"
},
{
"t": "small",
"v": "Review output"
},
{
"t": "small",
"v": "Review output and sources"
},
{
"t": "small",
"v": "Review before or after action"
},
{
"t": "small",
"v": "Ongoing oversight"
},
{
"t": "small",
"v": "Defined oversight and monitoring"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Assurance depth",
"big": false
},
{
"t": "small",
"v": "Lightweight"
},
{
"t": "small",
"v": "Lightweight"
},
{
"t": "small",
"v": "Standard"
},
{
"t": "small",
"v": "Enhanced"
},
{
"t": "small",
"v": "High"
},
{
"t": "small",
"v": "Specialist"
}
]
}
],
"box": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Generative, declarative and agentic: the three capability types"
},
{
"t": "p",
"v": "The capability classes say what the AI can do; the capability type says how it is built. Both set the assurance."
},
{
"t": "p",
"v": "goes deeper on choosing between them."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "auto_awesome"
},
{
"t": "title",
"v": "Generative",
"big": false
},
{
"t": "p",
"v": "Creates content in response to a prompt: text, summaries, images or code. A person decides what happens next."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "tune"
},
{
"t": "title",
"v": "Declarative",
"big": false
},
{
"t": "p",
"v": "Configured, not coded. Purpose, instructions, knowledge and permitted actions are declared; the platform supplies the model and orchestration."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "smart_toy"
},
{
"t": "title",
"v": "Agentic",
"big": false
},
{
"t": "p",
"v": "Plans and takes actions across tools under delegated authority, choosing its own path to a goal, sometimes on a schedule."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "table",
"rows": [
[
{
"h": true,
"v": "Class"
},
{
"h": true,
"v": "Usual capability type"
},
{
"h": true,
"v": "Minimum assurance"
}
],
[
{
"h": true,
"v": "Ask"
},
{
"h": false,
"v": "Generative"
},
{
"h": false,
"v": "Review the output."
}
],
[
{
"h": true,
"v": "Create"
},
{
"h": false,
"v": "Generative"
},
{
"h": false,
"v": "Review the output; label AI involvement."
}
],
[
{
"h": true,
"v": "Connect"
},
{
"h": false,
"v": "Generative, grounded · or Declarative with knowledge sources"
},
{
"h": false,
"v": "Test what the index and connectors expose; review sources."
}
],
[
{
"h": true,
"v": "Act"
},
{
"h": false,
"v": "Declarative with actions · or Agentic"
},
{
"h": false,
"v": "Approval before action; log every action; least-privilege connectors."
}
],
[
{
"h": true,
"v": "Delegate"
},
{
"h": false,
"v": "Agentic"
},
{
"h": false,
"v": "Full agent route: identity, tool permissions, approval points, behavioural tests."
}
],
[
{
"h": true,
"v": "Persist"
},
{
"h": false,
"v": "Agentic, scheduled or proactive"
},
{
"h": false,
"v": "Specialist assurance: defined oversight, monitoring and a kill switch."
}
]
]
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "What to assure: Generative",
"big": false
},
{
"t": "small",
"v": "Output quality, grounding sources, data entered, transparency labels."
},
{
"t": "title",
"v": "T3 · T4 · T6",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "What to assure: Declarative",
"big": false
},
{
"t": "p",
"v": "The declared configuration: instructions, knowledge sources, actions, owner, version and publishing scope. Reviewed like code."
},
{
"t": "title",
"v": "T7 · A2 prompt and skills record",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "What to assure: Agentic",
"big": false
},
{
"t": "p",
"v": "The whole human-agent system: identity, tools, approval points, escalation, logs, cost caps and the kill switch."
},
{
"t": "title",
"v": "T7 · T8 · T10 · agent register",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "Capability and risk classification"
},
{
"t": "title",
"v": "1. The most it will do in this use case",
"big": false
},
{
"t": "title",
"v": "2. What raises the stakes",
"big": false
},
{
"t": "title",
"v": "3. Deterministic alternative test",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Assurance route"
},
{
"t": "title",
"v": "Choose a capability class",
"big": true
},
{
"t": "p",
"v": "Start with the most the tool will do in this use case, not the most it could do."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Assessment questions"
},
{
"t": "small",
"v": "0 of 60 answered · 0 concerns · 0 not known"
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "small",
"v": "Feeds: **T1 Opportunity Brief · T2 Duplication Check**"
},
{
"t": "small",
"v": "What need does the tool meet?"
},
{
"t": "small",
"v": "Is an approved capability already available?"
},
{
"t": "small",
"v": "Is AI necessary?"
},
{
"t": "small",
"v": "Who benefits and who may be adversely affected?"
},
{
"t": "small",
"v": "What evidence supports expected value?"
},
{
"t": "small",
"v": "Can the outcome be evaluated?"
}
]
},
{
"t": "p",
"v": "Answers stay in this browser. “Concern” and “Not known” items carry into the decision record as conditions or open questions."
},
{
"t": "h",
"l": 2,
"v": "Decision record outcomes"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Approve",
"big": false
},
{
"t": "small",
"v": "Proceed within the documented scope and controls."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Approve with conditions",
"big": false
},
{
"t": "small",
"v": "Proceed once named conditions are met; conditions are tracked."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Pilot",
"big": false
},
{
"t": "small",
"v": "Controlled authority for a limited population, time and data set."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Restrict",
"big": false
},
{
"t": "small",
"v": "Narrow the users, data or actions permitted."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Extend",
"big": false
},
{
"t": "small",
"v": "Extend a pilot to gather the missing evidence."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Stop",
"big": false
},
{
"t": "small",
"v": "Do not proceed. Record the reason and lessons."
}
]
}
],
"box": false
}
]
},
{
"id": "criticality",
"part": "decide-and-prioritise",
"title": "Criticality & gates",
"icon": "radar",
"eyebrow": "Criticality & assurance gates",
"heading": "Usage is not the same as criticality",
"lede": "Evaluate across ten dimensions to identify the agents that need stronger governance, assurance and operational support. Score each 1 (very low) to 5 (very high); the total sets the criticality tier and the action.",
"body": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "1. Usage & adoption",
"big": false
},
{
"t": "small",
"v": "How many people use it and how often?"
},
{
"t": "title",
"v": "2. Audience & reach",
"big": false
},
{
"t": "p",
"v": "Who relies on it? Internal teams, partners, other organisations, or the public?"
},
{
"t": "title",
"v": "3. Business impact",
"big": false
},
{
"t": "p",
"v": "How important is it to achieving business outcomes or service delivery?"
},
{
"t": "title",
"v": "4. Decision impact",
"big": false
},
{
"t": "p",
"v": "Does it influence, recommend or make decisions that have significant consequences?"
},
{
"t": "title",
"v": "5. Data & information",
"big": false
},
{
"t": "p",
"v": "What is the sensitivity, value and regulatory significance of the data it uses or produces?"
},
{
"t": "title",
"v": "6. Autonomy level",
"big": false
},
{
"t": "p",
"v": "How independently can the agent act? Human-in-the-loop versus autonomous."
},
{
"t": "title",
"v": "7. Irreversibility",
"big": false
},
{
"t": "small",
"v": "How easy is it to reverse an error or unintended action?"
},
{
"t": "title",
"v": "8. Integration & dependencies",
"big": false
},
{
"t": "small",
"v": "How integrated is it with other systems, processes or agents?"
},
{
"t": "title",
"v": "9. Risk & compliance",
"big": false
},
{
"t": "small",
"v": "What are the legal, regulatory, security and reputational risks?"
},
{
"t": "title",
"v": "10. Public consequence",
"big": false
},
{
"t": "small",
"v": "Could failure cause harm, loss of trust or impact citizens?"
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "CRITICALITY SCORE"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "—",
"big": true
},
{
"t": "title",
"v": "—",
"big": false
}
]
},
{
"t": "small",
"v": "0 of 10 dimensions scored"
},
{
"t": "title",
"v": "Score every dimension for a criticality tier.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "From score to action"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "12–20",
"big": false
},
{
"t": "p",
"v": "**Low**Monitor, light governance, owner accountable"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "21–30",
"big": false
},
{
"t": "p",
"v": "**Moderate**Standard governance, regular review and quality checks"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "31–40",
"big": false
},
{
"t": "p",
"v": "**High**Enhanced assurance, monitoring, human oversight required"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "41–50",
"big": false
},
{
"t": "p",
"v": "**Very high**Enterprise governance, formal support, continuity and audit"
}
],
"box": false
}
]
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Scoring guide"
},
{
"t": "small",
"v": "**1 Very low** — Minimal impact or exposure"
},
{
"t": "small",
"v": "**2 Low** — Limited impact or exposure"
},
{
"t": "small",
"v": "**3 Moderate** — Noticeable impact or exposure"
},
{
"t": "small",
"v": "**4 High** — Significant impact or exposure"
},
{
"t": "small",
"v": "**5 Very high** — Critical impact or exposure"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Example agents across tiers"
},
{
"t": "p",
"v": "**Low**Personal research assistant used by creator only"
},
{
"t": "p",
"v": "**Moderate**Team policy search agent used by a department"
},
{
"t": "p",
"v": "**High**Case triage agent influencing case allocation and actions"
},
{
"t": "p",
"v": "**Very high**Citizen service agent that makes determinations or commitments"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key questions to ask"
},
{
"t": "small",
"v": "Who owns it and is responsible for its use?"
},
{
"t": "small",
"v": "What happens if it gives the wrong answer?"
},
{
"t": "small",
"v": "Where does it get its information from?"
},
{
"t": "small",
"v": "How is it tested, monitored and reviewed?"
},
{
"t": "small",
"v": "Does it support a critical business process?"
},
{
"t": "small",
"v": "Is it shared outside its original team or creator?"
},
{
"t": "small",
"v": "Could it become business or citizen critical?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Important reminders"
},
{
"t": "title",
"v": "High usage does not equal high criticality",
"big": false
},
{
"t": "title",
"v": "Low usage can still be business critical",
"big": false
},
{
"t": "title",
"v": "Reassess regularly as agents, usage and context change",
"big": false
},
{
"t": "title",
"v": "When in doubt, apply the higher level of control",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Right controls at the right level of risk and impact"
},
{
"t": "p",
"v": "Assurance increases with audience, autonomy, information sensitivity, criticality and consequence. Pick a tier to see the gate required before use."
},
{
"t": "row",
"items": [
{
"t": "p",
"v": "**Audience**Who is affected?"
},
{
"t": "p",
"v": "**Information**What data?"
},
{
"t": "p",
"v": "**Autonomy**What can it do?"
},
{
"t": "p",
"v": "**Criticality**How essential?"
},
{
"t": "p",
"v": "**Consequence**What is the impact?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Tier 2 · Departmental"
},
{
"t": "h",
"l": 3,
"v": "Departmental capability · more users, more data"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Audience"
},
{
"t": "title",
"v": "Department / multiple teams",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Autonomy"
},
{
"t": "title",
"v": "Acts with human approval",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Information sensitivity"
},
{
"t": "title",
"v": "Moderate",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Criticality"
},
{
"t": "title",
"v": "Medium",
"big": false
}
]
}
]
},
{
"t": "p",
"v": "**ASSURANCE GATE — REQUIRED BEFORE USE** **Formal assessment (risk & data review)**"
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "Key controls"
},
{
"t": "small",
"v": "Data source approval"
},
{
"t": "small",
"v": "Access controls"
},
{
"t": "small",
"v": "Human-in-the-loop"
},
{
"t": "small",
"v": "Quality checks"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Oversight & accountability****Business owner + data/info owner**"
},
{
"t": "p",
"v": "**Evidence & logging****Audit log · decisions recorded**"
},
{
"t": "p",
"v": "**Review frequency****Monthly**"
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Notes on using the tiers"
},
{
"t": "small",
"v": "Use the lowest tier that achieves the required outcome."
},
{
"t": "small",
"v": "Reassess tier when usage, data, autonomy or audience changes."
},
{
"t": "small",
"v": "When in doubt, escalate the risk and apply higher-tier controls."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance principles"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "tune"
},
{
"t": "title",
"v": "Proportionate to risk",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "fact_check"
},
{
"t": "title",
"v": "Risk-based and evidence-led",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "person"
},
{
"t": "title",
"v": "Human oversight at the right level",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "search"
},
{
"t": "title",
"v": "Transparent and explainable",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "autorenew"
},
{
"t": "title",
"v": "Continuously improved",
"big": false
}
]
}
],
"box": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Assurance by design, proportionate by risk, continuous by evidence"
},
{
"t": "p",
"v": "A holistic, cross-disciplinary approach ensuring AI systems and agents are safe, secure, reliable, fair — and deliver public value."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance domains"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Security",
"big": false
},
{
"t": "title",
"v": "Privacy & data protection",
"big": false
},
{
"t": "title",
"v": "Data & information governance",
"big": false
},
{
"t": "title",
"v": "Ethics & fairness",
"big": false
},
{
"t": "title",
"v": "Transparency & explainability",
"big": false
},
{
"t": "title",
"v": "Human oversight & accountability",
"big": false
},
{
"t": "title",
"v": "Reliability & safety",
"big": false
},
{
"t": "title",
"v": "Evaluation & testing",
"big": false
},
{
"t": "title",
"v": "Auditability & evidence",
"big": false
},
{
"t": "title",
"v": "Risk management",
"big": false
},
{
"t": "title",
"v": "Legal & compliance",
"big": false
},
{
"t": "title",
"v": "Accessibility & inclusion",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Applied across the lifecycle"
},
{
"t": "p",
"v": "**1. Plan & design**Define purpose, risks, requirements and success measures"
},
{
"t": "p",
"v": "**2. Build & deploy**Develop, test, assure and prepare for secure deployment"
},
{
"t": "p",
"v": "**3. Operate & monitor**Monitor performance, risks, usage and emerging issues"
},
{
"t": "p",
"v": "**4. Review & improve**Learn from evidence, update controls and improve continuously"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key assurance questions"
},
{
"t": "small",
"v": "What are we trying to achieve and for whom?"
},
{
"t": "small",
"v": "What could go wrong and who could be harmed?"
},
{
"t": "small",
"v": "Is the data appropriate, accurate and lawful?"
},
{
"t": "small",
"v": "How will decisions be explained and challenged?"
},
{
"t": "small",
"v": "Can we evidence compliance and responsible use?"
},
{
"t": "small",
"v": "How will we monitor, learn and improve?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence & assurance artefacts"
},
{
"t": "small",
"v": "Risk assessments"
},
{
"t": "small",
"v": "Data & DPIA records"
},
{
"t": "small",
"v": "Model cards & evaluations"
},
{
"t": "small",
"v": "Test results & red-team reports"
},
{
"t": "small",
"v": "Decision logs & audit trails"
},
{
"t": "small",
"v": "Policies, standards & guidance"
},
{
"t": "small",
"v": "Change logs & reviews"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The human oversight and accountability chain"
},
{
"t": "p",
"v": "Agents must operate within clearly defined accountability and oversight boundaries, with named owners and transparent decision trails. Clear accountability at every step; human judgement at the right points."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Source owner",
"big": false
},
{
"t": "small",
"v": "Owns the authoritative information and data made available."
},
{
"t": "small",
"v": "Ensure accuracy and currency"
},
{
"t": "small",
"v": "Define permitted use"
},
{
"t": "small",
"v": "Manage access and permissions"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Agent owner",
"big": false
},
{
"t": "small",
"v": "Owns the agent's purpose, behaviour and operation."
},
{
"t": "small",
"v": "Define purpose and scope"
},
{
"t": "small",
"v": "Configure behaviour and guardrails"
},
{
"t": "small",
"v": "Monitor performance and usage"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Reviewer",
"big": false
},
{
"t": "small",
"v": "Reviews AI outputs, recommendations and actions."
},
{
"t": "small",
"v": "Validate accuracy and relevance"
},
{
"t": "small",
"v": "Check for bias, risk and compliance"
},
{
"t": "small",
"v": "Decide to approve, edit or escalate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Decision-maker",
"big": false
},
{
"t": "small",
"v": "Accountable for the decision and its consequences."
},
{
"t": "small",
"v": "Make or authorise the decision"
},
{
"t": "small",
"v": "Consider impact on people and services"
},
{
"t": "small",
"v": "Record the rationale"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Service owner",
"big": false
},
{
"t": "small",
"v": "Accountable for the service, outcomes and performance."
},
{
"t": "small",
"v": "Ensure outcomes are achieved"
},
{
"t": "small",
"v": "Manage risks, incidents and issues"
},
{
"t": "small",
"v": "Provide assurance and improvements"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Audit & redress",
"big": false
},
{
"t": "small",
"v": "Provides independent assurance and holds the organisation to account."
},
{
"t": "small",
"v": "Audit and review activity and controls"
},
{
"t": "small",
"v": "Ensure records, transparency and compliance"
},
{
"t": "small",
"v": "Enable challenge and redress"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Human oversight points"
},
{
"t": "p",
"v": "**1. Source approval**Information is approved for use by the agent."
},
{
"t": "p",
"v": "**2. Configuration review**Agent design and guardrails are reviewed and authorised."
},
{
"t": "p",
"v": "**3. Output review**Outputs are checked for quality, risk and compliance."
},
{
"t": "p",
"v": "**4. Decision authorisation**Final decision is made or authorised by a person."
},
{
"t": "p",
"v": "**5. Outcome monitoring**Outcomes and performance are monitored and reviewed."
},
{
"t": "p",
"v": "**6. Independent assurance**Independent review, audit and redress are available."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Benefits"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Clear accountability",
"big": false
},
{
"t": "title",
"v": "Better decisions",
"big": false
},
{
"t": "title",
"v": "Reduced risk",
"big": false
},
{
"t": "title",
"v": "Greater trust",
"big": false
},
{
"t": "title",
"v": "Stronger public value",
"big": false
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "maturity",
"part": "decide-and-prioritise",
"title": "Assurance tiers",
"icon": "shield",
"eyebrow": "Maturity & assurance · illustrative frameworks",
"heading": "Where you are, and how much control that needs",
"lede": "Two illustrative frameworks from the capability report: a seven-stage maturity model for organisations, and six assurance tiers for agents. Neither scores named organisations; both are diagnostic tools to calibrate against evidence. Over-control halts adoption; under-control creates unmanaged exposure.",
"body": [
{
"t": "h",
"l": 2,
"v": "The seven-stage maturity model"
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Observable evidence at stage 2"
},
{
"t": "p",
"v": "Bounded pilots; named sponsors; temporary funding; largely self-reported benefit evidence."
}
]
},
{
"t": "h",
"l": 2,
"v": "Evidence to collect before claiming progression"
},
{
"t": "p",
"v": "Named owners, repeatable gates, measured outcomes — not licensed tools."
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Leadership & service ownership",
"big": false
},
{
"t": "p",
"v": "Named accountable owners, clear outcomes, risk appetite and decision records."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "People & workforce",
"big": false
},
{
"t": "p",
"v": "Role-based training, protected learning time, inclusive adoption measures, staff consultation."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Technology & agents",
"big": false
},
{
"t": "p",
"v": "Register, architecture, approved platforms, integrations, autonomy level, observability."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Data & information",
"big": false
},
{
"t": "p",
"v": "Authoritative sources, metadata, permissions remediation, retention, provenance."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Security & assurance",
"big": false
},
{
"t": "p",
"v": "Threat model, DPIA where required, evaluation evidence, incident route, audit logs."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Commercial & supplier",
"big": false
},
{
"t": "p",
"v": "Licensing clarity, usage data, audit rights, exit, substitution, support terms."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Finance & FinOps",
"big": false
},
{
"t": "p",
"v": "Forecasts, showback, anomaly alerts, unit economics, value attribution."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Community learning",
"big": false
},
{
"t": "p",
"v": "Reusable artefacts, supersession, contribution workflow, evidence of reuse."
}
],
"box": true
}
]
},
{
"t": "h",
"l": 2,
"v": "Six agent assurance tiers — controls scale with autonomy"
},
{
"t": "p",
"v": "Bounded autonomy and recoverability adopted from Government of Canada agentic-AI guidance (May 2026)."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "num",
"v": "1"
},
{
"t": "p",
"v": "**Personal assistant**Drafting, summarisation and retrieval for one accountable user"
},
{
"t": "p",
"v": "Acceptable-use policy, permissions hygiene, human verification, retention and transparency guidance."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "num",
"v": "2"
},
{
"t": "p",
"v": "**Personal or team agent**Bounded knowledge and repeatable team tasks"
},
{
"t": "p",
"v": "Registered owner, approved sources, test prompts, sharing gate, usage telemetry."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "num",
"v": "3"
},
{
"t": "p",
"v": "**Departmental reusable agent**Multiple teams or a repeatable departmental process"
},
{
"t": "p",
"v": "Business, technical, information, risk and support owners; service level; change control; retirement plan."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "num",
"v": "4"
},
{
"t": "p",
"v": "**Operational agent**Acts on records, workflows or line-of-business systems"
},
{
"t": "p",
"v": "Distinct identity, least privilege, action logging, rollback, incident response, continuous monitoring."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "num",
"v": "5"
},
{
"t": "p",
"v": "**Citizen-impacting agent**Influences access, rights, safety, services or public communication"
},
{
"t": "p",
"v": "Impact assessment, legal basis, accessibility and equality assessment, transparency, redress, independent assurance."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "num",
"v": "6"
},
{
"t": "p",
"v": "**Autonomous / multi-agent**Delegated goals, tools, sub-agents or long-running actions"
},
{
"t": "p",
"v": "Sandboxing, constrained authority, short-lived credentials, kill switch, behavioural monitoring, dependency mapping."
}
],
"box": true
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Tier finder — two questions set the tier at intake"
},
{
"t": "p",
"v": "What may this agent do without asking, and what can be undone when it goes wrong?"
}
]
}
]
},
{
"id": "railifecycle",
"part": "design-and-build",
"title": "Lifecycle & templates",
"icon": "account_tree",
"eyebrow": "Responsible AI · implement AI",
"heading": "Twelve stages, twelve gates, one evidence chain",
"lede": "Each stage answers one question, produces one artefact and passes one gate. Select a stage to filter the templates. The Agent journey chapter follows a single agent through the same path.",
"body": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Explore"
},
{
"t": "eyebrow",
"v": "Assess and design"
},
{
"t": "eyebrow",
"v": "Deploy and enable"
},
{
"t": "eyebrow",
"v": "Operate and evolve"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Key gates"
},
{
"t": "list",
"o": true,
"items": [
[
{
"t": "title",
"v": "Validated need",
"big": false
},
{
"t": "small",
"v": "Stage 1"
}
],
[
{
"t": "title",
"v": "Reuse decision",
"big": false
},
{
"t": "small",
"v": "Stage 2"
}
],
[
{
"t": "title",
"v": "Assessment approval",
"big": false
},
{
"t": "small",
"v": "Stage 4"
}
],
[
{
"t": "title",
"v": "Pilot authority",
"big": false
},
{
"t": "small",
"v": "Stage 6"
}
],
[
{
"t": "title",
"v": "Deployment approval",
"big": false
},
{
"t": "small",
"v": "Stage 9"
}
],
[
{
"t": "title",
"v": "Continued authority",
"big": false
},
{
"t": "small",
"v": "Stage 10"
}
],
[
{
"t": "title",
"v": "Closure confirmed",
"big": false
},
{
"t": "small",
"v": "Stage 12"
}
]
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Where each phase lives, what it produces, who does it"
},
{
"t": "card",
"kids": [
{
"t": "table",
"rows": [
[
{
"h": true,
"v": "Portal section"
},
{
"h": true,
"v": "Key outputs"
},
{
"h": true,
"v": "Primary roles"
}
],
[
{
"h": true,
"v": "Start here"
},
{
"h": false,
"v": "Opportunity brief · initial questions · support and contact"
},
{
"h": false,
"v": "All colleagues"
}
],
[
{
"h": true,
"v": "Find an AI tool"
},
{
"h": false,
"v": "Catalogue search · tool comparison · reuse decision"
},
{
"h": false,
"v": "Users · IT and digital"
}
],
[
{
"h": true,
"v": "Assess a use case"
},
{
"h": false,
"v": "Risk classification · data and privacy assessments · supplier and model review · approval record"
},
{
"h": false,
"v": "Project team · risk, security, privacy · data and information"
}
],
[
{
"h": true,
"v": "Implement AI"
},
{
"h": false,
"v": "Design record · pilot plan · evaluation report · deployment checklist"
},
{
"h": false,
"v": "Product owner · technical team · users and change leads"
}
],
[
{
"h": true,
"v": "Operate and monitor"
},
{
"h": false,
"v": "Usage, cost and outcome data · incident and risk records · change-impact assessments · continuous improvement"
},
{
"h": false,
"v": "Service owner · operational support · governance and assurance"
}
],
[
{
"h": true,
"v": "Learn and reuse"
},
{
"h": false,
"v": "Lessons learned · updated guidance · patterns and templates · retirement record"
},
{
"h": false,
"v": "All teams · community of practice · centre of excellence"
}
]
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Templates and artefacts · 12 core, 7 added"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T1 · Core · stage 1"
},
{
"t": "title",
"v": "AI Opportunity Brief",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Reference, title, owner and team",
"Problem evidence and affected users",
"Current process and non-AI options",
"Desired outcome and success measures",
"Inputs, outputs, decisions and systems",
"Accessibility need, urgency and constraints"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T2 · Core · stage 2"
},
{
"t": "title",
"v": "Existing-Capability and Duplication Check",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Required capability",
"Catalogue and pattern searches",
"Existing tools, agents, prompts and skills reviewed",
"Gap and configuration assessment",
"Reuse, extend, build or buy rationale",
"Decision and consulted owners"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T3 · Core · stage 3"
},
{
"t": "title",
"v": "Capability and Risk Classification",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Ask, Create, Connect, Act, Delegate, Persist",
"Information sensitivity and personal data",
"System access and autonomy",
"Consequential decisions and vulnerable users",
"Scale, reversibility and public visibility",
"Risk tier and assurance route"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T4 · Core · stage 4"
},
{
"t": "title",
"v": "Information and Data Readiness Assessment",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Sources, owners and classification",
"Personal data and DPIA needs",
"Permissions and business need",
"Metadata, filenames, titles and indexing",
"Quality, authority, retention and deletion",
"Traceability, aggregation and training use"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T5 · Core · stage 4"
},
{
"t": "title",
"v": "Supplier, Model and Deployment Assessment",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Supplier, edition and deployment route",
"Models, provider and automatic selection",
"Subprocessors, locations and retention",
"Contract, policy and admin controls",
"Logging, change notice and deprecation",
"Exit plan and unresolved uncertainties"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T6 · Core · stage 4"
},
{
"t": "title",
"v": "Responsible AI and Human Oversight Assessment",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Task and decisions affected",
"Reviewer, competence and evidence access",
"Uncertainty, challenge and override",
"Automation bias and prohibited uses",
"Failure, refusal and conflict handling",
"Redress and oversight effectiveness"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T7 · Core · stage 5"
},
{
"t": "title",
"v": "Agent Design Record",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Purpose, owner, identifier and status",
"Model, instructions, skills and knowledge",
"Identity, permissions, tools and credentials",
"Triggers, schedules, actions and other agents",
"Approval points, escalation and stopping",
"Tests, logging, monitoring, kill and review"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T8 · Core · stage 6"
},
{
"t": "title",
"v": "Pilot Plan",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Hypothesis, population, duration and environment",
"Permitted data and tasks, and exclusions",
"Support and training",
"Quality, accessibility, cost, trust and risk measures",
"Comparison and baseline design",
"Stop criteria and evidence ownership"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T9 · Core · stage 7"
},
{
"t": "title",
"v": "Evaluation and Benefits Record",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Outcomes, quality and reliability",
"Accessibility, wellbeing and user outcome",
"Time, review and rework",
"Cost, adoption and sustainability",
"Unintended effects and limitations",
"Evidence status and recommendation"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T10 · Core · stage 9"
},
{
"t": "title",
"v": "Deployment Readiness Checklist",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Catalogue and ownership",
"Assessments and approvals",
"Permissions and indexing tests",
"Oversight and accessibility",
"Guidance, support and monitoring",
"Incident, kill, retention, review and retirement"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T11 · Core · stage 11"
},
{
"t": "title",
"v": "Change-Impact Assessment",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Trigger: model, supplier, data, permission, action, interface, cost or policy",
"Impact on risks and controls",
"Evidence and documents to update",
"No, partial or full reassessment",
"Continue, restrict, suspend or retire decision"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "T12 · Core · stage 12"
},
{
"t": "title",
"v": "Retirement Record",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Reason and replacement",
"User communication",
"Disable agents and actions, and revoke access",
"Disconnect integrations and dispose of data",
"Update records and supplier access",
"Residual risks, lessons and closure approval"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A1 · Added · stage 9, 10"
},
{
"t": "title",
"v": "Adoption and capability plan",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Senior sponsor and local champions",
"Protected learning time",
"Local communications",
"Accessible training and reinforcement",
"Support"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A2 · Added · stage 5, 10"
},
{
"t": "title",
"v": "Prompt and skills product record",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Owner, version and purpose",
"Approved users",
"Test set and security checks",
"Cost profile",
"Retirement"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A3 · Added · stage 5, 9"
},
{
"t": "title",
"v": "Support and service record",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Support route and service dependencies",
"Escalation",
"Known limitations",
"Incident response",
"Service-level position"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A4 · Added · stage 6, 7"
},
{
"t": "title",
"v": "Comparative task-test template",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Baseline tool versus proposed capability",
"Task, time and cost",
"Review effort and output quality",
"Failure modes",
"Accessibility"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A5 · Added · stage 4, 7"
},
{
"t": "title",
"v": "Ways-of-working impact assessment",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Delegation and role development",
"Workload and meeting volume",
"Professional review",
"Change-management effects"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A6 · Added · stage 5, 9"
},
{
"t": "title",
"v": "Guest and cross-tenant access plan",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Resource location",
"Permissions",
"Alternate access route",
"Accessible distribution"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "A7 · Added · stage 11"
},
{
"t": "title",
"v": "Release and guidance change record",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Product terminology and screenshots",
"Training assets",
"Affected audiences",
"Communication date"
]
}
]
}
],
"box": false
}
]
},
{
"id": "capclasses",
"part": "design-and-build",
"title": "Capability classes",
"icon": "category",
"eyebrow": "Design and build · capability classes",
"heading": "Generative, agentic and declarative",
"lede": "Three ways of using the same underlying models, with different failure modes and therefore different controls. Most real deployments combine all three, which is exactly why the class needs naming before the controls are chosen.",
"body": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Generative"
},
{
"t": "p",
"v": "Produces content, analysis or options on request. Nothing changes until a person acts on the output."
},
{
"t": "small",
"v": "**You specify:** A prompt, grounding sources and an output format."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Drafting, summarising, translating",
"big": false
},
{
"t": "title",
"v": "Analysis and option generation",
"big": false
},
{
"t": "title",
"v": "Search and question answering",
"big": false
},
{
"t": "title",
"v": "Code and test generation",
"big": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Primary failure** Plausible but wrong output accepted without review. **Tier tendency** Usually low to moderate: consequence sits with the person who acts on the output."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Controls that carry the weight"
},
{
"t": "small",
"v": "Accuracy and grounding evaluation"
},
{
"t": "small",
"v": "Citation and provenance"
},
{
"t": "small",
"v": "Disclosure that AI was used"
},
{
"t": "small",
"v": "Human review before use"
}
]
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Agentic"
},
{
"t": "p",
"v": "Takes actions across tools and systems under delegated authority, deciding its own path to a goal."
},
{
"t": "p",
"v": "**You specify:** A goal, tools, permissions, guardrails and approval thresholds."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Triage and routing",
"big": false
},
{
"t": "title",
"v": "Multi-step workflow execution",
"big": false
},
{
"t": "title",
"v": "System updates within limits",
"big": false
},
{
"t": "title",
"v": "Orchestration across services",
"big": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Primary failure** Correct-looking action taken outside intended scope, or an irreversible action taken without approval. **Tier tendency** Moderate to high: the agent itself changes the world, so authority and reversibility dominate."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Controls that carry the weight"
},
{
"t": "small",
"v": "Capability separated from authority"
},
{
"t": "small",
"v": "Least-privilege permissions and recertification"
},
{
"t": "small",
"v": "Approval gates for consequential actions"
},
{
"t": "small",
"v": "Full action logging and rollback"
}
]
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Declarative"
},
{
"t": "p",
"v": "You state the outcome, rules and boundaries; the platform resolves how they are met and keeps behaviour inside the declaration."
},
{
"t": "p",
"v": "**You specify:** Instructions, knowledge sources, permitted actions and constraints — as configuration, not code."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Configured assistants defined by a manifest",
"big": false
},
{
"t": "title",
"v": "Policy and rules expressed as configuration",
"big": false
},
{
"t": "title",
"v": "Low-code flows with declared triggers",
"big": false
},
{
"t": "title",
"v": "Templated agents published to a team",
"big": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Primary failure** The declaration drifts from intent, or nobody owns the specification that is now running in production. **Tier tendency** Varies with what is declared: the specification is the control surface, so review it as you would code."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Controls that carry the weight"
},
{
"t": "small",
"v": "Version control over the declaration"
},
{
"t": "small",
"v": "Named owner for the specification"
},
{
"t": "small",
"v": "Change assessment on every edit"
},
{
"t": "small",
"v": "Regression tests against the declared outcome"
}
]
}
],
"box": true
}
]
},
{
"t": "h",
"l": 2,
"v": "Side by side"
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Generative"
},
{
"t": "eyebrow",
"v": "Agentic"
},
{
"t": "eyebrow",
"v": "Declarative"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "What you supply",
"big": false
},
{
"t": "small",
"v": "Prompt and sources"
},
{
"t": "small",
"v": "Goal, tools and authority"
},
{
"t": "small",
"v": "Instructions, sources and constraints"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Who chooses the path",
"big": false
},
{
"t": "small",
"v": "The person"
},
{
"t": "small",
"v": "The agent, within guardrails"
},
{
"t": "small",
"v": "The platform, within the declaration"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "What changes",
"big": false
},
{
"t": "small",
"v": "Nothing until a person acts"
},
{
"t": "small",
"v": "Systems and records directly"
},
{
"t": "small",
"v": "Whatever the declaration permits"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Primary failure",
"big": false
},
{
"t": "small",
"v": "Wrong content believed"
},
{
"t": "small",
"v": "Wrong action taken"
},
{
"t": "small",
"v": "Wrong specification running"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Assurance emphasis",
"big": false
},
{
"t": "small",
"v": "Output quality and provenance"
},
{
"t": "small",
"v": "Permissions, approval and rollback"
},
{
"t": "small",
"v": "Change control over the declaration"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Register fields that matter most",
"big": false
},
{
"t": "small",
"v": "Sources, evaluation, review point"
},
{
"t": "small",
"v": "Permitted actions, authority level, escalation"
},
{
"t": "small",
"v": "Version, owner, constraints, publication scope"
}
]
},
{
"t": "h",
"l": 2,
"v": "Which class is this?"
},
{
"t": "p",
"v": "Three questions. The answer sets the controls that matter most — not whether the work is allowed."
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Does it change anything outside the conversation?",
"big": false
},
{
"t": "title",
"v": "Who decides the steps?",
"big": false
},
{
"t": "title",
"v": "Can its effects be undone without a person?",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Most deployments are all three at once"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Declarative shell"
},
{
"t": "p",
"v": "The declaration sets purpose, sources, permitted actions and limits — and is the artefact under change control."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Generative core"
},
{
"t": "p",
"v": "Inside those limits, the model drafts, summarises and reasons; evaluation and provenance apply here."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Agentic actions"
},
{
"t": "p",
"v": "Only declared actions run, each with its permission, approval threshold and audit trail."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "Record the class in the register alongside purpose and authority. It tells a reviewer which controls to look for before they open the evidence — and it makes the difference between an agent that drafts and an agent that acts visible without reading the configuration."
}
]
}
]
},
{
"id": "skills",
"part": "design-and-build",
"title": "Skill studio",
"icon": "construction",
"eyebrow": "Skill studio · modular instruction block",
"heading": "Build the instruction. Inject the skill.",
"lede": "A reusable way to brief any assistant: seven parts, a directives layer that always applies, validation that is always on, and a library loop so good prompts get kept, shared and versioned — not rediscovered.",
"body": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "The block — seven parts"
},
{
"t": "eyebrow",
"v": "Skill focus — pick any"
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "System directives — always applied"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Be accurate & truthful",
"big": false
},
{
"t": "title",
"v": "Be helpful & relevant",
"big": false
},
{
"t": "title",
"v": "Be safe & secure",
"big": false
},
{
"t": "title",
"v": "Be fair & inclusive",
"big": false
},
{
"t": "title",
"v": "Cite sources when possible",
"big": false
},
{
"t": "title",
"v": "Declare limits & assumptions",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Assembled block"
},
{
"t": "p",
"v": "SYSTEM DIRECTIVES: be accurate and truthful; be helpful and relevant; be safe and secure; be fair and inclusive; cite sources when possible; declare limits and assumptions."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Validation — always on · 0 of 8 passed"
}
]
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Governance & safety rails"
},
{
"t": "small",
"v": "**Confidentiality & privacy.** respect data boundaries and permissions"
},
{
"t": "p",
"v": "**Security & compliance.** follow policies, standards and legal requirements"
},
{
"t": "small",
"v": "**Trust & transparency.** show reasoning, sources and confidence levels"
},
{
"t": "p",
"v": "**Fairness & non-discrimination.** avoid bias; promote equity and inclusion"
},
{
"t": "small",
"v": "**Human oversight.** AI supports; humans decide and take accountability"
},
{
"t": "p",
"v": "**Monitoring & audit.** logs, reviews, feedback and continuous improvement"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Risk & harm prevention"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Misinformation",
"big": false
},
{
"t": "title",
"v": "Privacy breach",
"big": false
},
{
"t": "title",
"v": "Overreliance",
"big": false
},
{
"t": "title",
"v": "Bias & exclusion",
"big": false
},
{
"t": "title",
"v": "Unsafe content",
"big": false
},
{
"t": "title",
"v": "Hallucination",
"big": false
}
]
},
{
"t": "eyebrow",
"v": "DEIA considerations"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Accessibility by design",
"big": false
},
{
"t": "title",
"v": "Inclusive language",
"big": false
},
{
"t": "title",
"v": "Cultural & regional awareness",
"big": false
},
{
"t": "title",
"v": "Multiple perspectives",
"big": false
},
{
"t": "title",
"v": "Bias detection & mitigation",
"big": false
},
{
"t": "title",
"v": "Cognitive load & readability",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Learning & improvement loop"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Use",
"big": false
},
{
"t": "small",
"v": "Apply the prompt in real work"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Review",
"big": false
},
{
"t": "small",
"v": "Evaluate output and experience"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Refine",
"big": false
},
{
"t": "small",
"v": "Improve prompt and instructions"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Capture",
"big": false
},
{
"t": "small",
"v": "Add to library; share learnings"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Evolve",
"big": false
},
{
"t": "small",
"v": "Retire, update or version prompts"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "The reusable library holds role-based, task-based and domain prompts, scenario templates, anti-pattern examples — tested and versioned, with supersession history."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Success signals"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Useful",
"big": false
},
{
"t": "title",
"v": "Trusted",
"big": false
},
{
"t": "title",
"v": "Inclusive",
"big": false
},
{
"t": "title",
"v": "Accessible",
"big": false
},
{
"t": "title",
"v": "Traceable",
"big": false
},
{
"t": "title",
"v": "Actionable",
"big": false
},
{
"t": "title",
"v": "Impactful",
"big": false
}
]
},
{
"t": "p",
"v": "Principles at the core: human-centred · inclusive by design · transparent · accountable · reliable · respectful · secure · sustainable · equitable."
}
]
}
],
"box": false
}
]
},
{
"id": "raiagents",
"part": "design-and-build",
"title": "Agent assurance X-ray",
"icon": "smart_toy",
"eyebrow": "Responsible AI · agents and skills",
"heading": "Assure the whole human-agent system, not only the output",
"lede": "What starts an agent, how it plans, what it can reach, where people stay in control, and what it leaves behind. Skill studio builds the instruction block; Autonomy boundaries sets the limits.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "1. Trigger",
"big": false
},
{
"t": "small",
"v": "What starts the agent?"
},
{
"t": "list",
"o": false,
"items": [
"User request",
"Scheduled task",
"Event: new email, meeting or file",
"System trigger: ticket or alert",
"API or webhook",
"Another agent"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "2. Agent core",
"big": false
},
{
"t": "small",
"v": "How it thinks, plans and decides"
},
{
"t": "list",
"o": false,
"items": [
"Instructions and goals: system prompt, skills, guardrails",
"Model: chosen models and settings",
"Knowledge and context: organisational data, memory, external sources",
"Reasoning and orchestration: plans tasks, selects tools, handles errors"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "3. Tools and actions",
"big": false
},
{
"t": "small",
"v": "What it can reach and do"
},
{
"t": "list",
"o": false,
"items": [
"Search and retrieve across approved sources",
"Call APIs and applications",
"Delegate to other agents",
"Create and update content, records, messages and code"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "4. Human oversight",
"big": false
},
{
"t": "small",
"v": "Where people stay in control"
},
{
"t": "list",
"o": false,
"items": [
"Approval before high-risk actions",
"Review and validation of outputs and sources",
"Challenge and feedback",
"Stop or override at any point"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "5. Outcome",
"big": false
},
{
"t": "small",
"v": "What it delivers"
},
{
"t": "list",
"o": false,
"items": [
"Completed task or output",
"System update or action taken",
"Notification or communication",
"Log and evidence: what happened, when and why"
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Cross-cutting control rails"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Identity and permissions",
"big": false
},
{
"t": "small",
"v": "Who the agent is and what it can reach"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Information and data controls",
"big": false
},
{
"t": "small",
"v": "Classification and retention"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Monitoring and logging",
"big": false
},
{
"t": "small",
"v": "Activity, actions and decisions"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Cost and consumption",
"big": false
},
{
"t": "small",
"v": "Budgets, limits and alerts"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Security and privacy",
"big": false
},
{
"t": "small",
"v": "Risks, DPIA and incident response"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Accessibility and inclusion",
"big": false
},
{
"t": "small",
"v": "Usable for all"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Sustainability",
"big": false
},
{
"t": "small",
"v": "Environmental impact"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Kill switch and retirement",
"big": false
},
{
"t": "small",
"v": "Stop and close safely"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Risks the agent route must test"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Prompt and skill supply chain",
"big": false
},
{
"t": "p",
"v": "A shared skill or instruction changed without review changes every agent that loads it."
},
{
"t": "eyebrow",
"v": "Proposed"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "External input and prompt injection",
"big": false
},
{
"t": "p",
"v": "Documents, web pages and emails can carry instructions that redirect the agent."
},
{
"t": "eyebrow",
"v": "Proposed"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Identity, delegation and shared accounts",
"big": false
},
{
"t": "p",
"v": "Agents acting through shared or delegated contexts can reach more than the requester could."
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Unexpected discovery",
"big": false
},
{
"t": "p",
"v": "Permissions alone may not stop titles, filenames and metadata surfacing through the index."
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Multi-agent orchestration",
"big": false
},
{
"t": "p",
"v": "Agents calling agents need identity, approval points and automated records at each hop."
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Telemetry gaps",
"big": false
},
{
"t": "small",
"v": "Agent activity logs may not show enough to reconstruct what happened."
},
{
"t": "eyebrow",
"v": "Reported"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Continuous assurance"
},
{
"t": "list",
"o": true,
"items": [
[
{
"t": "title",
"v": "1. Test before deployment",
"big": false
},
{
"t": "small",
"v": "Behaviour, safety and access"
}
],
[
{
"t": "title",
"v": "2. Monitor in operation",
"big": false
},
{
"t": "small",
"v": "Usage, outcomes and incidents"
}
],
[
{
"t": "title",
"v": "3. Learn and improve",
"big": false
},
{
"t": "small",
"v": "Feedback and evaluation"
}
],
[
{
"t": "title",
"v": "4. Reassess on change",
"big": false
},
{
"t": "small",
"v": "Model, tools, data and permissions"
}
],
[
{
"t": "title",
"v": "5. Retire when no longer needed",
"big": false
},
{
"t": "small",
"v": "Revoke access, close data, capture lessons"
}
]
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Prompts and skills as organisational products"
},
{
"t": "p",
"v": "A personal prompt stays personal. Once a prompt, skill or agent instruction is shared, it needs an owner, a version, a test set, security checks, a cost profile and a retirement date, and it goes in the prompt and skills register."
},
{
"t": "list",
"o": false,
"items": [
"Is the prompt or skill personal or organisational?",
"Does it need version control and approval?",
"Has it been tested for safety, consistency and cost?",
"Could deterministic automation do the job better?",
"Can it be shared or changed without review?",
"How will it be withdrawn?"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Deterministic alternative test"
},
{
"t": "title",
"v": "Before building a generative skill or agent, check whether conventional automation is safer, cheaper and more consistent.",
"big": false
},
{
"t": "p",
"v": "If a rule, form, flow or script can do the job reliably, use it. Generative AI earns its place where judgement over unstructured material is the actual task."
}
]
}
],
"box": false
}
]
},
{
"id": "raiconnectors",
"part": "design-and-build",
"title": "Connectors, MCP & integrations",
"icon": "hub",
"eyebrow": "Design and build · connectors, MCP and integrations",
"heading": "Every connector is a new data path and, often, a new action",
"lede": "MCP servers, connectors, plugins, add-ins and APIs turn a generative tool into one that can reach, and change, other systems. Each needs an owner, a scope and a log.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "hub"
},
{
"t": "p",
"v": "**MCP server**An open-protocol endpoint that exposes tools and resources to any MCP-capable client. Tools can be discovered at runtime, so capability can change without the agent being republished."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "cable"
},
{
"t": "p",
"v": "**Connector**Brings an external system’s content into an AI tool’s index or grounding, such as Copilot connectors."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "extension"
},
{
"t": "p",
"v": "**Plugin or action**A packaged custom action, such as an MCP or API plugin in a declarative agent, that lets an agent read or change data."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "apps"
},
{
"t": "p",
"v": "**Add-in or app**An extension that runs inside a host application, such as an Office add-in or a chat app."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "api"
},
{
"t": "p",
"v": "**API and webhook**Direct integration through REST APIs, OpenAPI descriptions or event webhooks."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "router"
},
{
"t": "p",
"v": "**Integrator or gateway**A platform that brokers many connections, applies policy and logs traffic, such as an MCP gateway or integration platform."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Platform directory"
},
{
"t": "p",
"v": "Graded claims from supplier documentation. Where a route is unresolved, confirm it with the supplier before connecting."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Microsoft 365, Copilot and Copilot Studio",
"big": false
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "p",
"v": "Declarative agents call MCP servers and REST APIs through MCP and API plugins. Copilot Studio adds MCP servers through an onboarding wizard, and servers can be registered in Agent 365."
},
{
"t": "title",
"v": "Routes: MCP plugin · API plugin · Copilot connectors · Power Platform connectors",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Tools are discovered at runtime unless pinned in the manifest. Register MCP servers in Agent 365 and review each tool’s read and write scope."
},
{
"t": "title",
"v": "MCP and API plugins for declarative agents ↗",
"big": false
},
{
"t": "title",
"v": "Connect Copilot Studio to an MCP server ↗",
"big": false
},
{
"t": "title",
"v": "Declarative agents for Microsoft 365 Copilot ↗",
"big": false
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "GitHub",
"big": false
},
{
"t": "eyebrow",
"v": "Verified"
}
]
},
{
"t": "p",
"v": "GitHub maintains a remote MCP server for repositories, issues and pull requests. Microsoft’s walkthrough wraps it as a plugin in a declarative agent."
},
{
"t": "title",
"v": "Routes: Remote MCP server · GitHub Apps · Actions",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** One OAuth app per organisation, repositories scoped, no write tools without approval, and no secrets in prompts."
},
{
"t": "title",
"v": "Build a declarative agent plugin from the GitHub MCP server ↗",
"big": false
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Atlassian: Jira, Confluence, Jira Service Management, Bitbucket, Loom",
"big": false
},
{
"t": "eyebrow",
"v": "Corroborated"
}
],
"box": false
},
{
"t": "p",
"v": "The official Rovo MCP server reached general availability on 4 February 2026. It is cloud-hosted and authenticates with OAuth 2.1 or API tokens."
},
{
"t": "title",
"v": "Routes: Remote MCP server, v2 endpoint · Rovo agents",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Admins choose which AI tools and domains may connect, destructive tool groups stay off until enabled, and every tool call is written to the audit log. Atlassian Cloud only, not Data Center."
},
{
"t": "title",
"v": "Getting started with the Rovo MCP server ↗",
"big": false
},
{
"t": "title",
"v": "atlassian-mcp-server on GitHub ↗",
"big": false
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Anthropic Claude",
"big": false
},
{
"t": "eyebrow",
"v": "Reported"
}
]
},
{
"t": "p",
"v": "Claude connects to Microsoft 365 and other systems through connectors built on MCP. Its Microsoft 365 connector reads Outlook, Teams, SharePoint and OneDrive and writes to the last two."
},
{
"t": "title",
"v": "Routes: Connectors · remote MCP servers · Skills",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Set connector write scope per user, keep an admin allow-list, and record each connector in the data-source register."
},
{
"t": "small",
"v": "No verified documentation linked yet."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Google Workspace and Gemini",
"big": false
},
{
"t": "eyebrow",
"v": "Unresolved"
}
]
},
{
"t": "p",
"v": "Gemini works with Workspace content and connected data. Confirm the current MCP and extension routes with Google before use."
},
{
"t": "title",
"v": "Routes: Workspace grounding · extensions · APIs",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Confirm data location, admin controls and audit logging for each extension."
},
{
"t": "small",
"v": "No verified documentation linked yet."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Slack",
"big": false
},
{
"t": "eyebrow",
"v": "Unresolved"
}
]
},
{
"t": "p",
"v": "Slack can be a source or a destination for agent work. Confirm the official MCP or app route and its data residency before connecting."
},
{
"t": "title",
"v": "Routes: App · MCP server · webhooks",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Scope channels, keep private channels out without approval, and align retention."
},
{
"t": "small",
"v": "No verified documentation linked yet."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Service and line-of-business systems",
"big": false
},
{
"t": "eyebrow",
"v": "Proposed"
}
]
},
{
"t": "p",
"v": "ServiceNow, Salesforce, Dynamics 365 and case-management systems publish connectors, APIs or MCP servers. Each is a new data path and a new action surface."
},
{
"t": "title",
"v": "Routes: Connector · API plugin · MCP server · integration platform",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Treat any write tool as Act class, require approval before change, and log to the incident trail."
},
{
"t": "small",
"v": "No verified documentation linked yet."
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Model Context Protocol, the standard",
"big": false
},
{
"t": "eyebrow",
"v": "Corroborated"
}
]
},
{
"t": "p",
"v": "An open standard that lets an AI client call tools and read resources on an outside system through one interface."
},
{
"t": "title",
"v": "Routes: Specification · registry · gateways",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Controls:** Prefer official servers, pin tools where possible, review tool descriptions for injected instructions and route calls through a gateway that logs them."
},
{
"t": "title",
"v": "modelcontextprotocol.io ↗",
"big": false
}
]
}
],
"box": true
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Nine controls for every connector, plugin and MCP server"
},
{
"t": "list",
"o": true,
"items": [
"Register every connector, plugin and MCP server in the tool and data-source registers, with a named owner.",
"Record read and write scope separately. Any write tool makes the use case at least Act class.",
"Prefer official, supplier-maintained servers over community ones, and record which is in use.",
"Pin tools or versions where the platform allows. Dynamic discovery means capability can change without republishing.",
"Authenticate with OAuth and least privilege. No shared service accounts for user-level data.",
"Log every tool call to an audit trail the service owner can read.",
"Review tool descriptions and returned content for prompt injection.",
"Check where each connected system stores and processes data, and whether it sits inside your data boundary.",
"Keep a revoke route: users can withdraw authorisations and admins can disable a connector at once."
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "A connector can change the capability class"
},
{
"t": "p",
"v": "A generative chat tool with a read connector is Connect class. Add a write tool and it is Act class; let it run on a schedule and it is Persist. Reclassify whenever a connector or tool is added, and record it in the change register."
}
]
}
]
},
{
"id": "architecture",
"part": "design-and-build",
"title": "Architecture & quality",
"icon": "gavel",
"eyebrow": "Architecture & quality",
"heading": "Principles with tests, gates with teeth",
"lede": "A three-tier canon governs everything that influences how something is built — prototypes explicitly included, because most principle breaches enter the estate through prototypes that were never intended to survive, and then did.",
"body": [
{
"t": "h",
"l": 2,
"v": "Tier 1 — mandated, cited not restated"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "T1-01 Secure by Design",
"big": false
},
{
"t": "title",
"v": "T1-02 CAF & GovAssure",
"big": false
},
{
"t": "title",
"v": "T1-03 Technology Code of Practice",
"big": false
},
{
"t": "title",
"v": "T1-04 GDS Service Standard",
"big": false
},
{
"t": "title",
"v": "T1-05 AI Playbook",
"big": false
},
{
"t": "title",
"v": "T1-06 Data protection & public records",
"big": false
},
{
"t": "title",
"v": "T1-07 WCAG 2.2 AA",
"big": false
}
]
},
{
"t": "p",
"v": "Tier 1 conformance is binary: the principle applies and is met, or a recorded exception exists. No informal partial conformance."
},
{
"t": "h",
"l": 2,
"v": "Tier 2 — the core seven, each with a test"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-01",
"big": false
},
{
"t": "title",
"v": "Federated, not centralised",
"big": false
},
{
"t": "p",
"v": "**Test:** Can each element be traced to a named authoritative owner outside the central platform?**Looks compliant, is not:** \"We will federate later, once everything is in one place.\""
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-02",
"big": false
},
{
"t": "title",
"v": "Discovery separated from access",
"big": false
},
{
"t": "p",
"v": "**Test:** Can a user discover an asset exists without thereby gaining access to it?**Looks compliant, is not:** Endpoint or credential details published as discovery metadata for convenience."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-03",
"big": false
},
{
"t": "title",
"v": "Human-in-the-loop by design",
"big": false
},
{
"t": "p",
"v": "**Test:** Can any automated component write to the record without recorded human approval?**Looks compliant, is not:** \"The user can always review it\" — optional review presented as a control."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-04",
"big": false
},
{
"t": "title",
"v": "Traceability & provenance mandatory",
"big": false
},
{
"t": "p",
"v": "**Test:** Take any output at random: can you reconstruct inputs, owner, decision and evidence without asking a person?**Looks compliant, is not:** Evidence linked at closure rather than at capture."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-05",
"big": false
},
{
"t": "title",
"v": "Reuse before build",
"big": false
},
{
"t": "p",
"v": "**Test:** Is there a recorded reuse assessment that predates the build decision?**Looks compliant, is not:** Reuse assessment performed after the build decision, to justify it."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-06",
"big": false
},
{
"t": "title",
"v": "Accessible & inclusive by design",
"big": false
},
{
"t": "p",
"v": "**Test:** Is accessibility evidence produced before release? Does every status carry a non-colour cue?**Looks compliant, is not:** \"We will run an accessibility audit before go-live\" as the only accessibility activity."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "AP-07",
"big": false
},
{
"t": "title",
"v": "Guardrails enforced, not documented",
"big": false
},
{
"t": "p",
"v": "**Test:** Attempt to breach the control through the normal interface. Does it succeed?**Looks compliant, is not:** A mandatory gate with no exception path — guaranteeing informal workarounds."
}
],
"box": true
},
{
"t": "p",
"v": "**Prototype promotion gate.** A prototype may not be promoted toward live use until Tier 1 and Tier 2 conformance is assessed and recorded. Record intended lifespan and disposal point at creation; record deferred principles; re-assess them at the gate. Human-in-the-loop applies in full even at prototype stage — autonomy introduced early is rarely removed later. Promotion is a decision, not a drift."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "The assurance-gated agentic pattern"
},
{
"t": "p",
"v": "From the AI Engineering Lab's reference implementation: the transferable innovation is not any one use case — it is the operating model around it. Specialist agents routed by an orchestrator; a challenger agent that reviews outputs; a compliance agent; assurance gates executed in the workflow; immutable audit trails; human approval before anything writes to the record. The same pattern serves capability management, catalogue operations, workforce planning and architecture governance."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Specialist agents",
"big": false
},
{
"t": "title",
"v": "Orchestrator routing",
"big": false
},
{
"t": "title",
"v": "Challenger agent",
"big": false
},
{
"t": "title",
"v": "Compliance agent",
"big": false
},
{
"t": "title",
"v": "Assurance gates in-workflow",
"big": false
},
{
"t": "title",
"v": "Immutable audit trail",
"big": false
},
{
"t": "title",
"v": "Human approval to write",
"big": false
},
{
"t": "title",
"v": "Escalation paths",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "The quality gate — score prototypes like you mean it"
},
{
"t": "p",
"v": "A gate, not a form: run at the end of every iteration. Tier 0 is the brake — the only tier that can argue for less; a veto is the instrument doing its job. Scores are capped by evidence strength, unscoreable items go on an open register published with the score, and demo drift — a level-3 prototype someone useful quietly relies on — is the most common governance failure in this class of work."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Quick index — ten dimensions, 1–5"
},
{
"t": "title",
"v": "Security",
"big": false
},
{
"t": "title",
"v": "Privacy & data protection",
"big": false
},
{
"t": "title",
"v": "Accessibility",
"big": false
},
{
"t": "title",
"v": "Usability",
"big": false
},
{
"t": "title",
"v": "Performance & resilience",
"big": false
},
{
"t": "title",
"v": "Maintainability",
"big": false
},
{
"t": "title",
"v": "Observability",
"big": false
},
{
"t": "title",
"v": "Data quality & provenance",
"big": false
},
{
"t": "title",
"v": "Trust & transparency",
"big": false
},
{
"t": "title",
"v": "Sustainability & handover",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "—",
"big": true
},
{
"t": "title",
"v": "Sophistication index — Σ scores ÷ 50 × 100",
"big": false
},
{
"t": "small",
"v": "Score the dimensions for a reading"
}
]
},
{
"t": "p",
"v": "**The ladder:** Tier 0 proportionality gate → Tiers 1–6 architecture compass → Tier 7 self-* autonomic → Tier 8 human & institutional sustainability. Evidence-rung caps; accessibility floors at human testing, not automated checks; an empty removal ledger three iterations running is itself a Tier 0 veto. Dimensions here are representative — attach the full template for real assessments."
}
]
}
],
"box": false
}
]
},
{
"id": "knowledge",
"part": "design-and-build",
"title": "Knowledge foundation",
"icon": "menu_book",
"eyebrow": "Knowledge foundation",
"heading": "From weekly call to institutional memory",
"lede": "Transforming conversations into trusted, discoverable knowledge that drives action and public value. Capture once, reuse everywhere — the seven-step pipeline and the record it produces.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "1. Capture",
"big": false
},
{
"t": "title",
"v": "Capture the moment",
"big": false
},
{
"t": "small",
"v": "Record meeting"
},
{
"t": "small",
"v": "Capture chat"
},
{
"t": "small",
"v": "Collect shared files"
},
{
"t": "small",
"v": "Note live questions"
},
{
"t": "small",
"v": "Identify demos"
},
{
"t": "title",
"v": "Raw recording, chat and artefacts",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "2. Transcribe",
"big": false
},
{
"t": "title",
"v": "Make it searchable",
"big": false
},
{
"t": "small",
"v": "Generate transcript"
},
{
"t": "small",
"v": "Speaker identification"
},
{
"t": "small",
"v": "Time-stamped content"
},
{
"t": "small",
"v": "Redact sensitive information"
},
{
"t": "title",
"v": "Transcript with timestamps",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "3. Summarise",
"big": false
},
{
"t": "title",
"v": "Extract the essence",
"big": false
},
{
"t": "small",
"v": "Executive summary"
},
{
"t": "small",
"v": "Key points"
},
{
"t": "small",
"v": "Decisions"
},
{
"t": "small",
"v": "Unresolved questions"
},
{
"t": "small",
"v": "Action items"
},
{
"t": "title",
"v": "Structured summary and outcomes",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "4. Curate",
"big": false
},
{
"t": "title",
"v": "Add context and quality",
"big": false
},
{
"t": "small",
"v": "Add metadata and taxonomy"
},
{
"t": "small",
"v": "Link to related topics"
},
{
"t": "small",
"v": "Classify sensitivity"
},
{
"t": "small",
"v": "Verify facts and sources"
},
{
"t": "small",
"v": "Assign owners"
},
{
"t": "title",
"v": "Curated, classified session record",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "5. Connect",
"big": false
},
{
"t": "title",
"v": "Link and enrich",
"big": false
},
{
"t": "small",
"v": "Link to use cases, agents and data assets"
},
{
"t": "small",
"v": "Classify sensitivity"
},
{
"t": "small",
"v": "Verify facts and sources"
},
{
"t": "small",
"v": "Assign owners"
},
{
"t": "title",
"v": "Connected knowledge with relationships",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "6. Publish",
"big": false
},
{
"t": "title",
"v": "Make it discoverable",
"big": false
},
{
"t": "small",
"v": "Publish to knowledge hub"
},
{
"t": "small",
"v": "Update indexes"
},
{
"t": "small",
"v": "Set permissions"
},
{
"t": "small",
"v": "Notify communities"
},
{
"t": "small",
"v": "Version control"
},
{
"t": "title",
"v": "Discoverable knowledge asset",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "7. Reuse & learn",
"big": false
},
{
"t": "title",
"v": "Drive action and value",
"big": false
},
{
"t": "small",
"v": "Apply in projects and services"
},
{
"t": "small",
"v": "Share improvements"
},
{
"t": "small",
"v": "Capture feedback and results"
},
{
"t": "small",
"v": "Update and improve"
},
{
"t": "title",
"v": "Impact, feedback and new learning",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What we capture"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Discussions and insights",
"big": false
},
{
"t": "title",
"v": "Questions and challenges",
"big": false
},
{
"t": "title",
"v": "Decisions and agreements",
"big": false
},
{
"t": "title",
"v": "Actions and owners",
"big": false
},
{
"t": "title",
"v": "Shared links and resources",
"big": false
},
{
"t": "title",
"v": "Demos and use cases",
"big": false
},
{
"t": "title",
"v": "Risks and issues",
"big": false
},
{
"t": "title",
"v": "Ideas and opportunities",
"big": false
},
{
"t": "title",
"v": "Lessons learned",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers of institutional memory"
},
{
"t": "small",
"v": "Clear governance and ownership"
},
{
"t": "small",
"v": "Consistent metadata and taxonomy"
},
{
"t": "small",
"v": "Security, privacy and access control"
},
{
"t": "small",
"v": "Quality assurance and validation"
},
{
"t": "small",
"v": "Retention and lifecycle management"
},
{
"t": "small",
"v": "Community curation and care"
},
{
"t": "small",
"v": "Usage analytics and feedback"
},
{
"t": "small",
"v": "Integrated tools and platforms"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "The benefits"
},
{
"t": "p",
"v": "**Easier to find**Searchable, organised knowledge at your fingertips"
},
{
"t": "p",
"v": "**Avoid duplication**Learn from others' experience and mistakes"
},
{
"t": "p",
"v": "**Stronger decisions**Better informed choices backed by evidence"
},
{
"t": "p",
"v": "**Faster delivery**Reuse patterns, guidance and assets to go faster"
},
{
"t": "p",
"v": "**Assured and trusted**Quality, governance and provenance you can rely on"
},
{
"t": "p",
"v": "**Greater public value**Better services, outcomes and trust for citizens"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The session evidence pack"
},
{
"t": "p",
"v": "Twelve components that turn a conversation into a reusable organisational asset. Session record · community owned · governed and secure · discoverable and reusable · continuously improved."
},
{
"t": "table",
"rows": [
[
{
"h": false,
"v": "**1. Session details** Title & description · Date, time, duration · Community / forum · Participants & roles · Host / facilitators"
},
{
"h": false,
"v": "**2. Recording & transcript** Recording link · Transcript / captions · Time-stamped version · Accessibility info"
},
{
"h": false,
"v": "**3. Summary & key points** Executive summary · Key discussion points · Important context · What was agreed"
},
{
"h": false,
"v": "**4. Questions & answers** Questions raised · Answers provided · Outstanding questions · Follow-up needed"
}
],
[
{
"h": false,
"v": "**5. Demonstrations & artefacts** Demos & walkthroughs · Screens / diagrams · Code / configs · Resources shared"
},
{
"h": false,
"v": "**6. Decisions & agreements** Decisions taken · Agreements reached · Scope confirmations · Changes approved"
},
{
"h": false,
"v": "**7. Actions & owners** Action items · Owners · Due dates · Status / progress"
},
{
"h": false,
"v": "**8. Linked resources & references** Related publications · Previous sessions · External references"
}
],
[
{
"h": false,
"v": "**9. Classification & sensitivity** Information classification · Sensitivity level · Handling instructions · Access permissions"
},
{
"h": false,
"v": "**10. Evidence & insights** Evidence referenced · Impact / value assessed · Metrics mentioned · Outcomes expected"
},
{
"h": false,
"v": "**11. Lessons & insights** Key takeaways · Update history · Supersedes / superseded · Recommendations"
},
{
"h": false,
"v": "**12. Review & lifecycle** Review cycle · Update history · Supersedes / superseded · Retention"
}
]
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How it flows"
},
{
"t": "small",
"v": "**1. Capture** — Capture during or after the session"
},
{
"t": "small",
"v": "**2. Structure** — Organise using the standard evidence pack template"
},
{
"t": "small",
"v": "**3. Classify** — Apply classification, tags and sensitivity controls"
},
{
"t": "small",
"v": "**4. Publish** — Store in governed repositories and make discoverable"
},
{
"t": "small",
"v": "**5. Link** — Connect to related assets, agents, use cases, and more"
},
{
"t": "small",
"v": "**6. Reuse** — Teams find, reuse and build on the knowledge"
},
{
"t": "small",
"v": "**7. Improve** — Feedback and updates improve content continuously"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence status on every claim"
},
{
"t": "p",
"v": "**Verified**Independently confirmed"
},
{
"t": "p",
"v": "**Reported / emerging**Reported by a source, not yet validated"
},
{
"t": "p",
"v": "**In progress / unvalidated**Being worked through"
},
{
"t": "p",
"v": "**Contradictory / needs review**Sources disagree"
},
{
"t": "p",
"v": "**Outdated / superseded**Replaced by newer content"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The knowledge lifecycle"
},
{
"t": "p",
"v": "Good knowledge is captured once, curated well, reused often and improved continuously — and its status is always visible."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Create & capture** **1. Capture** Capture knowledge, evidence and experiences from sessions, projects or operations."
},
{
"t": "p",
"v": "**Create & capture** **2. Classify & tag** Add metadata, taxonomy, topics, sensitivity and links to related assets."
},
{
"t": "p",
"v": "**Curate & govern** **3. Curate & quality** Check accuracy, completeness, relevance and quality. Remove duplicates."
},
{
"t": "p",
"v": "**Curate & govern** **4. Validate & approve** Subject matter and governance review. Approve for use and assign lifecycle status."
},
{
"t": "p",
"v": "**Share & use** **5. Publish & discover** Publish to trusted repositories and make discoverable through search, portals and APIs."
},
{
"t": "p",
"v": "**Share & use** **6. Use & reuse** Apply knowledge in work, decisions and services. Share feedback and outcomes."
},
{
"t": "p",
"v": "**Share & use** **7. Review & update** Review on a regular cycle or when triggered by change. Update content, links, owners and status."
},
{
"t": "p",
"v": "**Share & use** **8. Sunset, supersede or archive** Retire outdated content, mark as superseded or archive with retention and audit trail."
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence status through the lifecycle"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Draft** Newly captured. Not yet reviewed."
},
{
"t": "p",
"v": "**Reviewed** Peer or owner reviewed. Accuracy and relevance checked."
},
{
"t": "p",
"v": "**Validated** Approved by subject matter and governance for use."
},
{
"t": "p",
"v": "**Published** Officially published and visible to intended audience."
},
{
"t": "p",
"v": "**Superseded** Outdated or replaced by newer content. Retained for reference."
},
{
"t": "p",
"v": "**Archived** No longer active. Retained in accordance with policy."
}
],
"box": false
}
]
},
{
"t": "h",
"l": 2,
"v": "One discovery layer, many trusted sources"
},
{
"t": "p",
"v": "Information is scattered across teams, tools and tenancies, which causes duplication, inconsistency and poor discoverability. Federate rather than migrate: connect the sources, add a metadata and governance layer, and put one discovery layer on top."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Today — fragmented and hard to find"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Chats & calls",
"big": false
},
{
"t": "title",
"v": "SharePoint sites",
"big": false
},
{
"t": "title",
"v": "Personal OneDrives",
"big": false
},
{
"t": "title",
"v": "Email & attachments",
"big": false
},
{
"t": "title",
"v": "Departmental file shares",
"big": false
},
{
"t": "title",
"v": "GitHub repositories",
"big": false
},
{
"t": "title",
"v": "Slack / other channels",
"big": false
},
{
"t": "title",
"v": "Knowledge hub",
"big": false
},
{
"t": "title",
"v": "Spreadsheets & lists",
"big": false
},
{
"t": "title",
"v": "Recordings & transcripts",
"big": false
},
{
"t": "title",
"v": "Documents & guidance",
"big": false
},
{
"t": "title",
"v": "External links & references",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Target — federated knowledge service"
},
{
"t": "p",
"v": "**Unified discovery layer**Single search & navigation · Smart filters & facets · Recommendations & related content · Personalised views"
},
{
"t": "p",
"v": "**Metadata & governance layer**Common taxonomy · Metadata standards · Classification & sensitivity · Ownership & accountability · Lifecycle management"
},
{
"t": "p",
"v": "**Federated source connectors**Source systems remain in place — connect, do not migrate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Tomorrow — connected and valuable"
},
{
"t": "p",
"v": "**Find faster**One place to search across the estate with better results"
},
{
"t": "p",
"v": "**Trust & assurance**Clear ownership, quality signals, classification and provenance"
},
{
"t": "p",
"v": "**Reuse & collaborate**Easily share, reference and reuse patterns, guidance and code"
},
{
"t": "p",
"v": "**Learn & improve**Insights and lessons flow back to improve practice continuously"
},
{
"t": "p",
"v": "**Deliver more value**Better decisions, less duplication, faster delivery and stronger outcomes"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The agent information supply chain"
},
{
"t": "p",
"v": "High-quality information in, trusted retrieval and reasoning, accurate answers or actions, human review, better sources. Every link has a quality gate."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "1. Authoritative sources",
"big": false
},
{
"t": "small",
"v": "Trusted, approved information assets"
},
{
"t": "small",
"v": "Policies & procedures"
},
{
"t": "small",
"v": "Knowledge bases"
},
{
"t": "small",
"v": "Case law & guidance"
},
{
"t": "small",
"v": "Data & reference sets"
},
{
"t": "small",
"v": "Forms & templates"
},
{
"t": "small",
"v": "Expert directories"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "2. Metadata & discovery",
"big": false
},
{
"t": "small",
"v": "Make content findable and understandable"
},
{
"t": "small",
"v": "Metadata & tagging"
},
{
"t": "small",
"v": "Taxonomy & keywords"
},
{
"t": "small",
"v": "Semantic models"
},
{
"t": "small",
"v": "Lineage & provenance"
},
{
"t": "small",
"v": "Classification & sensitivity"
},
{
"t": "small",
"v": "Permissions & access"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "3. Retrieval",
"big": false
},
{
"t": "small",
"v": "Find the right information in the right context"
},
{
"t": "small",
"v": "Hybrid search (keyword + semantic)"
},
{
"t": "small",
"v": "Filters (date, owner, source, role)"
},
{
"t": "small",
"v": "Rank by relevance & authority"
},
{
"t": "small",
"v": "Permission check"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "4. Agent reasoning & processing",
"big": false
},
{
"t": "small",
"v": "Synthesise and reason using retrieved content"
},
{
"t": "small",
"v": "Understand intent"
},
{
"t": "small",
"v": "Ground in source content"
},
{
"t": "small",
"v": "Reason & generate"
},
{
"t": "small",
"v": "Cite sources"
},
{
"t": "small",
"v": "Guardrails & policies"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "5. Response / action proposal",
"big": false
},
{
"t": "small",
"v": "Answer with citations, drafts or recommended actions"
},
{
"t": "small",
"v": "Answer with citations"
},
{
"t": "small",
"v": "Draft content"
},
{
"t": "small",
"v": "Suggest next steps"
},
{
"t": "small",
"v": "Propose system actions"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "6. Human review & oversight",
"big": false
},
{
"t": "small",
"v": "Humans verify, approve and take responsibility"
},
{
"t": "small",
"v": "Check accuracy"
},
{
"t": "small",
"v": "Validate sources"
},
{
"t": "small",
"v": "Confirm applicability"
},
{
"t": "small",
"v": "Decision logged"
},
{
"t": "small",
"v": "Approve / reject"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "7. Feedback & capture",
"big": false
},
{
"t": "small",
"v": "Capture feedback, use and outcomes"
},
{
"t": "small",
"v": "Was it useful?"
},
{
"t": "small",
"v": "Was it correct?"
},
{
"t": "small",
"v": "What was missing?"
},
{
"t": "small",
"v": "New knowledge identified"
},
{
"t": "small",
"v": "Add & monitor"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "8. Source improvement & governance",
"big": false
},
{
"t": "small",
"v": "Improve sources, policies and models"
},
{
"t": "small",
"v": "Update content"
},
{
"t": "small",
"v": "Fix gaps"
},
{
"t": "small",
"v": "Improve metadata"
},
{
"t": "small",
"v": "Retire outdated"
},
{
"t": "small",
"v": "Review & approve"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key metrics for the chain"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Answer accuracy rate",
"big": false
},
{
"t": "title",
"v": "Citation coverage",
"big": false
},
{
"t": "title",
"v": "User satisfaction score",
"big": false
},
{
"t": "title",
"v": "Time to answer",
"big": false
},
{
"t": "title",
"v": "Re-use rate of content",
"big": false
},
{
"t": "title",
"v": "Feedback captured",
"big": false
},
{
"t": "title",
"v": "Source improvements",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Where should the information live?"
},
{
"t": "p",
"v": "Work down the questions in order. The first \"yes\" is your answer — the goal is the right information, to the right people, at the right time, securely and efficiently."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "1. Is the information already in an authoritative source (for example an existing SharePoint site, policy library, DMS)?",
"big": false
},
{
"t": "p",
"v": "**Yes →** Use existing authoritative location — keep content in place, ensure agent uses existing permissions, maintain single source of truth, lower duplication and maintenance risk."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "2. Can the agent be granted access via existing permissions?",
"big": false
},
{
"t": "p",
"v": "**Yes →** Grant access via existing permissions — respect least privilege, use groups and roles already in place, review access periodically."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "3. Can permissions be adjusted for a new site (or team) without creating excessive oversharing or security risk?",
"big": false
},
{
"t": "p",
"v": "**Yes →** Adjust permissions or create a secure location — create dedicated site or library if needed, configure granular permissions, review with the information owner."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "4. Is copying or moving content necessary for the agent to function effectively?",
"big": false
},
{
"t": "p",
"v": "**Yes →** Copy or move content with strong controls — risk of duplication and divergence, increased maintenance and update process, maintain links back to source of truth, confirm retention and records handling."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "If every answer is no — reconsider the approach"
},
{
"t": "small",
"v": "Use metadata, views, filters or connectors"
},
{
"t": "small",
"v": "Use automated search over authoritative sources"
},
{
"t": "small",
"v": "Build a knowledge layer or index"
},
{
"t": "small",
"v": "Avoid unnecessary duplication"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key considerations"
},
{
"t": "small",
"v": "**Audience** — Who will use the agent and what access do they need?"
},
{
"t": "small",
"v": "**Permissions** — Does the agent respect existing SharePoint permissions?"
},
{
"t": "p",
"v": "**Information governance** — Is the content authoritative, current and controlled?"
},
{
"t": "p",
"v": "**Duplication risk** — Will copying create multiple versions or conflicting answers?"
},
{
"t": "p",
"v": "**Maintenance** — Can the team keep copies current and in sync with the source?"
},
{
"t": "p",
"v": "**Licensing & cost** — Do users have the right licences? Any connector or storage costs?"
},
{
"t": "p",
"v": "**Records & retention** — Where is the record of business? How long must it be kept?"
},
{
"t": "p",
"v": "**Security & risk** — Does the agent minimise risk of overexposure or data leakage?"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "A knowledge graph and a shared prompt layer"
},
{
"t": "p",
"v": "Connecting people, communities, data, decisions and capabilities to power trusted AI — and a common language layer that improves responses, saves time and builds organisational memory."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What the graph connects"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Questions it answers"
},
{
"t": "small",
"v": "Who owns this AI agent and what use cases is it used for?"
},
{
"t": "small",
"v": "What data sources ground the answers provided by this agent?"
},
{
"t": "small",
"v": "Which policies and controls apply to this use case?"
},
{
"t": "small",
"v": "What outcomes are we expecting and how are we measuring them?"
},
{
"t": "small",
"v": "What other capabilities or agents could be reused?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Why a knowledge graph"
},
{
"t": "p",
"v": "**Connects what sits apart**Links people, data, agents and outcomes"
},
{
"t": "p",
"v": "**Improves discovery and reuse**Find what already exists before building"
},
{
"t": "p",
"v": "**Provides context for AI**Grounded, related, permissioned context"
},
{
"t": "p",
"v": "**Supports governance, assurance and value**Ownership, lineage and measurable benefit"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Organisational prompt library** Curated collection of high-quality prompts · Categorised by task, role and domain · Example prompts and use guidance"
},
{
"t": "p",
"v": "**2. Prompt lab & testing** Experiment and refine prompts safely · Compare results and measure quality · Promote best prompts to the library"
},
{
"t": "p",
"v": "**3. Business glossary & acronym dictionary** Common definitions and terminology · Acronyms, abbreviations and plain-government terms · Versioned and community-maintained"
},
{
"t": "p",
"v": "**4. Domain & policy knowledge packs** Topical guidance, policies, frameworks · Approved content summaries for AI use · Content packs for specific domains"
},
{
"t": "p",
"v": "**5. Technology-agnostic prompt patterns** Reusable patterns that work across tools and models · Task-based templates and structures · Adapts to different AI platforms and models"
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How the layer works"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Create & contribute** Users contribute prompts, definitions and knowledge"
},
{
"t": "p",
"v": "**2. Curate & review** Experts review for quality, accuracy and safety"
},
{
"t": "p",
"v": "**3. Publish & tag** Approved items are tagged and published to the layer"
},
{
"t": "p",
"v": "**4. Use & apply** AI tools and users discover and apply in their work"
},
{
"t": "p",
"v": "**5. Feedback & improve** Results and feedback feed back to improve content"
}
],
"box": false
}
]
}
]
},
{
"id": "autonomy",
"part": "design-and-build",
"title": "Autonomy boundaries",
"icon": "tune",
"eyebrow": "Responsible autonomy boundaries",
"heading": "The right level of autonomy for the right task at the right risk",
"lede": "Autonomy is a privilege, not a default. Human oversight increases with risk, impact and consequence — and the human role is named at every level, not implied.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Human control**People remain accountable"
},
{
"t": "p",
"v": "**Proportionate by risk**Match autonomy to impact"
},
{
"t": "p",
"v": "**Transparent & explainable**Make actions understandable"
},
{
"t": "p",
"v": "**Safe by design & default**Guardrails built in"
},
{
"t": "p",
"v": "**Continuously monitored**Detect, learn and improve"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "LOW AUTONOMY"
},
{
"t": "eyebrow",
"v": "HIGH AUTONOMY"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Level 2 of 6"
},
{
"t": "h",
"l": 2,
"v": "Recommend"
},
{
"t": "title",
"v": "Suggest options",
"big": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "small",
"v": "Agent recommends options or actions for human consideration."
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Recommend next action"
},
{
"t": "small",
"v": "Suggest content or draft"
},
{
"t": "small",
"v": "Flag risks or anomalies"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "The human role"
},
{
"t": "title",
"v": "Human decision",
"big": false
},
{
"t": "small",
"v": "User reviews and decides"
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Factors that increase autonomy"
},
{
"t": "small",
"v": "Low risk to people"
},
{
"t": "small",
"v": "Reversible outcomes"
},
{
"t": "small",
"v": "High data quality"
},
{
"t": "small",
"v": "Clear objectives"
},
{
"t": "small",
"v": "Strong monitoring"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Factors that reduce autonomy"
},
{
"t": "small",
"v": "High impact or risk"
},
{
"t": "small",
"v": "Irreversible outcomes"
},
{
"t": "small",
"v": "Sensitive data"
},
{
"t": "small",
"v": "Uncertain environments"
},
{
"t": "small",
"v": "Weak monitoring"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Decision factors"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "crisis"
},
{
"t": "title",
"v": "Risk & impact",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "account_balance"
},
{
"t": "title",
"v": "Legal / regulatory",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lock"
},
{
"t": "title",
"v": "Data sensitivity",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "hub"
},
{
"t": "title",
"v": "Complexity",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "repeat"
},
{
"t": "title",
"v": "Frequency",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "volunteer_activism"
},
{
"t": "title",
"v": "Public trust",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Guardrails always on, and the override set"
},
{
"t": "title",
"v": "0 of 7 guardrails and 0 of 4 override controls in place. Guardrails are always on — autonomy is a privilege, not a default.",
"big": false
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "GUARDRAILS (ALWAYS ON)"
},
{
"t": "eyebrow",
"v": "ESCALATION & OVERRIDE"
}
]
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Accountability"
},
{
"t": "icon",
"v": "person"
},
{
"t": "title",
"v": "Named owners for agents",
"big": false
},
{
"t": "icon",
"v": "gavel"
},
{
"t": "title",
"v": "Clear responsibility for decisions",
"big": false
},
{
"t": "icon",
"v": "timeline"
},
{
"t": "title",
"v": "Traceable actions and outcomes",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key reminder"
},
{
"t": "title",
"v": "Increase autonomy only when the value outweighs the risk, controls are effective and humans can intervene.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"id": "raicontrols",
"part": "assure-and-govern",
"title": "Control plane & evidence",
"icon": "shield_lock",
"eyebrow": "Responsible AI · govern and assure",
"heading": "One control plane for any tool, model or agent",
"lede": "Consistent principles, proportionate assurance and reusable components. Evidence and transparency records sit with the controls because they are how each control is shown to work.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "balance"
},
{
"t": "title",
"v": "Principles and risk",
"big": false
},
{
"t": "small",
"v": "Capability-based classification sets the assurance depth."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "database"
},
{
"t": "title",
"v": "Information and data",
"big": false
},
{
"t": "small",
"v": "Readiness, metadata, indexing, provenance and retention."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lock"
},
{
"t": "title",
"v": "Security and privacy",
"big": false
},
{
"t": "small",
"v": "Threat model, DPIA, transfers, subprocessors and DLP."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "supervisor_account"
},
{
"t": "title",
"v": "Human oversight",
"big": false
},
{
"t": "small",
"v": "Reviewer competence, evidence access, override and stop."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "accessibility_new"
},
{
"t": "title",
"v": "Accessibility and inclusion",
"big": false
},
{
"t": "p",
"v": "Accessibility of AI, AI for accessibility, alternatives and adjustments."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "payments"
},
{
"t": "title",
"v": "Cost and FinOps",
"big": false
},
{
"t": "small",
"v": "User-visible cost, caps, alerts, budget owners and value criteria."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "eco"
},
{
"t": "title",
"v": "Sustainability",
"big": false
},
{
"t": "small",
"v": "Supplier evidence, usage indicators and proportionate use."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "receipt_long"
},
{
"t": "title",
"v": "Transparency and records",
"big": false
},
{
"t": "small",
"v": "AI involvement statements, labelling and audit evidence."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "monitor_heart"
},
{
"t": "title",
"v": "Monitoring and assurance",
"big": false
},
{
"t": "small",
"v": "Control effectiveness, incidents and reviews."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "history"
},
{
"t": "title",
"v": "Change and lifecycle",
"big": false
},
{
"t": "small",
"v": "Change watchlist, reassessment triggers and retirement."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Who owns what"
},
{
"t": "card",
"kids": [
{
"t": "table",
"rows": [
[
{
"h": true,
"v": "Role"
},
{
"h": true,
"v": "Accountable for"
}
],
[
{
"h": true,
"v": "Senior responsible owner"
},
{
"h": false,
"v": "Risk acceptance, sponsorship and the decision to proceed or stop."
}
],
[
{
"h": true,
"v": "Service owner"
},
{
"h": false,
"v": "Day-to-day operation, monitoring, support and retirement."
}
],
[
{
"h": true,
"v": "Technical owner"
},
{
"h": false,
"v": "Configuration, identity, permissions, release and change."
}
],
[
{
"h": true,
"v": "Commercial owner"
},
{
"h": false,
"v": "Contract, licences, consumption, supplier commitments and exit."
}
],
[
{
"h": true,
"v": "Assurance owner"
},
{
"h": false,
"v": "Assessments, evidence, exceptions and reviews."
}
],
[
{
"h": true,
"v": "Budget owner"
},
{
"h": false,
"v": "Caps, alerts and spend against value evidence."
}
],
[
{
"h": true,
"v": "Agent, prompt or skill owner"
},
{
"h": false,
"v": "Version, tests, approved users and withdrawal."
}
]
]
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Exceptions and the shadow AI route"
},
{
"t": "p",
"v": "Unmanaged external use falls when asking for an approved alternative is quicker than working around the rules."
},
{
"t": "list",
"o": true,
"items": [
"Search your organisation's catalogue for an approved alternative.",
"If none fits, request one through a use case with the reason it is needed.",
"Get an interim decision within an agreed time, with conditions.",
"Record the exception in the decision register with an expiry date.",
"Review at expiry: approve, restrict or close."
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Communication authenticity"
},
{
"t": "p",
"v": "Keep clear labels on AI-sent messages and disclose material AI involvement. Removing a label is prohibited where transparency is required."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Evidence and transparency records"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "AI involvement statements",
"big": false
},
{
"t": "small",
"v": "What AI did in producing a document, decision or message."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Source records",
"big": false
},
{
"t": "small",
"v": "The sources an output drew on, so a reader can check them."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Human verification",
"big": false
},
{
"t": "small",
"v": "Who checked the output, against what, and when."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Decision-support records",
"big": false
},
{
"t": "small",
"v": "How AI informed a decision and who made it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Code records",
"big": false
},
{
"t": "small",
"v": "Provenance and review of AI-generated code."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Agent action logs",
"big": false
},
{
"t": "small",
"v": "What an agent did, when and why."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Audit evidence",
"big": false
},
{
"t": "small",
"v": "Configuration, approvals and test results for assurance."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Evidence grades for product and supplier claims"
},
{
"t": "p",
"v": "These sit alongside the guide's own conventions (verified, interpretation, proposal, illustrative). They grade claims about products and suppliers, many of which come from community calls."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Verified",
"big": false
},
{
"t": "p",
"v": "Confirmed against authoritative supplier documentation or a live configuration."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Corroborated",
"big": false
},
{
"t": "small",
"v": "Reported by more than one independent source or session."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Reported",
"big": false
},
{
"t": "small",
"v": "Raised in a community call or channel. Not yet confirmed."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Proposed",
"big": false
},
{
"t": "small",
"v": "A design recommendation in this portal, not a finding."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Unresolved",
"big": false
},
{
"t": "small",
"v": "An open question. Do not rely on it until answered."
}
]
}
],
"box": false
}
]
},
{
"id": "raidata",
"part": "assure-and-govern",
"title": "Data & information governance",
"icon": "database",
"eyebrow": "Assure and govern · data and information governance",
"heading": "AI governance is data governance with actions attached",
"lede": "Whatever an AI tool can reach, it can surface, summarise and act on. Data and information governance sets what it should reach, and treats what it produces as records.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "badge"
},
{
"t": "p",
"v": "**Ownership and accountability**Every source has an information asset owner; the senior information risk owner accepts residual risk."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "label"
},
{
"t": "p",
"v": "**Classification and handling**Classification travels with content into prompts, indexes and outputs."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "privacy_tip"
},
{
"t": "p",
"v": "**Data protection**UK GDPR lawful basis, DPIA, transfers, rights and service-level privacy notices."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "verified"
},
{
"t": "p",
"v": "**Quality and authority**Grounding sources are authoritative, current and fit for the decision they inform."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "travel_explore"
},
{
"t": "p",
"v": "**Metadata and discoverability**Titles, filenames and metadata can surface through AI. Discovery is not access."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "inventory_2"
},
{
"t": "p",
"v": "**Records and retention**Prompts, outputs, transcripts and agent logs are records with retention rules."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lock"
},
{
"t": "p",
"v": "**Access and permissions**Least privilege, oversharing clean-up, delegated and shared contexts tested."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "share"
},
{
"t": "p",
"v": "**Sharing and reuse**Data-sharing agreements, data boundaries and trusted reuse such as data matchmaking."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Data controls at each lifecycle stage"
},
{
"t": "card",
"kids": [
{
"t": "table",
"rows": [
[
{
"h": true,
"v": "Stage"
},
{
"h": true,
"v": "Data and information control"
},
{
"h": true,
"v": "Where it is recorded"
}
],
[
{
"h": true,
"v": "1 Discover"
},
{
"h": false,
"v": "Identify the information assets involved and their owners"
},
{
"h": false,
"v": "T1 Opportunity Brief"
}
],
[
{
"h": true,
"v": "2 Check"
},
{
"h": false,
"v": "Reuse existing datasets, catalogues and indexes before creating new ones"
},
{
"h": false,
"v": "T2 Duplication Check"
}
],
[
{
"h": true,
"v": "3 Classify"
},
{
"h": false,
"v": "Classification, personal data and special category data"
},
{
"h": false,
"v": "T3 Classification"
}
],
[
{
"h": true,
"v": "4 Assess"
},
{
"h": false,
"v": "DPIA, lawful basis, transfers and subprocessors, data quality and authority"
},
{
"h": false,
"v": "T4 Data Readiness · T5 Supplier"
}
],
[
{
"h": true,
"v": "5 Design"
},
{
"h": false,
"v": "Metadata, index scope, permissions, connectors and retention rules"
},
{
"h": false,
"v": "T7 Agent Design Record · data-source register"
}
],
[
{
"h": true,
"v": "6 Pilot"
},
{
"h": false,
"v": "Minimisation, test data, consent and privacy notices"
},
{
"h": false,
"v": "T8 Pilot Plan"
}
],
[
{
"h": true,
"v": "10 Operate"
},
{
"h": false,
"v": "Access reviews, DLP, records of AI outputs and agent logs"
},
{
"h": false,
"v": "Monitoring and assurance record"
}
],
[
{
"h": true,
"v": "11 Reassess"
},
{
"h": false,
"v": "New sources, connectors or locations trigger review"
},
{
"h": false,
"v": "T11 Change-Impact Assessment"
}
],
[
{
"h": true,
"v": "12 Retire"
},
{
"h": false,
"v": "Disposal, records transfer and revoked access"
},
{
"h": false,
"v": "T12 Retirement Record"
}
]
]
}
]
},
{
"t": "h",
"l": 2,
"v": "What AI creates is information too"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Prompts and instructions",
"big": false
},
{
"t": "p",
"v": "Retain where they shaped a decision or a published output; organisational prompts are versioned products."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "AI outputs",
"big": false
},
{
"t": "p",
"v": "Treat as drafts until a person verifies them; keep the verified version with an AI involvement statement."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Transcripts and recordings",
"big": false
},
{
"t": "p",
"v": "Consent notice, retention period and deletion route set before recording starts."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Agent action logs",
"big": false
},
{
"t": "small",
"v": "Keep for audit and incident review; owner can read them."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Indexes and embeddings",
"big": false
},
{
"t": "p",
"v": "Derived copies of source data; inherit the source classification and retention."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Evaluation data",
"big": false
},
{
"t": "small",
"v": "Test sets and results are evidence for the decision record."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Frameworks this aligns with"
},
{
"t": "p",
"v": "[[https://www.gov.uk/government/publications/data-ethics-framework|Data Ethics Framework ↗]]UK government framework for using data responsibly in policy and services."
},
{
"t": "p",
"v": "[[https://www.gov.uk/government/publications/the-government-data-quality-framework|Government Data Quality Framework ↗]]Principles and practices for assessing and improving data quality."
},
{
"t": "p",
"v": "[[https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/artificial-intelligence/|ICO guidance on AI and data protection ↗]]How UK GDPR applies to AI, including fairness, transparency and DPIAs."
},
{
"t": "p",
"v": "[[https://www.executiveoffice-ni.gov.uk/news/executive-office-launches-consultation-artificial-intelligence-strategy|Northern Ireland draft AI Strategy ↗]]Data governance is one of the eight proposed principles."
},
{
"t": "p",
"v": "[[https://www.gov.scot/publications/scotlands-ai-strategy-2026-2031/pages/8/|Scotland’s AI Strategy 2026–2031 ↗]]Data framed as a public asset, with a data matchmaking pilot."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Seen in practice"
},
{
"t": "p",
"v": "Kingston Council ran a DPIA on data flows, storage and access before going live with AI note-taking, and updated its privacy notice. Hartlepool publishes a service-level privacy notice for the same kind of tool. Belfast City Council put data and AI governance under one Data & Analytics Board."
}
]
}
],
"box": false
}
]
},
{
"id": "trust",
"part": "assure-and-govern",
"title": "Architecture of trust",
"icon": "verified_user",
"eyebrow": "Architecture of trust",
"heading": "Governing AI agents as digital workers",
"lede": "A continuous cycle that governs AI agents as digital workers to deliver trusted public services and outcomes. The agent at the centre is purpose driven, accountable, supervised, transparent, secure and ethical. Trust at every step: right agent, right authority, right outcome.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Identity",
"big": false
},
{
"t": "title",
"v": "Know the agent",
"big": false
},
{
"t": "small",
"v": "Define purpose and role"
},
{
"t": "small",
"v": "Register and record"
},
{
"t": "small",
"v": "Assign owners"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Capability",
"big": false
},
{
"t": "title",
"v": "What it can do",
"big": false
},
{
"t": "small",
"v": "Define capabilities"
},
{
"t": "small",
"v": "Tools, data, knowledge"
},
{
"t": "small",
"v": "Evaluate and test"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Authority",
"big": false
},
{
"t": "title",
"v": "What it is allowed to do",
"big": false
},
{
"t": "small",
"v": "Set boundaries"
},
{
"t": "small",
"v": "Permissions & limits"
},
{
"t": "small",
"v": "Approval thresholds"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Policy & guardrails",
"big": false
},
{
"t": "title",
"v": "How it must operate",
"big": false
},
{
"t": "small",
"v": "Policies & rules"
},
{
"t": "small",
"v": "Ethics & responsible AI"
},
{
"t": "small",
"v": "Guardrails & constraints"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Approval",
"big": false
},
{
"t": "title",
"v": "Authorise to operate",
"big": false
},
{
"t": "small",
"v": "Risk assessment"
},
{
"t": "small",
"v": "Assurance sign-off"
},
{
"t": "small",
"v": "Go-live approval"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Execution",
"big": false
},
{
"t": "title",
"v": "Act within context",
"big": false
},
{
"t": "small",
"v": "Perform tasks"
},
{
"t": "small",
"v": "Interact with users"
},
{
"t": "small",
"v": "Create outputs"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "7"
},
{
"t": "title",
"v": "Evidence & traceability",
"big": false
},
{
"t": "title",
"v": "Prove what happened",
"big": false
},
{
"t": "small",
"v": "Logging and audit trails"
},
{
"t": "small",
"v": "Data provenance"
},
{
"t": "small",
"v": "Decisions explained"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "8"
},
{
"t": "title",
"v": "Audit & assurance",
"big": false
},
{
"t": "title",
"v": "Check and challenge",
"big": false
},
{
"t": "small",
"v": "Monitor & review"
},
{
"t": "small",
"v": "Independent assurance"
},
{
"t": "small",
"v": "Compliance checks"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "9"
},
{
"t": "title",
"v": "Learning & improvement",
"big": false
},
{
"t": "title",
"v": "Get better over time",
"big": false
},
{
"t": "small",
"v": "Feedback loops"
},
{
"t": "small",
"v": "Model & process updates"
},
{
"t": "small",
"v": "Continuous improvement"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "10"
},
{
"t": "title",
"v": "Exit & retirement",
"big": false
},
{
"t": "title",
"v": "End safely and cleanly",
"big": false
},
{
"t": "small",
"v": "Revoke access"
},
{
"t": "small",
"v": "Archive & retain records"
},
{
"t": "small",
"v": "Decommission & review"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Guiding principles"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Human accountability",
"big": false
},
{
"t": "title",
"v": "Safety, security and privacy",
"big": false
},
{
"t": "title",
"v": "Lawful, ethical and fair",
"big": false
},
{
"t": "title",
"v": "Transparency and explainability",
"big": false
},
{
"t": "title",
"v": "Inclusive and accessible",
"big": false
},
{
"t": "title",
"v": "Proportional and risk-based",
"big": false
},
{
"t": "title",
"v": "Evidence based and measurable",
"big": false
},
{
"t": "title",
"v": "Sustainable and value for money",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Foundation standards"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Service Standard",
"big": false
},
{
"t": "title",
"v": "Technology Code of Practice",
"big": false
},
{
"t": "title",
"v": "AI Playbook & Responsible AI",
"big": false
},
{
"t": "title",
"v": "Data & AI Ethics Framework",
"big": false
},
{
"t": "title",
"v": "Security & Privacy Standards",
"big": false
},
{
"t": "title",
"v": "ATRS & Transparency",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Outcomes for public services"
},
{
"t": "p",
"v": "**Better services**Simpler, faster and more responsive"
},
{
"t": "p",
"v": "**Trusted decisions**Fair, consistent and well governed"
},
{
"t": "p",
"v": "**Efficient operations**Automate safely, reduce burden"
},
{
"t": "p",
"v": "**Empowered workforce**Augmented capability, not replaced"
},
{
"t": "p",
"v": "**Improved outcomes**Better impact for citizens, business and communities"
},
{
"t": "p",
"v": "**Stronger public trust**Accountable, transparent and responsible AI use"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Enabling foundations — built in, not bolted on"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Security by design** Protect systems, data and users"
},
{
"t": "p",
"v": "**Privacy by design** Respect and protect personal data"
},
{
"t": "p",
"v": "**Human in control** Meaningful oversight at all times"
},
{
"t": "p",
"v": "**Impact aware** Consider equality, ethics and public impact"
},
{
"t": "p",
"v": "**Transparent** Open about how and why agents are used"
},
{
"t": "p",
"v": "**Measured & monitored** Track performance, risk and value"
},
{
"t": "p",
"v": "**Continuously improved** Learn, adapt and evolve responsibly"
}
],
"box": false
}
]
}
]
},
{
"id": "govmodel",
"part": "assure-and-govern",
"title": "Governance operating model",
"icon": "layers",
"eyebrow": "Enterprise AI governance operating model",
"heading": "Nine layers, one accountable line of sight",
"lede": "A layered model for governing AI agents as digital workers to deliver trusted public services. Each layer answers one question — from why we do it, to the impact we create together.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Strategy & public value",
"big": false
},
{
"t": "p",
"v": "Set direction, objectives and principles aligned to mission, public value and national priorities."
},
{
"t": "title",
"v": "Why we do it",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Policy & governance",
"big": false
},
{
"t": "p",
"v": "Establish policies, roles, responsibilities, decision rights and governance structures."
},
{
"t": "title",
"v": "Who is accountable and how",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Standards & guidance",
"big": false
},
{
"t": "p",
"v": "Apply mandatory standards and guidance (Service Standard, TCoP, AI Playbook, Responsible AI, data and security standards, ATRS)."
},
{
"t": "title",
"v": "What good looks like",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Assurance & risk management",
"big": false
},
{
"t": "p",
"v": "Identify, assess, mitigate and monitor risks across the AI lifecycle. Assure compliance, safety, ethics and performance."
},
{
"t": "title",
"v": "How we manage risk and quality",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Platforms & foundations",
"big": false
},
{
"t": "p",
"v": "Provide secure, scalable, reliable platforms, data foundations, tools, shared services and reusable components."
},
{
"t": "title",
"v": "What we build on",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Services & solutions",
"big": false
},
{
"t": "p",
"v": "Design and deliver AI-enabled services and solutions that solve real problems and meet user needs."
},
{
"t": "title",
"v": "What we deliver",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "AI agents as digital workers",
"big": false
},
{
"t": "p",
"v": "Managed AI agents with defined purpose, capabilities, authority, guardrails, tools, data access and human oversight."
},
{
"t": "title",
"v": "Who or what performs the work",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Users & workforce",
"big": false
},
{
"t": "p",
"v": "Civil servants, partners and operational teams who work with AI agents, provide feedback and maintain oversight."
},
{
"t": "title",
"v": "Who uses and oversees",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Public outcomes & impact",
"big": false
},
{
"t": "p",
"v": "Better services, fairer decisions, improved efficiency and outcomes for citizens, businesses and communities."
},
{
"t": "title",
"v": "The impact we create together",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Enablers across all layers"
},
{
"t": "p",
"v": "**People & capability**Roles, skills, training, culture and change enablement."
},
{
"t": "p",
"v": "**Security & privacy**Secure by design, privacy by design, data protection and resilience."
},
{
"t": "p",
"v": "**Ethics & inclusion**Fairness, equality, accessibility, human rights and inclusion."
},
{
"t": "p",
"v": "**Transparency & ATRS**Transparency records, explainability, traceability and open publication where appropriate."
},
{
"t": "p",
"v": "**Data & knowledge**Quality data, governance, lineage, provenance and knowledge management."
},
{
"t": "p",
"v": "**Technology & architecture**Interoperability, open standards, cloud first, modular and secure architectures."
},
{
"t": "p",
"v": "**Continuous improvement**Measure, learn, iterate and improve services, models and operations."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Outcomes for citizens, businesses and society"
},
{
"t": "small",
"v": "Accessible, inclusive and easy to use services"
},
{
"t": "small",
"v": "Safe, secure and trustworthy use of AI"
},
{
"t": "small",
"v": "Fairer decisions and improved experiences"
},
{
"t": "small",
"v": "Better outcomes, efficiency and productivity"
},
{
"t": "small",
"v": "Value for money and responsible use of resources"
},
{
"t": "small",
"v": "Stronger public trust, legitimacy and democratic accountability"
},
{
"t": "title",
"v": "Our commitment: governed AI agents that are safe, effective and accountable — empowering our workforce to deliver better public services.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"id": "risks",
"part": "assure-and-govern",
"title": "Risks & mitigations",
"icon": "warning",
"eyebrow": "Key risks & mitigations matrix",
"heading": "Identify early. Control appropriately. Monitor continuously. Improve relentlessly.",
"lede": "Proactive risk management so AI agents operate safely, lawfully and effectively as digital workers. Human accountability remains non-delegable. Residual ratings assume the controls listed are in place and effective.",
"body": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "1. Inaccurate outputs (hallucination)"
},
{
"t": "eyebrow",
"v": "Residual: Low"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "small",
"v": "Agent generates incorrect, fabricated or misleading information."
},
{
"t": "small",
"v": "Poor decisions, loss of trust, reputational and service impact."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Factual errors"
},
{
"t": "small",
"v": "Fabricated citations"
},
{
"t": "small",
"v": "Incorrect summaries"
},
{
"t": "small",
"v": "Outdated information"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Grounding with trusted sources"
},
{
"t": "small",
"v": "Retrieval + citation requirements"
},
{
"t": "small",
"v": "Confidence thresholds"
},
{
"t": "small",
"v": "Human review for high-risk outputs"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Output evaluation"
},
{
"t": "small",
"v": "Spot checks & sampling"
},
{
"t": "small",
"v": "User feedback loops"
},
{
"t": "small",
"v": "Red-teaming"
},
{
"t": "title",
"v": "Well controlled with evidence of accuracy and oversight.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "2. Data privacy & leakage"
},
{
"t": "eyebrow",
"v": "Residual: Low"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "p",
"v": "Sensitive or personal data is exposed, misused or shared inappropriately."
},
{
"t": "small",
"v": "Breach of privacy law, fines, individual harm, loss of trust."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Prompt injection"
},
{
"t": "small",
"v": "Unintended data exposure"
},
{
"t": "small",
"v": "Insecure data handling"
},
{
"t": "small",
"v": "Over-broad access"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Data classification & handling rules"
},
{
"t": "small",
"v": "Least privilege access"
},
{
"t": "small",
"v": "Data masking & filtering"
},
{
"t": "small",
"v": "Secure environments & logging"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Access monitoring"
},
{
"t": "small",
"v": "Data loss prevention (DLP)"
},
{
"t": "small",
"v": "Privacy impact reviews"
},
{
"t": "small",
"v": "Audit trails"
},
{
"t": "title",
"v": "Strong controls and continuous monitoring in place.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "3. Automation bias & over-reliance"
},
{
"t": "eyebrow",
"v": "Residual: Medium"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "p",
"v": "Users over-trust agent outputs and fail to challenge or validate them."
},
{
"t": "small",
"v": "Poor decisions, unintended harm, reduced critical thinking."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Acceptance without review"
},
{
"t": "small",
"v": "Over-trust in suggestions"
},
{
"t": "small",
"v": "Deskilling of staff"
},
{
"t": "small",
"v": "Suppressed escalation"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Human-in-the-loop for key decisions"
},
{
"t": "small",
"v": "Clear UI prompts & explanations"
},
{
"t": "small",
"v": "Challenge culture & training"
},
{
"t": "small",
"v": "Escalation pathways"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "User behaviour analytics"
},
{
"t": "small",
"v": "Review rates"
},
{
"t": "small",
"v": "Escalation metrics"
},
{
"t": "small",
"v": "User training effectiveness"
},
{
"t": "title",
"v": "Managed through oversight and user behaviour.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "4. Over-authorisation & scope creep"
},
{
"t": "eyebrow",
"v": "Residual: Medium"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "small",
"v": "Agent given more authority or access than necessary or approved."
},
{
"t": "small",
"v": "Unauthorised actions, data misuse, policy breach, financial loss."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Excessive permissions"
},
{
"t": "small",
"v": "Unapproved action types"
},
{
"t": "small",
"v": "Gradual expansion"
},
{
"t": "small",
"v": "Weak change control"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Least privilege & role-based access"
},
{
"t": "small",
"v": "Capability versus authority separation"
},
{
"t": "small",
"v": "Change management process"
},
{
"t": "small",
"v": "Approval for new actions"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Permission reviews"
},
{
"t": "small",
"v": "Access recertification"
},
{
"t": "small",
"v": "Change audit logs"
},
{
"t": "small",
"v": "Control effectiveness testing"
},
{
"t": "title",
"v": "Requires ongoing review and re-authorisation.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "5. Bias & fairness issues"
},
{
"t": "eyebrow",
"v": "Residual: Medium"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "small",
"v": "Agent outputs unfair, discriminatory or disproportionate results."
},
{
"t": "p",
"v": "Inequitable outcomes, legal challenge, reputational harm, loss of legitimacy."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Biased training data"
},
{
"t": "small",
"v": "Proxy discrimination"
},
{
"t": "small",
"v": "Unequal performance"
},
{
"t": "small",
"v": "Lack of representation"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Fairness assessments"
},
{
"t": "small",
"v": "Diverse data & testing"
},
{
"t": "small",
"v": "Bias detection tools"
},
{
"t": "small",
"v": "Human oversight & appeal"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Fairness metrics"
},
{
"t": "small",
"v": "Outcome monitoring"
},
{
"t": "small",
"v": "Equality impact reviews"
},
{
"t": "small",
"v": "Independent assurance"
},
{
"t": "title",
"v": "Continuing evaluation reduces but does not eliminate risk.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "6. Security threats & misuse"
},
{
"t": "eyebrow",
"v": "Residual: Medium"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "small",
"v": "Agent exploited, manipulated or used maliciously by attackers."
},
{
"t": "p",
"v": "System compromise, data theft, service disruption, fraudulent actions."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Prompt injection"
},
{
"t": "small",
"v": "Model exploitation"
},
{
"t": "small",
"v": "Adversarial attacks"
},
{
"t": "small",
"v": "Tool misuse"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Secure by design"
},
{
"t": "small",
"v": "Input validation & guardrails"
},
{
"t": "small",
"v": "Threat modelling"
},
{
"t": "small",
"v": "Anomaly detection"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Security monitoring"
},
{
"t": "small",
"v": "Pen testing & red teaming"
},
{
"t": "small",
"v": "Vulnerability scans"
},
{
"t": "small",
"v": "Incident response testing"
},
{
"t": "title",
"v": "Risk managed with robust security controls.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "7. Model drift & performance degradation"
},
{
"t": "eyebrow",
"v": "Residual: Low"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "p",
"v": "Agent performance declines over time as data, context or patterns change."
},
{
"t": "small",
"v": "Reduced accuracy, poor outcomes, inefficiency, user dissatisfaction."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "Changing data patterns"
},
{
"t": "small",
"v": "Outdated knowledge"
},
{
"t": "small",
"v": "Domain shifts"
},
{
"t": "small",
"v": "Feedback ignored"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Continuous evaluation"
},
{
"t": "small",
"v": "Monitoring & alerting"
},
{
"t": "small",
"v": "Retraining & updates"
},
{
"t": "small",
"v": "Feedback integration"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Performance dashboards"
},
{
"t": "small",
"v": "Drift detection"
},
{
"t": "small",
"v": "KPI tracking"
},
{
"t": "small",
"v": "Periodic revalidation"
},
{
"t": "title",
"v": "Active monitoring keeps risk low.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "8. Accountability & governance gaps"
},
{
"t": "eyebrow",
"v": "Residual: Low"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Description & impact"
},
{
"t": "small",
"v": "Unclear ownership, missing controls or weak governance arrangements."
},
{
"t": "p",
"v": "Unclear responsibility, compliance breach, regulatory scrutiny, loss of trust."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Examples"
},
{
"t": "small",
"v": "No named owner"
},
{
"t": "small",
"v": "Weak oversight"
},
{
"t": "small",
"v": "Missing documentation"
},
{
"t": "small",
"v": "Poor auditability"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key mitigations / controls"
},
{
"t": "small",
"v": "Named owners & RACI"
},
{
"t": "small",
"v": "Governance framework"
},
{
"t": "small",
"v": "Mandatory documentation (ATRS)"
},
{
"t": "small",
"v": "Clear policies & escalation"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance & monitoring"
},
{
"t": "small",
"v": "Governance reviews"
},
{
"t": "small",
"v": "Audit & compliance checks"
},
{
"t": "small",
"v": "ATRS records reviewed"
},
{
"t": "small",
"v": "Board reporting"
},
{
"t": "title",
"v": "Strong governance minimises risk.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Foundational principles that reduce all risks"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Human accountability** People remain answerable."
},
{
"t": "p",
"v": "**Lawful & ethical use** Compliant, fair and responsible."
},
{
"t": "p",
"v": "**Transparency & explainability** Understandable actions and outputs."
},
{
"t": "p",
"v": "**Security & privacy by design** Protect data and systems."
},
{
"t": "p",
"v": "**Inclusive & equitable** Designed for all users and groups."
},
{
"t": "p",
"v": "**Measure & improve** Evidence based continuous improvement."
},
{
"t": "p",
"v": "**Monitor & adapt** Stay responsive to change."
},
{
"t": "p",
"v": "**Document & record** Keep clear, complete and auditable records."
}
],
"box": false
}
]
}
]
},
{
"id": "registers",
"part": "assure-and-govern",
"title": "Registers & lifecycle",
"icon": "inventory_2",
"eyebrow": null,
"heading": "The full catalogue schema — eighteen fields",
"lede": "A single source of truth for AI agents: discoverable, governed and actionable. The four-group record above is the minimum; this is the full blueprint, with an illustrative value for each field.",
"body": [
{
"t": "row",
"items": [
{
"t": "p",
"v": "**Why**Visibility, control, safe adoption and reuse"
},
{
"t": "p",
"v": "**Who**Owners, users and support"
},
{
"t": "p",
"v": "**What**Purpose, function and behaviour"
},
{
"t": "p",
"v": "**Where**Data sources, systems and environment"
},
{
"t": "p",
"v": "**How**Controls, assurance and monitoring"
},
{
"t": "p",
"v": "**Cost & value**Usage, cost and outcomes"
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Identity** Agent ID (unique) · Name · Description · Version · Status e.g. AG-2026-015"
},
{
"t": "p",
"v": "**2. Purpose & function** Purpose · Primary functions · Typical tasks · Intended outcomes e.g. Summarise policy documents and answer internal questions"
},
{
"t": "p",
"v": "**3. Ownership** Business owner · Creator / builder · Technical owner · Information owner · Risk owner · Service owner e.g. Policy Team Lead"
},
{
"t": "p",
"v": "**4. Users & audience** Primary users · User groups · Audience size · Geographies / organisations e.g. 250 staff in 2 departments"
},
{
"t": "p",
"v": "**5. Platform & model** Platform (e.g. M365, Copilot Studio) · Model(s) used · Model version · Region / residency e.g. GPT-4o via Microsoft 365 Copilot"
},
{
"t": "p",
"v": "**6. Data & information sources** Source systems · Information assets · Grounding method (RAG, connectors) · Data sensitivity e.g. SharePoint libraries, Policy DB, HR system"
},
{
"t": "p",
"v": "**7. Access & permissions** Access model · Authentication · Permissions set · Least privilege confirmed e.g. Entra ID group based access"
},
{
"t": "p",
"v": "**8. Sharing & distribution** Sharing scope (personal / team / department / cross-gov) · Channels / locations · Public-facing? Y/N e.g. Teams tab, shared with 3 teams"
},
{
"t": "p",
"v": "**9. Risk & criticality** Risk category · Criticality (low/med/high) · Impact areas · Consequences e.g. Medium — operational impact"
},
{
"t": "p",
"v": "**10. Controls & assurance** Guardrails / controls · Human oversight points · Testing completed · Assurance status e.g. Human review before action"
},
{
"t": "p",
"v": "**11. Usage & performance** Usage level · Key metrics · Quality indicators · Last reviewed e.g. 1,250 queries / month, 92% helpful"
},
{
"t": "p",
"v": "**12. Cost & consumption** Licensing / pricing model · Tokens / credits used · Estimated monthly cost · Cost owner e.g. Pay-as-you-go, £320 / month, IT"
},
{
"t": "p",
"v": "**13. Integrations & dependencies** Connected systems · APIs / connectors · Dependencies · Third parties e.g. Graph API, Power Automate"
},
{
"t": "p",
"v": "**14. Monitoring & observability** Logging enabled · Monitoring tools · Alerts & thresholds · Incident contact e.g. Purview logs, alerts via Sentinel"
},
{
"t": "p",
"v": "**15. Change & release** Change process · Release history · Last updated · Next review date e.g. v1.2 — 12 Aug 2026, review 12 Nov 2026"
},
{
"t": "p",
"v": "**16. Support & operations** Support model · Support owner · SLA / hours · Escalation path e.g. IT Service Desk P1–P3, 24×7"
},
{
"t": "p",
"v": "**17. Benefits & value** Expected benefits · Measured outcomes · Baseline / target · Value owner e.g. Saves 10 hrs/week, faster responses"
},
{
"t": "p",
"v": "**18. Retirement & archiving** Retirement criteria · Data retention · Archive location · Decommission plan e.g. Retire if inactive 180 days"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Status"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Active",
"big": false
},
{
"t": "title",
"v": "In review",
"big": false
},
{
"t": "title",
"v": "Planned",
"big": false
},
{
"t": "title",
"v": "Retired",
"big": false
},
{
"t": "title",
"v": "Draft",
"big": false
}
]
},
{
"t": "h",
"l": 3,
"v": "Quick filters"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "By owner",
"big": false
},
{
"t": "title",
"v": "By department",
"big": false
},
{
"t": "title",
"v": "By status",
"big": false
},
{
"t": "title",
"v": "By risk",
"big": false
},
{
"t": "title",
"v": "By platform",
"big": false
},
{
"t": "title",
"v": "By data sensitivity",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Governance at a glance"
},
{
"t": "p",
"v": "**Policy & standards**Aligned to organisational AI policy, standards and acceptable use."
},
{
"t": "p",
"v": "**Approval & publishing**Clear gates for publication and sharing at each tier of use."
},
{
"t": "p",
"v": "**Ongoing review**Regular reviews of risk, performance, cost and relevance."
},
{
"t": "p",
"v": "**Discovery & reuse**Easily discoverable to avoid duplication and share good practice."
},
{
"t": "p",
"v": "**Audit & accountability**Traceable ownership, decisions, changes and evidence."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Agent lifecycle overview"
},
{
"t": "small",
"v": "**1. Idea** — Identify need or opportunity"
},
{
"t": "small",
"v": "**2. Build** — Create and test in a safe space"
},
{
"t": "small",
"v": "**3. Share** — Share with team or small group"
},
{
"t": "small",
"v": "**4. Assess** — Evaluate risk, quality and value"
},
{
"t": "small",
"v": "**5. Approve** — Approve for wider use"
},
{
"t": "small",
"v": "**6. Operate** — Monitor, support and optimise"
},
{
"t": "small",
"v": "**7. Review** — Regular review of performance and risk"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Three registers, working together"
},
{
"t": "p",
"v": "Integrated, not duplicated. Each serves a distinct purpose and links to the others for discovery, control and reuse — over one set of shared foundational services. Single source of truth · integrated, not duplicated · accessible and discoverable · governed and accountable."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Information asset register"
},
{
"t": "p",
"v": "Maintain authoritative inventory of information assets for governance, compliance and risk management."
},
{
"t": "h",
"l": 4,
"v": "Contains"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Policies, standards & procedures",
"big": false
},
{
"t": "title",
"v": "Documents & records",
"big": false
},
{
"t": "title",
"v": "Data sets & databases",
"big": false
},
{
"t": "title",
"v": "Forms & templates",
"big": false
},
{
"t": "title",
"v": "Knowledge bases",
"big": false
},
{
"t": "title",
"v": "Reference data",
"big": false
},
{
"t": "title",
"v": "Expert directories",
"big": false
}
]
},
{
"t": "h",
"l": 4,
"v": "Key attributes"
},
{
"t": "p",
"v": "Owner & accountability · Classification & sensitivity · Retention & disposal · Legal & regulatory · Quality & governance"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Data catalogue"
},
{
"t": "p",
"v": "Make data easy to find and understand to promote reuse, reduce duplication and improve decision-making."
},
{
"t": "h",
"l": 4,
"v": "Contains"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Data sets & data products",
"big": false
},
{
"t": "title",
"v": "APIs & data services",
"big": false
},
{
"t": "title",
"v": "Metrics & KPIs",
"big": false
},
{
"t": "title",
"v": "Data dictionaries & models",
"big": false
},
{
"t": "title",
"v": "Knowledge bases",
"big": false
},
{
"t": "title",
"v": "Lineage & relationships",
"big": false
},
{
"t": "title",
"v": "Usage & popularity",
"big": false
}
]
},
{
"t": "h",
"l": 4,
"v": "Key attributes"
},
{
"t": "p",
"v": "Data definition · Owner & steward · Lineage & relationships · Quality & freshness · Access & usage"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Agent register / catalogue"
},
{
"t": "p",
"v": "Register and govern AI agents and agentic capabilities to ensure safe, accountable and effective use."
},
{
"t": "h",
"l": 4,
"v": "Contains"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agents & agent teams",
"big": false
},
{
"t": "title",
"v": "Purpose & use cases",
"big": false
},
{
"t": "title",
"v": "Intended users & audience",
"big": false
},
{
"t": "title",
"v": "Data sources & tools",
"big": false
},
{
"t": "title",
"v": "Risk & criticality",
"big": false
},
{
"t": "title",
"v": "Controls & assurance",
"big": false
},
{
"t": "title",
"v": "Lifecycle & status",
"big": false
}
]
},
{
"t": "h",
"l": 4,
"v": "Key attributes"
},
{
"t": "p",
"v": "Agent owner & creator · Purpose & scope · Data sources & tools · Risk & criticality · Lifecycle & review"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Shared foundational services"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Identity & access management",
"big": false
},
{
"t": "title",
"v": "Security & privacy controls",
"big": false
},
{
"t": "title",
"v": "Metadata & standards (terms, definitions)",
"big": false
},
{
"t": "title",
"v": "Integration & APIs (data & systems)",
"big": false
},
{
"t": "title",
"v": "Audit & monitoring (usage, changes)",
"big": false
},
{
"t": "title",
"v": "Quality management (validation, rules)",
"big": false
},
{
"t": "title",
"v": "Search & discovery (federated)",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How they work together"
},
{
"t": "p",
"v": "**1. Create / ingest** — New information, data or agents are identified and registered"
},
{
"t": "p",
"v": "**2. Link & relate** — Relationships are captured between assets, data and agents"
},
{
"t": "p",
"v": "**3. Use & leverage** — People and agents find and use trusted sources and capabilities"
},
{
"t": "p",
"v": "**4. Monitor & assure** — Activity, quality, risk and compliance are monitored across the landscape"
},
{
"t": "p",
"v": "**5. Improve & evolve** — Feedback drives better sources, smarter agents and stronger controls"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The observability control tower"
},
{
"t": "p",
"v": "Real-time visibility of agents, usage, performance, risk and value — insight into every agent, control where it matters, confidence in outcomes. These are the panels the register needs to earn its keep."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Agents by status** Active · In review · Retired"
},
{
"t": "p",
"v": "**Usage over time** Total agents against active agents, last 30 days"
},
{
"t": "p",
"v": "**Top agents by interactions** Ranked list with trend sparklines"
},
{
"t": "p",
"v": "**Risk distribution** Very high · High · Medium · Low"
},
{
"t": "p",
"v": "**Agents by owner type** Individual · Team · Department · Enterprise"
},
{
"t": "p",
"v": "**Spend by agent** Cost per agent, highest first"
},
{
"t": "p",
"v": "**Quality & reliability** Success rate · Response time · Feedback score · Error rate"
},
{
"t": "p",
"v": "**Sharing & discovery** Shared · Public to organisation · Cross-organisation"
},
{
"t": "p",
"v": "**Alerts & needs attention** Agents with no active owner · Breached sharing thresholds · Not reviewed in 60+ days · Rising error rate"
},
{
"t": "p",
"v": "**Recommended actions** Assign ownership and confirm purpose · Review sharing scope · Confirm continued value and risks · Validate cost and usage patterns"
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Filter by dimension"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner",
"big": false
},
{
"t": "title",
"v": "Department",
"big": false
},
{
"t": "title",
"v": "Agent type",
"big": false
},
{
"t": "title",
"v": "Platform / tool",
"big": false
},
{
"t": "title",
"v": "Data sensitivity",
"big": false
},
{
"t": "title",
"v": "Risk level",
"big": false
},
{
"t": "title",
"v": "Lifecycle status",
"big": false
},
{
"t": "title",
"v": "Tags / capability",
"big": false
},
{
"t": "title",
"v": "Geography",
"big": false
}
]
}
]
}
]
},
{
"id": "atrs",
"part": "assure-and-govern",
"title": "Internal ATRS registry",
"icon": "description",
"eyebrow": "Internal ATRS registry · AI agent transparency record",
"heading": "A transparency record for every agent used inside the organisation",
"lede": "Based on the principles of the Algorithmic Transparency Recording Standard. Purpose: to ensure transparency, accountability and responsible use of AI agents by recording key information about their purpose, authority, data, risks, controls and lifecycle. The registry integrates with enterprise systems for identity, security, monitoring, risk, audit and knowledge management.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Agent identity",
"big": false
},
{
"t": "small",
"v": "Agent name"
},
{
"t": "small",
"v": "Unique identifier"
},
{
"t": "small",
"v": "Version"
},
{
"t": "small",
"v": "Owners (SRO, service, technical)"
},
{
"t": "small",
"v": "Sponsor / business unit"
},
{
"t": "small",
"v": "Service supported"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Business context & purpose",
"big": false
},
{
"t": "small",
"v": "User needs addressed"
},
{
"t": "small",
"v": "Problem statement"
},
{
"t": "small",
"v": "Intended outcomes"
},
{
"t": "small",
"v": "Users and stakeholders"
},
{
"t": "small",
"v": "Public impact assessment"
},
{
"t": "small",
"v": "Success measures"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Authority & permissions",
"big": false
},
{
"t": "small",
"v": "Actions permitted"
},
{
"t": "small",
"v": "Actions prohibited"
},
{
"t": "small",
"v": "Decisions supported"
},
{
"t": "small",
"v": "Escalation triggers"
},
{
"t": "small",
"v": "Human approval requirements"
},
{
"t": "small",
"v": "Delegated authority level"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Data & knowledge",
"big": false
},
{
"t": "small",
"v": "Data sources & types"
},
{
"t": "small",
"v": "Data classification"
},
{
"t": "small",
"v": "Knowledge sources"
},
{
"t": "small",
"v": "Retention & disposal"
},
{
"t": "small",
"v": "Provenance & lineage"
},
{
"t": "small",
"v": "Data sharing & flows"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Responsible AI considerations",
"big": false
},
{
"t": "small",
"v": "Fairness & bias assessment"
},
{
"t": "small",
"v": "Equality & inclusion impact"
},
{
"t": "small",
"v": "Accessibility considerations"
},
{
"t": "small",
"v": "Ethical considerations"
},
{
"t": "small",
"v": "Human oversight approach"
},
{
"t": "small",
"v": "Potential harms & mitigations"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Assurance & controls",
"big": false
},
{
"t": "small",
"v": "Security review"
},
{
"t": "small",
"v": "DPIA completed"
},
{
"t": "small",
"v": "Risk assessment"
},
{
"t": "small",
"v": "Evaluation results"
},
{
"t": "small",
"v": "Red-teaming"
},
{
"t": "small",
"v": "Compliance checks"
},
{
"t": "small",
"v": "Approval & sign-off"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "7"
},
{
"t": "title",
"v": "Operational management",
"big": false
},
{
"t": "small",
"v": "Monitoring & logging"
},
{
"t": "small",
"v": "Performance metrics"
},
{
"t": "small",
"v": "Incident management"
},
{
"t": "small",
"v": "Alerting & thresholds"
},
{
"t": "small",
"v": "Kill switch / disable"
},
{
"t": "small",
"v": "Rollback process"
},
{
"t": "small",
"v": "Support & escalation"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "8"
},
{
"t": "title",
"v": "Lifecycle & status",
"big": false
},
{
"t": "p",
"v": "Lifecycle stage (proposed, pilot, live, restricted, suspended, retired)"
},
{
"t": "small",
"v": "Start date"
},
{
"t": "small",
"v": "Review frequency"
},
{
"t": "small",
"v": "End of life plan"
},
{
"t": "small",
"v": "Decommission date"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Benefits"
},
{
"t": "small",
"v": "Builds trust and accountability"
},
{
"t": "small",
"v": "Supports assurance and audit"
},
{
"t": "small",
"v": "Enables oversight and control"
},
{
"t": "small",
"v": "Promotes continuous improvement"
},
{
"t": "small",
"v": "Aligns with public sector standards"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Outputs & uses"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Central inventory of AI agents",
"big": false
},
{
"t": "title",
"v": "Transparency for oversight",
"big": false
},
{
"t": "title",
"v": "Evidence for audits & assurance",
"big": false
},
{
"t": "title",
"v": "Risk management & reporting",
"big": false
},
{
"t": "title",
"v": "Informed decision making",
"big": false
},
{
"t": "title",
"v": "Public sector standards compliance",
"big": false
},
{
"t": "title",
"v": "Learning & improvement",
"big": false
}
]
},
{
"t": "p",
"v": "Record quality: keep the record accurate, up to date and complete. Review at defined intervals or when significant changes occur."
}
]
}
],
"box": false
}
]
},
{
"id": "recovery",
"part": "assure-and-govern",
"title": "Failure & recovery",
"icon": "restart_alt",
"eyebrow": "Failure & recovery loop",
"heading": "Learn fast. Fix the issue. Prevent recurrence. Share the lesson.",
"lede": "Turning errors into stronger systems and better outcomes. Nine stages, and the ninth feeds the first: every failure is a source of learning. Work a live issue through the loop and record where you are.",
"body": [
{
"t": "row",
"items": [
{
"t": "p",
"v": "**Transparency**Surface issues quickly"
},
{
"t": "p",
"v": "**Accountability**Own and act on problems"
},
{
"t": "p",
"v": "**Learning**Capture and share lessons"
},
{
"t": "p",
"v": "**Continuous improvement**Strengthen continuously"
},
{
"t": "p",
"v": "**Proportionate**Respond to risk and impact"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "0 of 9 stages recorded",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Stage 1 of 9"
},
{
"t": "h",
"l": 2,
"v": "Detect & identify"
},
{
"t": "small",
"v": "Identify that something has gone wrong."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Errors, unexpected outputs",
"big": false
},
{
"t": "title",
"v": "User feedback",
"big": false
},
{
"t": "title",
"v": "Monitoring & alerts",
"big": false
},
{
"t": "title",
"v": "Quality checks",
"big": false
}
]
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Typical triggers"
},
{
"t": "icon",
"v": "priority_high"
},
{
"t": "small",
"v": "Inaccurate or incorrect outputs"
},
{
"t": "icon",
"v": "database"
},
{
"t": "small",
"v": "Outdated or incorrect data sources"
},
{
"t": "icon",
"v": "psychology"
},
{
"t": "small",
"v": "Prompt or model misbehaviour"
},
{
"t": "icon",
"v": "person"
},
{
"t": "small",
"v": "User confusion or misuse"
},
{
"t": "icon",
"v": "settings"
},
{
"t": "small",
"v": "System or integration failure"
},
{
"t": "icon",
"v": "balance"
},
{
"t": "small",
"v": "Policy or compliance breach"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Success indicators"
},
{
"t": "small",
"v": "Issues detected early"
},
{
"t": "small",
"v": "Risks contained"
},
{
"t": "small",
"v": "Root causes addressed"
},
{
"t": "small",
"v": "Controls strengthened"
},
{
"t": "small",
"v": "Lessons shared and applied"
},
{
"t": "small",
"v": "Fewer repeat incidents"
},
{
"t": "small",
"v": "Improved outcomes and trust"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "title",
"v": "Leadership sponsorship",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "handshake"
},
{
"t": "title",
"v": "Clear roles & responsibilities",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "policy"
},
{
"t": "title",
"v": "Policies & standards",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "database"
},
{
"t": "title",
"v": "Quality data & governance",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "shield"
},
{
"t": "title",
"v": "Security & privacy by design",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "monitor_heart"
},
{
"t": "title",
"v": "Observability & monitoring",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "school"
},
{
"t": "title",
"v": "Skilled teams & training",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "forum"
},
{
"t": "title",
"v": "Open feedback culture",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "build"
},
{
"t": "title",
"v": "Tools & automation support",
"big": false
}
]
}
],
"box": false
}
]
}
]
},
{
"id": "tower",
"part": "operate-and-oversee",
"title": "Portfolio control tower",
"icon": "monitor_heart",
"eyebrow": "AI governance control tower · illustrative dashboard",
"heading": "AI agent portfolio control tower",
"lede": "Govern. Assure. Monitor. Improve. Every figure below is illustrative, showing the dashboard a governed portfolio produces — last updated 20 May 2025 09:30.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "smart_toy"
},
{
"t": "stat",
"v": "128"
},
{
"t": "title",
"v": "AI agents registered",
"big": false
},
{
"t": "small",
"v": "↑ 12 this month"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check_circle"
},
{
"t": "stat",
"v": "87"
},
{
"t": "title",
"v": "Live in production",
"big": false
},
{
"t": "small",
"v": "↑ 7 this month"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "assignment"
},
{
"t": "stat",
"v": "45"
},
{
"t": "title",
"v": "In testing / pilot",
"big": false
},
{
"t": "small",
"v": "↓ 3 this month"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "pause_circle"
},
{
"t": "stat",
"v": "5"
},
{
"t": "title",
"v": "Suspended",
"big": false
},
{
"t": "small",
"v": "— No change"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "delete"
},
{
"t": "stat",
"v": "12"
},
{
"t": "title",
"v": "Retired",
"big": false
},
{
"t": "small",
"v": "↑ 2 this month"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "description"
},
{
"t": "stat",
"v": "116"
},
{
"t": "title",
"v": "ATRS records published",
"big": false
},
{
"t": "small",
"v": "↑ 10 this month"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Agents by lifecycle stage"
},
{
"t": "title",
"v": "128 total",
"big": true
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Proposed"
},
{
"t": "num",
"v": "8"
},
{
"t": "small",
"v": "6%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Live"
},
{
"t": "num",
"v": "87"
},
{
"t": "small",
"v": "68%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Pilot / testing"
},
{
"t": "num",
"v": "45"
},
{
"t": "small",
"v": "35%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Suspended"
},
{
"t": "num",
"v": "5"
},
{
"t": "small",
"v": "4%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Retired"
},
{
"t": "num",
"v": "12"
},
{
"t": "small",
"v": "9%"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Assurance status"
},
{
"t": "title",
"v": "128 total",
"big": true
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Assured"
},
{
"t": "num",
"v": "72"
},
{
"t": "small",
"v": "56%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Partially assured"
},
{
"t": "num",
"v": "28"
},
{
"t": "small",
"v": "22%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Not assured"
},
{
"t": "num",
"v": "18"
},
{
"t": "small",
"v": "14%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Not required"
},
{
"t": "num",
"v": "10"
},
{
"t": "small",
"v": "8%"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Risk overview"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "High risk"
},
{
"t": "num",
"v": "12"
},
{
"t": "small",
"v": "9%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Medium risk"
},
{
"t": "num",
"v": "34"
},
{
"t": "small",
"v": "27%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Low risk"
},
{
"t": "num",
"v": "62"
},
{
"t": "small",
"v": "48%"
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Very low risk"
},
{
"t": "num",
"v": "20"
},
{
"t": "small",
"v": "16%"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "error"
},
{
"t": "p",
"v": "12 agents are high risk and require action.**View risk register →**"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Agents by domain"
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "groups"
},
{
"t": "small",
"v": "Citizen services"
},
{
"t": "num",
"v": "32"
},
{
"t": "small",
"v": "25%"
}
]
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "apartment"
},
{
"t": "small",
"v": "Corporate services"
},
{
"t": "num",
"v": "24"
},
{
"t": "small",
"v": "19%"
}
]
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "description"
},
{
"t": "small",
"v": "Policy & strategy"
},
{
"t": "num",
"v": "18"
},
{
"t": "small",
"v": "14%"
}
]
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "settings"
},
{
"t": "small",
"v": "Operations"
},
{
"t": "num",
"v": "16"
},
{
"t": "small",
"v": "13%"
}
]
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "payments"
},
{
"t": "small",
"v": "Finance"
},
{
"t": "num",
"v": "14"
},
{
"t": "small",
"v": "11%"
}
]
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "badge"
},
{
"t": "small",
"v": "HR & people"
},
{
"t": "num",
"v": "12"
},
{
"t": "small",
"v": "9%"
}
]
},
{
"t": "row",
"items": [
{
"t": "icon",
"v": "more_horiz"
},
{
"t": "small",
"v": "Other"
},
{
"t": "num",
"v": "12"
},
{
"t": "small",
"v": "9%"
}
]
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "AI agent register"
},
{
"t": "title",
"v": "View all agents →",
"big": false
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "AGENT NAME",
"big": false
},
{
"t": "title",
"v": "SERVICE / DOMAIN",
"big": false
},
{
"t": "title",
"v": "PURPOSE",
"big": false
},
{
"t": "title",
"v": "OWNER",
"big": false
},
{
"t": "title",
"v": "LIFECYCLE",
"big": false
},
{
"t": "title",
"v": "RISK",
"big": false
},
{
"t": "title",
"v": "ASSURANCE",
"big": false
},
{
"t": "title",
"v": "ATRS RECORD",
"big": false
},
{
"t": "title",
"v": "LAST REVIEWED",
"big": false
},
{
"t": "title",
"v": "NEXT REVIEW",
"big": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Policy Briefing Assistant",
"big": false
},
{
"t": "small",
"v": "Policy & Strategy"
},
{
"t": "p",
"v": "Supports policy analysts by summarising key documents and drafting briefing notes."
},
{
"t": "small",
"v": "Sarah Johnson (Director)"
},
{
"t": "title",
"v": "Live",
"big": false
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "title",
"v": "Assured",
"big": false
},
{
"t": "title",
"v": "ATRS-00045",
"big": false
},
{
"t": "small",
"v": "15 May 2025"
},
{
"t": "small",
"v": "15 Aug 2025"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Casework Triage Agent",
"big": false
},
{
"t": "small",
"v": "Citizen Services"
},
{
"t": "small",
"v": "Prioritises and routes incoming casework to the right teams."
},
{
"t": "small",
"v": "Mark Patel (Head of Ops)"
},
{
"t": "title",
"v": "Live",
"big": false
},
{
"t": "title",
"v": "Medium",
"big": false
},
{
"t": "title",
"v": "Partially assured",
"big": false
},
{
"t": "title",
"v": "ATRS-00032",
"big": false
},
{
"t": "small",
"v": "10 May 2025"
},
{
"t": "small",
"v": "10 Aug 2025"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Finance Anomaly Detector",
"big": false
},
{
"t": "small",
"v": "Finance"
},
{
"t": "small",
"v": "Detects unusual transactions and supports investigation."
},
{
"t": "small",
"v": "Alex Kim (Finance Lead)"
},
{
"t": "title",
"v": "Live",
"big": false
},
{
"t": "title",
"v": "Medium",
"big": false
},
{
"t": "title",
"v": "Assured",
"big": false
},
{
"t": "title",
"v": "ATRS-00028",
"big": false
},
{
"t": "small",
"v": "08 May 2025"
},
{
"t": "small",
"v": "08 Aug 2025"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "HR Policy Advisor",
"big": false
},
{
"t": "small",
"v": "HR & People"
},
{
"t": "small",
"v": "Provides guidance on HR policies and employee queries."
},
{
"t": "small",
"v": "Lucy Bennett (HR Manager)"
},
{
"t": "title",
"v": "Pilot",
"big": false
},
{
"t": "title",
"v": "Low",
"big": false
},
{
"t": "title",
"v": "Assured",
"big": false
},
{
"t": "title",
"v": "ATRS-00061",
"big": false
},
{
"t": "small",
"v": "12 May 2025"
},
{
"t": "small",
"v": "12 Aug 2025"
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "IT Support Copilot",
"big": false
},
{
"t": "small",
"v": "Corporate Services"
},
{
"t": "small",
"v": "Assists IT staff with troubleshooting and knowledge retrieval."
},
{
"t": "small",
"v": "Tom Edwards (IT Manager)"
},
{
"t": "title",
"v": "Pilot",
"big": false
},
{
"t": "title",
"v": "Low",
"big": false
},
{
"t": "title",
"v": "Not required",
"big": false
},
{
"t": "title",
"v": "—",
"big": false
},
{
"t": "small",
"v": "18 May 2025"
},
{
"t": "small",
"v": "18 Aug 2025"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Comms Drafting Agent",
"big": false
},
{
"t": "small",
"v": "Corporate Services"
},
{
"t": "small",
"v": "Drafts communications and public messages."
},
{
"t": "small",
"v": "Nina Shah (Comms Lead)"
},
{
"t": "title",
"v": "Suspended",
"big": false
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "title",
"v": "Partially assured",
"big": false
},
{
"t": "title",
"v": "ATRS-00017",
"big": false
},
{
"t": "small",
"v": "02 May 2025"
},
{
"t": "small",
"v": "Review pending"
}
]
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "database"
},
{
"t": "p",
"v": "**46**Data sources connected"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lock"
},
{
"t": "p",
"v": "**27**High sensitivity data sources"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "p",
"v": "**96%**Agents have named owners"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "assignment_turned_in"
},
{
"t": "p",
"v": "**100%**DPIA completed (in scope)"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "shield"
},
{
"t": "p",
"v": "**94%**Security reviews up to date"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "insights"
},
{
"t": "p",
"v": "**85%**Agents have defined success metrics"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "eco"
},
{
"t": "p",
"v": "**65%**Agents have sustainability assessment"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Control tower navigation"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "speed"
},
{
"t": "title",
"v": "Overview",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "smart_toy"
},
{
"t": "title",
"v": "AI agents",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "description"
},
{
"t": "title",
"v": "ATRS registry",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "verified_user"
},
{
"t": "title",
"v": "Assurance",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "warning"
},
{
"t": "title",
"v": "Risks",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "notifications"
},
{
"t": "title",
"v": "Incidents",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "bar_chart"
},
{
"t": "title",
"v": "Reports",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "title",
"v": "Analytics",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "settings"
},
{
"t": "title",
"v": "Settings",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "help"
},
{
"t": "title",
"v": "Help",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Quick actions"
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "add"
},
{
"t": "title",
"v": "Register new agent",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "add"
},
{
"t": "title",
"v": "Create ATRS record",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "add"
},
{
"t": "title",
"v": "Request assurance",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "add"
},
{
"t": "title",
"v": "Report an issue",
"big": false
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "raioperate",
"part": "operate-and-oversee",
"title": "Monitoring & reassessment",
"icon": "update",
"eyebrow": "Responsible AI · operate and improve",
"heading": "Approval stays current only while the evidence does",
"lede": "Monitoring, incidents, the change watchlist, reassessment and retirement, run as one learning loop. The Portfolio control tower shows the estate view; this chapter defines what feeds it.",
"body": [
{
"t": "h",
"l": 2,
"v": "Dashboard domains"
},
{
"t": "card",
"kids": [
{
"t": "table",
"rows": [
[
{
"h": true,
"v": "Domain"
},
{
"h": true,
"v": "What it shows"
},
{
"h": true,
"v": "Fed by"
},
{
"h": true,
"v": "Act when"
}
],
[
{
"h": true,
"v": "Adoption and usage"
},
{
"h": false,
"v": "Who uses which capability, and how often, by group."
},
{
"h": false,
"v": "Tool, use-case registers"
},
{
"h": false,
"v": "Uptake is uneven across groups or roles."
}
],
[
{
"h": true,
"v": "Cost and consumption"
},
{
"h": false,
"v": "Spend by tool, team and use case against caps."
},
{
"h": false,
"v": "Tool, benefits registers"
},
{
"h": false,
"v": "A cap or alert is hit, or cost grows faster than use."
}
],
[
{
"h": true,
"v": "Outcome quality and value"
},
{
"h": false,
"v": "Verified outcomes, not prompt counts or time saved alone."
},
{
"h": false,
"v": "Benefits register"
},
{
"h": false,
"v": "Value evidence is missing at review."
}
],
[
{
"h": true,
"v": "Correction and rework"
},
{
"h": false,
"v": "Review effort and how much output is edited or discarded."
},
{
"h": false,
"v": "Benefits register"
},
{
"h": false,
"v": "Rework cancels the time saved."
}
],
[
{
"h": true,
"v": "Accessibility and inclusion"
},
{
"h": false,
"v": "Barriers reported, adjustments supported, alternatives used."
},
{
"h": false,
"v": "Incident, use-case registers"
},
{
"h": false,
"v": "A barrier is reported without an alternative."
}
],
[
{
"h": true,
"v": "Risk, incidents and near misses"
},
{
"h": false,
"v": "Incidents by tool, model, use case and control."
},
{
"h": false,
"v": "Incident register"
},
{
"h": false,
"v": "Repeat incidents on one control."
}
],
[
{
"h": true,
"v": "Sustainability indicators"
},
{
"h": false,
"v": "Supplier environmental evidence and usage indicators."
},
{
"h": false,
"v": "Tool, supplier registers"
},
{
"h": false,
"v": "Evidence is missing or usage is disproportionate."
}
],
[
{
"h": true,
"v": "Control effectiveness"
},
{
"h": false,
"v": "Whether controls worked when tested or triggered."
},
{
"h": false,
"v": "Incident, review registers"
},
{
"h": false,
"v": "A control failed or was bypassed."
}
],
[
{
"h": true,
"v": "Change and deprecation exposure"
},
{
"h": false,
"v": "Live use affected by upcoming supplier changes."
},
{
"h": false,
"v": "Change, model registers"
},
{
"h": false,
"v": "A deprecation lands inside the review window."
}
],
[
{
"h": true,
"v": "Support demand and recurring issues"
},
{
"h": false,
"v": "Tickets and repeat questions by tool."
},
{
"h": false,
"v": "Support and service records"
},
{
"h": false,
"v": "The same issue recurs across teams."
}
]
]
}
]
},
{
"t": "p",
"v": "No figures are shown. Each domain is defined so live data can be connected without redesign."
},
{
"t": "h",
"l": 2,
"v": "Changes that trigger reassessment"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Model",
"big": false
},
{
"t": "small",
"v": "A new, changed or automatically selected model."
},
{
"t": "eyebrow",
"v": "Partial or full"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Supplier",
"big": false
},
{
"t": "small",
"v": "New subprocessor, location, terms or ownership."
},
{
"t": "eyebrow",
"v": "Full"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Data",
"big": false
},
{
"t": "small",
"v": "New sources, connectors or classifications."
},
{
"t": "eyebrow",
"v": "Partial or full"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Permission",
"big": false
},
{
"t": "small",
"v": "Wider access, new identities or delegation."
},
{
"t": "eyebrow",
"v": "Full"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Action",
"big": false
},
{
"t": "small",
"v": "New tools, actions, schedules or agent links."
},
{
"t": "eyebrow",
"v": "Full"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Interface",
"big": false
},
{
"t": "small",
"v": "Changed terminology, layout or defaults."
},
{
"t": "eyebrow",
"v": "Guidance update"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Cost",
"big": false
},
{
"t": "small",
"v": "New pricing, metering or consumption pattern."
},
{
"t": "eyebrow",
"v": "Partial"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Policy",
"big": false
},
{
"t": "small",
"v": "New law, policy or guidance."
},
{
"t": "eyebrow",
"v": "Partial or full"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Change and deprecation watchlist"
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "17–20 Nov 2026",
"big": false
},
{
"t": "p",
"v": "**Microsoft Ignite**Several preview items in the service atlas are likely to change. Refresh entries after the event."
},
{
"t": "eyebrow",
"v": "Proposed"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Ongoing",
"big": false
},
{
"t": "p",
"v": "**Staggered Copilot interface rollout**Users see different interfaces at once. Keep guidance by release channel and show preview versus general availability."
},
{
"t": "eyebrow",
"v": "Reported"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Each release",
"big": false
},
{
"t": "p",
"v": "**Default-enabled features**New features can arrive switched on. Check admin defaults after each release."
},
{
"t": "eyebrow",
"v": "Reported"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Open",
"big": false
},
{
"t": "p",
"v": "**Claude models reached through Microsoft**Confirm subprocessor terms and whether processing sits inside the data boundary."
},
{
"t": "eyebrow",
"v": "Unresolved"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Open",
"big": false
},
{
"t": "p",
"v": "**Microsoft UK data processing commitment**Confirm in your admin centre before relying on it."
},
{
"t": "eyebrow",
"v": "Unresolved"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Ongoing",
"big": false
},
{
"t": "p",
"v": "**Indexed metadata exposure**Titles, filenames and metadata can surface despite permissions. Re-test indexing on change."
},
{
"t": "eyebrow",
"v": "Reported"
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Sep 2026",
"big": false
},
{
"t": "p",
"v": "**Cowork cost and security**Concerns raised about credit consumption and security. Set caps before any pilot."
},
{
"t": "eyebrow",
"v": "Reported"
}
],
"box": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Governance as a continuous learning system"
},
{
"t": "list",
"o": true,
"items": [
[
{
"t": "title",
"v": "1. Pilot",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Define scope and success measures",
"Test with real users",
"Put controls in place",
"Set stop criteria"
]
}
],
[
{
"t": "title",
"v": "2. Evidence",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Gather quantitative and qualitative data",
"Assess outcomes, quality and value",
"Capture accessibility, cost and risk",
"Record unintended effects"
]
}
],
[
{
"t": "title",
"v": "3. Decision",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Approve, approve with conditions, extend, restrict or stop",
"Record the rationale",
"Update registers",
"Tell users"
]
}
],
[
{
"t": "title",
"v": "4. Operation",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Deploy and support users",
"Monitor controls and compliance",
"Provide guidance and training",
"Maintain human oversight"
]
}
],
[
{
"t": "title",
"v": "5. Monitoring",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Track usage and adoption",
"Monitor cost and consumption",
"Identify risks and anomalies",
"Measure outcomes and value"
]
}
],
[
{
"t": "title",
"v": "6. Incident, change or learning",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Respond to incidents and near misses",
"Investigate and resolve",
"Capture lessons",
"Watch product, model and policy changes"
]
}
],
[
{
"t": "title",
"v": "7. Reassessment",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Assess the impact of change",
"Update the risk and control position",
"Refresh approvals if needed",
"Decide: continue, restrict or suspend"
]
}
],
[
{
"t": "title",
"v": "8. Retire",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Disable access and actions",
"Remove data and integrations",
"Inform users",
"Close records and capture lessons"
]
}
]
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Turning experience into impact"
},
{
"t": "row",
"items": [
{
"t": "p",
"v": "**Better decisions**Based on real outcomes"
},
{
"t": "p",
"v": "**Stronger controls**Informed by actual risks"
},
{
"t": "p",
"v": "**Greater value**For users and public services"
},
{
"t": "p",
"v": "**More inclusive**Identifying and removing barriers"
},
{
"t": "p",
"v": "**Smarter spending**Understanding cost and benefit"
},
{
"t": "p",
"v": "**Shared learning**Reusable patterns and guidance"
}
]
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Report an AI concern, incident or near miss"
},
{
"t": "p",
"v": "Report anything that went wrong or nearly did: a wrong output that was acted on, an exposed file, an unexpected action, a cost spike, an accessibility barrier. Record:"
},
{
"t": "list",
"o": false,
"items": [
"Tool, model and deployment route",
"Use case and what happened",
"Who was affected, and whether it was acted on",
"The control that should have prevented it",
"Corrective action and owner"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Suspension and retirement"
},
{
"t": "list",
"o": true,
"items": [
"Record the reason and the replacement.",
"Tell users, with an accessible alternative.",
"Disable agents and actions, and revoke access.",
"Disconnect integrations and dispose of data.",
"Update registers and remove supplier access.",
"Record residual risks and lessons; approve closure."
]
}
]
}
],
"box": false
}
]
},
{
"id": "agentjourney",
"part": "operate-and-oversee",
"title": "Agent journey",
"icon": "smart_toy",
"eyebrow": "Agent journey",
"heading": "From idea to retirement, with a gate at every stage",
"lede": "An agent that helps one person and an agent that decides a citizen's case are not the same thing, and should not carry the same controls. Eight stages, each with a gate question, the control that applies, the role accountable and the outcome expected.",
"body": [
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Stage 1 of 8"
},
{
"t": "title",
"v": "Personal / low risk",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Idea & discovery"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "What happens"
},
{
"t": "small",
"v": "Identify a need or opportunity"
},
{
"t": "small",
"v": "Explore feasibility"
},
{
"t": "small",
"v": "Check existing solutions and policies"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Gate 1 — decision point"
},
{
"t": "title",
"v": "Is there a clear need and no existing solution?",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 3,
"v": "Control applied"
},
{
"t": "small",
"v": "Security & identity — least privilege, access control and monitoring"
},
{
"t": "h",
"l": 3,
"v": "Key accountability"
},
{
"t": "small",
"v": "Creator — builds and tests the agent; owns initial purpose"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Outcome at this stage"
},
{
"t": "title",
"v": "Better understanding of needs and opportunities",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Success factors across the journey"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Clear purpose and value",
"big": false
},
{
"t": "title",
"v": "Right controls at the right time",
"big": false
},
{
"t": "title",
"v": "User engagement and feedback",
"big": false
},
{
"t": "title",
"v": "Sustainable cost model",
"big": false
},
{
"t": "title",
"v": "Continuous learning and improvement",
"big": false
},
{
"t": "title",
"v": "Strong community and reuse",
"big": false
},
{
"t": "title",
"v": "Better public value outcomes",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Personal experiment to enterprise service"
},
{
"t": "p",
"v": "A managed progression that balances innovation, control, risk and value. Agents evolve from personal productivity tools to trusted, supported services — and ownership moves with them."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1** **Idea & personal experiment** Individual creator **Low risk · personal**"
},
{
"t": "p",
"v": "**2** **Share & team pilot** Team data owners **Low to moderate risk**"
},
{
"t": "p",
"v": "**3** **Managed team agent** Named team owner **Moderate risk**"
},
{
"t": "p",
"v": "**4** **Departmental capability** Departmental owner **Moderate to high risk**"
},
{
"t": "p",
"v": "**5** **Operational service** Service owner **High risk**"
},
{
"t": "p",
"v": "**6** **Enterprise service** Business / enterprise owner **High risk · high impact**"
},
{
"t": "p",
"v": "**7** **Cross-government capability** Cross-government sponsor & owners **High impact · strategic**"
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The Copilot-to-agent capability spectrum"
},
{
"t": "p",
"v": "A continuum of assistance and autonomy — from individual productivity to enterprise agentic services. Purpose: enable people to work better today, orchestrate intelligent workflows, and deliver trusted, efficient, inclusive public services."
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Tier 2 of 7 · Microsoft 365 Copilot"
},
{
"t": "h",
"l": 3,
"v": "M365 Copilot"
},
{
"t": "small",
"v": "AI in your apps to summarise, draft, analyse and explore"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**TYPICAL USERS****Individual**"
},
{
"t": "p",
"v": "**AUTONOMY LEVEL****Human-guided**"
},
{
"t": "p",
"v": "**SCOPE****Personal**"
},
{
"t": "p",
"v": "**EXAMPLES****Word drafting · Excel analysis · Meeting summary**"
},
{
"t": "p",
"v": "**GOVERNANCE****Org policies**"
}
],
"box": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Avoiding the “access database moment”"
},
{
"t": "p",
"v": "How a small, helpful agent becomes business critical without visibility or control. Small agents can create big value — and big risk. Visibility, ownership and governance early in the journey prevent unmanaged tools from becoming unmanageable dependencies."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1** **Personal experiment** **Low visibility** An individual creates an agent to solve a personal problem or save time."
},
{
"t": "p",
"v": "**2** **Shared with team** **Low to moderate visibility** Shared with colleagues or a small team to help with common tasks."
},
{
"t": "p",
"v": "**3** **Wider adoption** **Rising visibility** More teams discover it and start using it organically."
},
{
"t": "p",
"v": "**4** **Dependence builds** **Rising to high visibility** Critical work starts to rely on the agent and its outputs."
},
{
"t": "p",
"v": "**5** **Business critical** **High visibility** Used across the department or service. Business processes become dependent on it."
},
{
"t": "p",
"v": "**6** **Invisible criticality** **Gap in visibility** No formal ownership, limited documentation, unknown controls, unknown costs."
},
{
"t": "p",
"v": "**7** **High risk exposure** **Critical · unmanaged** Risk of failure, data breach, inaccurate decisions or service disruption."
},
{
"t": "p",
"v": "**8** **Event of failure** **Loss of trust and value** Agent breaks, data changes or owner leaves — causing a real operational impact."
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How to prevent it — early detection, ownership and routes to enterprise management"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Discover early** Agent inventory · Usage monitoring · Sharing visibility · Community reports"
},
{
"t": "p",
"v": "**2. Require description** Purpose & use case · Data sources · Support owner · Review dates"
},
{
"t": "p",
"v": "**3. Assign ownership** Business owner · Technical owner · Information owner · Risk owner"
},
{
"t": "p",
"v": "**4. Assess & classify** Risk & impact · Data sensitivity · Criticality score · Assurance needs"
},
{
"t": "p",
"v": "**5. Control & assure** Approved sources · Guardrails · Human oversight · Testing & validation"
},
{
"t": "p",
"v": "**6. Monitor & review** Usage & performance · Cost & consumption · Accuracy & quality · Incident tracking"
},
{
"t": "p",
"v": "**7. Promote or remediate** Move to managed platform (e.g. Copilot Studio) · Or retire / remediate"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What goes wrong when it is left unmanaged"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Shadow IT",
"big": false
},
{
"t": "title",
"v": "Inconsistent quality",
"big": false
},
{
"t": "title",
"v": "Lack of documentation",
"big": false
},
{
"t": "title",
"v": "Knowledge silos",
"big": false
},
{
"t": "title",
"v": "Unknown cost",
"big": false
},
{
"t": "title",
"v": "Compliance breach",
"big": false
},
{
"t": "title",
"v": "Loss of public trust",
"big": false
}
]
}
]
}
]
},
{
"id": "employment",
"part": "operate-and-oversee",
"title": "Agent employment",
"icon": "badge",
"eyebrow": "Agent employment lifecycle",
"heading": "Managing AI agents as governed digital workers",
"lede": "Agents go through a managed lifecycle — from intent to retirement — with controls, oversight and accountability at every stage. People remain accountable: agents augment human capability, they do not replace judgement, responsibility or accountability.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Recruitment (define need)",
"big": false
},
{
"t": "title",
"v": "Human analogy: Job requisition & role definition",
"big": false
},
{
"t": "small",
"v": "Identify the need, purpose and expected outcomes."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Business case"
},
{
"t": "small",
"v": "User need"
},
{
"t": "small",
"v": "Impact assessment"
},
{
"t": "small",
"v": "Initial risk screen"
},
{
"t": "title",
"v": "Clear mandate and value case",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Vetting & evaluation",
"big": false
},
{
"t": "title",
"v": "Human analogy: Recruitment, screening & interviews",
"big": false
},
{
"t": "p",
"v": "Assess capability, risks, data needs and alignment with policy and standards."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Capability evaluation"
},
{
"t": "small",
"v": "Risk assessment"
},
{
"t": "small",
"v": "Data assessment"
},
{
"t": "small",
"v": "Policy alignment"
},
{
"t": "small",
"v": "Assurance readiness"
},
{
"t": "title",
"v": "Fit for purpose and safe to proceed",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Onboarding & configuration",
"big": false
},
{
"t": "title",
"v": "Human analogy: Induction & onboarding",
"big": false
},
{
"t": "p",
"v": "Configure the agent, set permissions, guardrails and connect data and tools."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Guardrails"
},
{
"t": "small",
"v": "Permissions"
},
{
"t": "small",
"v": "Data access"
},
{
"t": "small",
"v": "Tool access"
},
{
"t": "small",
"v": "Prompt / policy setup"
},
{
"t": "title",
"v": "Ready, secure and configured correctly",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Probation (pilot)",
"big": false
},
{
"t": "title",
"v": "Human analogy: Probation period",
"big": false
},
{
"t": "p",
"v": "Operate in a controlled pilot with close human oversight and limited scope."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Limited scope"
},
{
"t": "small",
"v": "Human-in-the-loop"
},
{
"t": "small",
"v": "Enhanced monitoring"
},
{
"t": "small",
"v": "Evaluation criteria"
},
{
"t": "title",
"v": "Validated in real-world conditions",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Live operation (employment)",
"big": false
},
{
"t": "title",
"v": "Human analogy: Active employment",
"big": false
},
{
"t": "p",
"v": "Agent operates within its delegated authority to support people and processes."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Performance KPIs"
},
{
"t": "small",
"v": "Quality checks"
},
{
"t": "small",
"v": "Bias & fairness review"
},
{
"t": "small",
"v": "Value realisation review"
},
{
"t": "title",
"v": "Measurable performance and value",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Performance review",
"big": false
},
{
"t": "title",
"v": "Human analogy: Performance appraisal",
"big": false
},
{
"t": "p",
"v": "Regularly review performance, accuracy, impact, risks and value delivery."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Model updates"
},
{
"t": "small",
"v": "Knowledge refresh"
},
{
"t": "small",
"v": "Prompt tuning"
},
{
"t": "small",
"v": "Control improvements"
},
{
"t": "title",
"v": "Improved capability and reduced risk",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "7"
},
{
"t": "title",
"v": "Retrain & improve",
"big": false
},
{
"t": "title",
"v": "Human analogy: Training & professional development",
"big": false
},
{
"t": "p",
"v": "Update knowledge, models, prompts and controls based on learning and feedback."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Restrict authority"
},
{
"t": "small",
"v": "Increase oversight"
},
{
"t": "small",
"v": "Root cause analysis"
},
{
"t": "small",
"v": "Remediation plan"
},
{
"t": "title",
"v": "Risk managed and issues contained",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "8"
},
{
"t": "title",
"v": "Suspension / restriction",
"big": false
},
{
"t": "title",
"v": "Human analogy: Suspension",
"big": false
},
{
"t": "p",
"v": "Restrict or pause the agent if risks increase, rules are breached or performance drops."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Restrict authority"
},
{
"t": "small",
"v": "Increase oversight"
},
{
"t": "small",
"v": "Root cause analysis"
},
{
"t": "small",
"v": "Remediation plan"
},
{
"t": "title",
"v": "Risk managed and issues contained",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "9"
},
{
"t": "title",
"v": "Retirement & offboarding",
"big": false
},
{
"t": "title",
"v": "Human analogy: Resignation or termination",
"big": false
},
{
"t": "p",
"v": "Retire the agent when no longer needed or superseded. Revoke access and archive records."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Revoke access"
},
{
"t": "small",
"v": "Disconnect tools"
},
{
"t": "small",
"v": "Data cleanup"
},
{
"t": "small",
"v": "Archive agent artefacts"
},
{
"t": "small",
"v": "Stakeholder notification"
},
{
"t": "title",
"v": "Safe retirement and clean closure",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "10"
},
{
"t": "title",
"v": "Archive & audit",
"big": false
},
{
"t": "title",
"v": "Human analogy: HR records retention & reference",
"big": false
},
{
"t": "p",
"v": "Retain records for audit, transparency and learning in accordance with retention rules."
},
{
"t": "eyebrow",
"v": "Key controls"
},
{
"t": "small",
"v": "Records archive"
},
{
"t": "small",
"v": "Audit trail retention"
},
{
"t": "small",
"v": "Lessons learned"
},
{
"t": "small",
"v": "Knowledge capture"
},
{
"t": "title",
"v": "Accountable, transparent and auditable",
"big": false
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Applied throughout every stage"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Continuous monitoring",
"big": false
},
{
"t": "title",
"v": "Human oversight",
"big": false
},
{
"t": "title",
"v": "Audit logging",
"big": false
},
{
"t": "title",
"v": "Policy compliance",
"big": false
},
{
"t": "title",
"v": "Security",
"big": false
},
{
"t": "title",
"v": "Privacy",
"big": false
},
{
"t": "title",
"v": "Safety",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Capability is not authority"
},
{
"t": "p",
"v": "An agent may be able to do many things, but can only do what it is authorised to do. Capability is about potential; authority is about permission. Capability + authority + context = action — only when all controls are satisfied can the agent act."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What the agent can do — capability"
},
{
"t": "p",
"v": "Technical potential and skills: the full range of capabilities, knowledge, tools and actions the agent is built or enabled to perform."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Natural language understanding",
"big": false
},
{
"t": "title",
"v": "Tool use & API integrations",
"big": false
},
{
"t": "title",
"v": "Reasoning & problem solving",
"big": false
},
{
"t": "title",
"v": "Workflow orchestration & scheduling",
"big": false
},
{
"t": "title",
"v": "Data retrieval (RAG)",
"big": false
},
{
"t": "title",
"v": "Analysis & visualisation",
"big": false
},
{
"t": "title",
"v": "Code generation & automation",
"big": false
},
{
"t": "title",
"v": "Communication (drafting, email, chat)",
"big": false
},
{
"t": "title",
"v": "Content creation & summarisation",
"big": false
},
{
"t": "title",
"v": "Learning from feedback",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Governance controls"
},
{
"t": "small",
"v": "Bridging capability to appropriate action."
},
{
"t": "p",
"v": "**Policies & rules**Organisational policies, legal requirements and ethical guardrails define boundaries."
},
{
"t": "p",
"v": "**Delegated authority**Explicit permissions granted for specific tasks, data and systems."
},
{
"t": "p",
"v": "**Conditions & controls**Context, thresholds, approvals and risk checks must be satisfied."
},
{
"t": "p",
"v": "**Human oversight**People remain in control of decisions and can intervene or override."
},
{
"t": "p",
"v": "**Monitoring & audit**Actions are logged, monitored and reviewed for assurance and compliance."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What the agent is allowed to do — authority"
},
{
"t": "p",
"v": "Permitted actions in the right context, within defined boundaries, for specific purposes and users."
},
{
"t": "p",
"v": "**Draft policy briefing**Approved for Policy Team (sensitivity: OFFICIAL)"
},
{
"t": "p",
"v": "**Send standard email response**Approved with template and review"
},
{
"t": "p",
"v": "**Summarise documents up to OFFICIAL**Approved with data classification filter"
},
{
"t": "p",
"v": "**Update internal tracker or system**Approved with role-based permissions"
},
{
"t": "p",
"v": "**Escalate issue to human reviewer**Required when confidence below threshold"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Why separate capability from authority"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Reduces risk** Limits exposure by ensuring agents cannot act beyond intent or policy."
},
{
"t": "p",
"v": "**Improves accountability** Clear ownership and delegated authority for all actions."
},
{
"t": "p",
"v": "**Supports compliance** Meets legal, regulatory and ethical obligations."
},
{
"t": "p",
"v": "**Enables agility** Adjust permissions without rebuilding capabilities."
},
{
"t": "p",
"v": "**Builds trust** Transparent controls and oversight build confidence."
},
{
"t": "p",
"v": "**Adapts to context** Decisions and permissions can vary by role, risk and situation."
}
],
"box": false
}
]
}
]
},
{
"id": "cascade",
"part": "operate-and-oversee",
"title": "Public service impact",
"icon": "trending_up",
"eyebrow": "Public service impact cascade",
"heading": "How agents influence decisions and deliver public value",
"lede": "Seven links from capability to legitimacy — each with the elements that must hold and the risks that arise if they do not. Better governed AI agents lead to better decisions, better services, and better outcomes for everyone.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "AI capability (agent)",
"big": false
},
{
"t": "p",
"v": "The agent has defined capabilities, knowledge, tools and guardrails within its delegated authority."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Purpose & scope"
},
{
"t": "small",
"v": "Data & knowledge"
},
{
"t": "small",
"v": "Tools & actions"
},
{
"t": "small",
"v": "Guardrails & policies"
},
{
"t": "small",
"v": "Monitoring in place"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Inaccurate data"
},
{
"t": "small",
"v": "Bias in training"
},
{
"t": "small",
"v": "Over-broad scope"
},
{
"t": "small",
"v": "Weak guardrails"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Civil servant (user)",
"big": false
},
{
"t": "p",
"v": "Civil servants interact with the agent, using its insights, suggestions or automation to support their work."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Trained & supported"
},
{
"t": "small",
"v": "Understood context"
},
{
"t": "small",
"v": "Human judgement"
},
{
"t": "small",
"v": "Challenge & review"
},
{
"t": "small",
"v": "Accountability clear"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Over-reliance"
},
{
"t": "small",
"v": "Skill gaps"
},
{
"t": "small",
"v": "Poor prompting"
},
{
"t": "small",
"v": "Lack of challenge"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Operational decision",
"big": false
},
{
"t": "p",
"v": "The agent's output informs or automates parts of operational decisions and actions within the service process."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Decision boundaries"
},
{
"t": "small",
"v": "Human oversight"
},
{
"t": "small",
"v": "Documentation"
},
{
"t": "small",
"v": "Compliance checks"
},
{
"t": "small",
"v": "Audit trail captured"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Automation bias"
},
{
"t": "small",
"v": "Unclear accountability"
},
{
"t": "small",
"v": "Policy non-compliance"
},
{
"t": "small",
"v": "Incomplete records"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Service outcome (internal)",
"big": false
},
{
"t": "p",
"v": "Operational actions produce internal service results such as case resolution, guidance issued, or processing completed."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Process efficiency"
},
{
"t": "small",
"v": "Accuracy & quality"
},
{
"t": "small",
"v": "Timeliness"
},
{
"t": "small",
"v": "Consistency"
},
{
"t": "small",
"v": "Cost effectiveness"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Process error"
},
{
"t": "small",
"v": "Delay or backlog"
},
{
"t": "small",
"v": "Poor integration"
},
{
"t": "small",
"v": "Resource waste"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Public experience (end users)",
"big": false
},
{
"t": "p",
"v": "Citizens, businesses and communities experience the service through interactions, outcomes and communications."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Accessible & inclusive"
},
{
"t": "small",
"v": "Fair & equitable"
},
{
"t": "small",
"v": "Clear & understandable"
},
{
"t": "small",
"v": "Safe & secure"
},
{
"t": "small",
"v": "Easy to interact with"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Exclusion"
},
{
"t": "small",
"v": "Confusing comms"
},
{
"t": "small",
"v": "Privacy concerns"
},
{
"t": "small",
"v": "Unequal treatment"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Public value & outcomes",
"big": false
},
{
"t": "p",
"v": "Better outcomes are achieved for people and society, improving trust in services and contributing to public value."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Better outcomes"
},
{
"t": "small",
"v": "Reduced burden"
},
{
"t": "small",
"v": "Improved equity"
},
{
"t": "small",
"v": "Economic value"
},
{
"t": "small",
"v": "Confidence & trust"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Unintended harm"
},
{
"t": "small",
"v": "Widening inequity"
},
{
"t": "small",
"v": "Benefit not realised"
},
{
"t": "small",
"v": "Poor measurement"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "7"
},
{
"t": "title",
"v": "Trust & legitimacy",
"big": false
},
{
"t": "p",
"v": "Transparent, accountable and responsible use of AI strengthens public trust and democratic legitimacy."
},
{
"t": "eyebrow",
"v": "Key elements"
},
{
"t": "small",
"v": "Transparency"
},
{
"t": "small",
"v": "Accountability"
},
{
"t": "small",
"v": "Safety & ethics"
},
{
"t": "small",
"v": "Oversight & redress"
},
{
"t": "small",
"v": "Continuous improvement"
},
{
"t": "eyebrow",
"v": "Risks at this stage"
},
{
"t": "small",
"v": "Lack of transparency"
},
{
"t": "small",
"v": "Accountability gaps"
},
{
"t": "small",
"v": "Safety incidents"
},
{
"t": "small",
"v": "Erosion of trust"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Enablers & controls across the cascade"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Strong governance & ownership",
"big": false
},
{
"t": "title",
"v": "Policies & ethical use",
"big": false
},
{
"t": "title",
"v": "Security & privacy by design",
"big": false
},
{
"t": "title",
"v": "Quality data & provenance",
"big": false
},
{
"t": "title",
"v": "Monitoring & auditability",
"big": false
},
{
"t": "title",
"v": "Human oversight at every step",
"big": false
},
{
"t": "title",
"v": "Transparency (ATRS)",
"big": false
},
{
"t": "title",
"v": "Performance measurement",
"big": false
},
{
"t": "title",
"v": "Continuous improvement",
"big": false
}
]
}
]
}
]
},
{
"id": "observatory",
"part": "operate-and-oversee",
"title": "Observatory & barometer",
"icon": "insights",
"eyebrow": "Observatory & barometer",
"heading": "Published signals and a readiness assessment",
"lede": "The observatory tracks published signals of UK public-sector AI adoption; the model is the OECD.AI Policy Observatory, applied at home. Each figure carries its publisher and date. The barometer is your own assessment, stored on this device only.",
"body": [
{
"t": "h",
"l": 2,
"v": "Observatory — published signals"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "fact_check"
},
{
"t": "eyebrow",
"v": "OECD.AI · 2026"
},
{
"t": "title",
"v": "125+",
"big": true
},
{
"t": "p",
"v": "ATRS transparency records published. Mandatory reporting for central departments and ALBs since 2025."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "emoji_objects"
},
{
"t": "eyebrow",
"v": "Civil Service · 2026"
},
{
"t": "stat",
"v": "252"
},
{
"t": "p",
"v": "ideas entered in the Civil Service AI & Data Challenge 2026 — a 160% increase, in its fourth year."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "eyebrow",
"v": "DSIT analysis · 2026"
},
{
"t": "title",
"v": "£68bn",
"big": true
},
{
"t": "small",
"v": "AI investment pledged into the UK since January 2025."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "memory"
},
{
"t": "eyebrow",
"v": "delivery.ai.gov.uk · 2026"
},
{
"t": "title",
"v": "21 → 420",
"big": true
},
{
"t": "p",
"v": "UK AI compute in ExaFLOPs: 2025 actual to 2030 commitment (from 2 in 2024)."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "school"
},
{
"t": "eyebrow",
"v": "DSIT analysis · 2026"
},
{
"t": "stat",
"v": "200k"
},
{
"t": "small",
"v": "People studying AI-related higher-education programmes in the UK."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "account_balance"
},
{
"t": "eyebrow",
"v": "Heriot-Watt & GoLLM · 2025"
},
{
"t": "title",
"v": "1 in 3",
"big": true
},
{
"t": "p",
"v": "Scottish councils piloting or using AI. The leading UK nation for local-government adoption."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Barometer — your readiness"
},
{
"t": "p",
"v": "Rate each dimension 0–4. Saved locally on this device; nothing leaves the page."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "—",
"big": true
},
{
"t": "title",
"v": "Rate the dimensions for a reading",
"big": false
},
{
"t": "small",
"v": "0 of 8 dimensions rated"
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Strategy & leadership",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Governance & assurance",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Data readiness",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Skills & multidisciplinary team",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Security posture",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Ethics & transparency",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Delivery capability",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Monitoring & evaluation",
"big": false
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "value",
"part": "fund-and-buy",
"title": "Cost, value & investment",
"icon": "savings",
"eyebrow": "Cost, value & investment · illustrative figures",
"heading": "Make every AI pound count",
"lede": "A continuous cycle that connects business value, cost efficiency and responsible use. Align, optimise, govern — then deliver and reinvest. Figures on this page are illustrative: replace them with your own contract and telemetry data.",
"body": [
{
"t": "table",
"rows": [
[
{
"h": false,
"v": "**1** **Forecast & plan** Understand demand, use cases, models and unit economics."
},
{
"h": false,
"v": "**2** **Allocate & budget** Allocate costs, set budgets and guardrails."
},
{
"h": false,
"v": "**3** **Observe & measure** Monitor usage, costs, quality and performance in real time."
},
{
"h": false,
"v": "**4** **Analyse & evaluate** Analyse spend, quality, outcomes and business value by use case."
}
],
[
{
"h": false,
"v": "**5** **Optimise & improve** Optimise models, workloads, prompts and architecture."
},
{
"h": false,
"v": "**6** **Report & transparency** Report on spend, value, carbon and compliance."
},
{
"h": false,
"v": "**7** **Reinvest or rebalance** Reinvest in high-value initiatives or rebalance priorities."
},
{
"h": false,
"v": "**8** **Govern & control** Review policies, controls and risks. Adapt and evolve."
}
]
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "What we manage"
},
{
"t": "p",
"v": "**Consumption costs**Tokens, credits, API calls, capacity, compute, storage, network"
},
{
"t": "p",
"v": "**Technology costs**Models, platforms, connectors, licences, infrastructure"
},
{
"t": "p",
"v": "**Adoption costs**Training, change, champions, workflow redesign, support"
},
{
"t": "p",
"v": "**Data & integration costs**Data preparation, quality, integration, access and governance"
},
{
"t": "p",
"v": "**Assurance & risk costs**Security, privacy, testing, human review, audits, incident response"
},
{
"t": "p",
"v": "**Operational costs**Monitoring, tooling, run support, maintenance, model updates"
},
{
"t": "p",
"v": "**Transition & exit costs**Migration, parallel run, data portability, vendor switching"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Value we deliver"
},
{
"t": "p",
"v": "**Better public services**Improved outcomes for citizens"
},
{
"t": "p",
"v": "**Productivity gains**Time saved, efficiency improved"
},
{
"t": "p",
"v": "**Quality & consistency**Fewer errors, higher quality outputs"
},
{
"t": "p",
"v": "**Risk reduction**Stronger controls, fewer incidents"
},
{
"t": "p",
"v": "**Transparency & trust**Clear oversight, explainable decisions"
},
{
"t": "p",
"v": "**Sustainable impact**Lower waste, optimised resources, lower carbon"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Key practices"
},
{
"t": "small",
"v": "Tag and track by use case"
},
{
"t": "small",
"v": "Set budgets and alerts"
},
{
"t": "small",
"v": "Right-size models and workloads"
},
{
"t": "small",
"v": "Reduce token waste"
},
{
"t": "small",
"v": "Automate policies and guardrails"
},
{
"t": "small",
"v": "Review value regularly"
},
{
"t": "small",
"v": "Collaborate across business, technology, finance and security"
},
{
"t": "h",
"l": 3,
"v": "Who works together"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Business owners",
"big": false
},
{
"t": "title",
"v": "Product & delivery",
"big": false
},
{
"t": "title",
"v": "Finance / FinOps",
"big": false
},
{
"t": "title",
"v": "Engineering & data",
"big": false
},
{
"t": "title",
"v": "Security & privacy",
"big": false
},
{
"t": "title",
"v": "Procurement & commercial",
"big": false
},
{
"t": "title",
"v": "Operations & service management",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Unit economics: what one interaction costs, and what it returns"
},
{
"t": "p",
"v": "Total cost per interaction = token cost + infrastructure + tool and API cost + human effort + overhead. Multiply by volume for the period, and set it against the value delivered. The starting values are the source model's worked example."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cost inputs"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cost per interaction"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Input tokens",
"big": false
},
{
"t": "small",
"v": "2,000 tokens @ £0.002 / 1K"
},
{
"t": "title",
"v": "£0.0040",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Output tokens",
"big": false
},
{
"t": "small",
"v": "1,000 tokens @ £0.006 / 1K"
},
{
"t": "title",
"v": "£0.0060",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Tool / API calls",
"big": false
},
{
"t": "small",
"v": "External calls and connectors"
},
{
"t": "title",
"v": "£0.010",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Infrastructure",
"big": false
},
{
"t": "small",
"v": "Compute, storage and network allocation"
},
{
"t": "title",
"v": "£0.0050",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Human effort",
"big": false
},
{
"t": "small",
"v": "5 minutes @ £1 / min"
},
{
"t": "title",
"v": "£5.00",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Overheads",
"big": false
},
{
"t": "small",
"v": "Operations, support and monitoring"
},
{
"t": "title",
"v": "£0.500",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Total cost per interaction",
"big": false
},
{
"t": "stat",
"v": "£5.53"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "× 1,000 interactions in the period",
"big": false
},
{
"t": "title",
"v": "£5,525.00",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Value per interaction"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Total value",
"big": false
},
{
"t": "title",
"v": "£20.00",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "p",
"v": "**NET VALUE** **£14.48**"
},
{
"t": "p",
"v": "**RETURN** **262%**"
}
]
},
{
"t": "p",
"v": "Value to cost ratio 3.6× · value across the period £20,000.00. Human review time is usually the largest single line — design it down before changing model."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Workload / request",
"big": false
},
{
"t": "small",
"v": "User or system initiates a task"
},
{
"t": "title",
"v": "Examples",
"big": false
},
{
"t": "small",
"v": "Summarise document"
},
{
"t": "small",
"v": "Draft policy response"
},
{
"t": "small",
"v": "Code generation"
},
{
"t": "small",
"v": "Case classification"
},
{
"t": "small",
"v": "Data analysis"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Tokens & model consumption",
"big": false
},
{
"t": "small",
"v": "Tokens processed by the model"
},
{
"t": "title",
"v": "Components",
"big": false
},
{
"t": "small",
"v": "Input tokens (prompt)"
},
{
"t": "small",
"v": "Output tokens (response)"
},
{
"t": "small",
"v": "Reasoning / tool use tokens"
},
{
"t": "small",
"v": "Cached tokens"
},
{
"t": "small",
"v": "Embedding tokens"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Cost drivers",
"big": false
},
{
"t": "small",
"v": "What drives the cost of the interaction"
},
{
"t": "title",
"v": "Cost elements",
"big": false
},
{
"t": "small",
"v": "Model pricing per 1K tokens"
},
{
"t": "small",
"v": "Context length"
},
{
"t": "small",
"v": "Frequency & volume"
},
{
"t": "small",
"v": "Tool / API calls"
},
{
"t": "small",
"v": "Storage & compute"
},
{
"t": "small",
"v": "Network / egress"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Quality & outcomes",
"big": false
},
{
"t": "small",
"v": "Quality of output and outcome achieved"
},
{
"t": "title",
"v": "Quality factors",
"big": false
},
{
"t": "small",
"v": "Accuracy & relevance"
},
{
"t": "small",
"v": "Hallucination rate"
},
{
"t": "small",
"v": "Completeness"
},
{
"t": "small",
"v": "Safety & compliance"
},
{
"t": "small",
"v": "User satisfaction"
},
{
"t": "small",
"v": "Rework rate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Human effort & rework",
"big": false
},
{
"t": "small",
"v": "Human time to review, edit or redo"
},
{
"t": "title",
"v": "Effort factors",
"big": false
},
{
"t": "small",
"v": "Review time"
},
{
"t": "small",
"v": "Editing / refinement"
},
{
"t": "small",
"v": "Additional research"
},
{
"t": "small",
"v": "Escalations"
},
{
"t": "small",
"v": "Rework / repeat runs"
},
{
"t": "small",
"v": "Approval steps"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Business value",
"big": false
},
{
"t": "small",
"v": "Value delivered to the organisation"
},
{
"t": "title",
"v": "Value types",
"big": false
},
{
"t": "small",
"v": "Time saved"
},
{
"t": "small",
"v": "Productivity gain"
},
{
"t": "small",
"v": "Better decisions"
},
{
"t": "small",
"v": "Risk reduction"
},
{
"t": "small",
"v": "Improved quality"
},
{
"t": "small",
"v": "Citizen / service impact"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cost and value optimisation levers"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "**Right model**Match quality needs to the most cost-effective model"
},
{
"t": "p",
"v": "**Prompt engineering**Optimise prompts to reduce tokens and improve outcomes"
},
{
"t": "p",
"v": "**Context management**Right-size context, use retrieval and caching"
},
{
"t": "p",
"v": "**Routing & policies**Route to the best model or tier based on cost and risk"
},
{
"t": "p",
"v": "**Caching & reuse**Reuse responses, embeddings and results"
},
{
"t": "p",
"v": "**Evaluation & guardrails**Measure quality and set thresholds to avoid waste"
},
{
"t": "p",
"v": "**FinOps controls**Budgets, alerts, showback / chargeback and governance"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "KPIs to track"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Cost per 1,000 tokens",
"big": false
},
{
"t": "title",
"v": "Cost per interaction",
"big": false
},
{
"t": "title",
"v": "Cost per outcome",
"big": false
},
{
"t": "title",
"v": "Quality score",
"big": false
},
{
"t": "title",
"v": "Rework rate",
"big": false
},
{
"t": "title",
"v": "Time saved",
"big": false
},
{
"t": "title",
"v": "ROI / value per £",
"big": false
},
{
"t": "title",
"v": "Budget adherence",
"big": false
},
{
"t": "title",
"v": "Token utilisation efficiency",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Six cost layers, each with an owner"
},
{
"t": "p",
"v": "Fair, transparent, actionable. Allocate costs across the organisation to drive accountability, optimise spend and demonstrate value. Without a named owner, a layer is not allocated — it is absorbed."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "LAYER 1",
"big": false
},
{
"t": "h",
"l": 3,
"v": "Enterprise foundation"
},
{
"t": "small",
"v": "Core platform and shared capabilities enabling all AI use."
},
{
"t": "small",
"v": "Azure / cloud baseline"
},
{
"t": "small",
"v": "Identity & security"
},
{
"t": "small",
"v": "Networking"
},
{
"t": "small",
"v": "Monitoring platform"
},
{
"t": "small",
"v": "Central services"
},
{
"t": "p",
"v": "**OWNER** **CTO / enterprise** Direct allocation by actual cost"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "LAYER 2",
"big": false
},
{
"t": "h",
"l": 3,
"v": "Shared platform services"
},
{
"t": "small",
"v": "Reusable services consumed by multiple products and teams."
},
{
"t": "small",
"v": "AI platforms / services"
},
{
"t": "small",
"v": "Data platforms"
},
{
"t": "small",
"v": "Model hosting"
},
{
"t": "small",
"v": "Logging & observability"
},
{
"t": "small",
"v": "Shared tools & libraries"
},
{
"t": "p",
"v": "**OWNER** **Platform / centre of excellence** Cost pools and apportionment drivers"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "LAYER 3",
"big": false
},
{
"t": "h",
"l": 3,
"v": "Product / service"
},
{
"t": "small",
"v": "AI capabilities built into a product or service."
},
{
"t": "small",
"v": "Feature development"
},
{
"t": "small",
"v": "Data preparation"
},
{
"t": "small",
"v": "Models & fine-tuning"
},
{
"t": "small",
"v": "QA & evaluation"
},
{
"t": "small",
"v": "Product support"
},
{
"t": "p",
"v": "**OWNER** **Product / service owner** Showback or chargeback to products"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "LAYER 4",
"big": false
},
{
"t": "h",
"l": 3,
"v": "Team / workload"
},
{
"t": "small",
"v": "Teams running workloads to deliver features or operations."
},
{
"t": "small",
"v": "Training / inference"
},
{
"t": "small",
"v": "API calls / tokens"
},
{
"t": "small",
"v": "Experimentation"
},
{
"t": "small",
"v": "Environments"
},
{
"t": "small",
"v": "Team tooling"
},
{
"t": "p",
"v": "**OWNER** **Team lead** Team budgets and workload tagging"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "LAYER 5",
"big": false
},
{
"t": "h",
"l": 3,
"v": "Use case / consumer"
},
{
"t": "small",
"v": "A specific AI use case delivering value to a defined user group."
},
{
"t": "small",
"v": "End-user interactions"
},
{
"t": "small",
"v": "Business process cost"
},
{
"t": "small",
"v": "Human-in-the-loop"
},
{
"t": "small",
"v": "Quality assurance"
},
{
"t": "small",
"v": "Change & adoption"
},
{
"t": "p",
"v": "**OWNER** **Use case owner** Usage-based allocation (tokens, calls, transactions)"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "LAYER 6",
"big": false
},
{
"t": "h",
"l": 3,
"v": "End user / beneficiary"
},
{
"t": "small",
"v": "Cost associated with delivery to end users or citizens."
},
{
"t": "small",
"v": "Per user / transaction"
},
{
"t": "small",
"v": "Support & helpdesk"
},
{
"t": "small",
"v": "Feedback & outcomes"
},
{
"t": "small",
"v": "Accessibility support"
},
{
"t": "small",
"v": "Service impact"
},
{
"t": "p",
"v": "**OWNER** **Business / service** Unit cost per user or beneficiary"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Allocation flow"
},
{
"t": "title",
"v": "1. Capture actual costs",
"big": false
},
{
"t": "title",
"v": "2. Tag & classify costs",
"big": false
},
{
"t": "title",
"v": "3. Allocate using defined rules",
"big": false
},
{
"t": "title",
"v": "4. Report & analyse",
"big": false
},
{
"t": "title",
"v": "5. Optimise & act",
"big": false
},
{
"t": "title",
"v": "6. Value realised",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What we gain"
},
{
"t": "small",
"v": "Clear visibility of total cost"
},
{
"t": "small",
"v": "Fair and consistent allocation"
},
{
"t": "small",
"v": "Informed decision making"
},
{
"t": "small",
"v": "Cost optimisation"
},
{
"t": "small",
"v": "Accountability and ownership"
},
{
"t": "small",
"v": "Evidence of value and return"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Data & enablers"
},
{
"t": "small",
"v": "Consistent tagging & metadata"
},
{
"t": "small",
"v": "Usage & cost data feeds"
},
{
"t": "small",
"v": "Access & permissions"
},
{
"t": "small",
"v": "Cost ownership model"
},
{
"t": "small",
"v": "Policies & standards"
},
{
"t": "small",
"v": "Automation & integration"
},
{
"t": "small",
"v": "Data quality & validation"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Governance & controls"
},
{
"t": "small",
"v": "Cost policies & guardrails"
},
{
"t": "small",
"v": "Budgeting & approvals"
},
{
"t": "small",
"v": "Monitoring & alerts"
},
{
"t": "small",
"v": "Audit & reconciliation"
},
{
"t": "small",
"v": "Roles & responsibilities"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Visible costs are the tip of the iceberg"
},
{
"t": "p",
"v": "Licences, tokens and compute are usually budgeted and tracked, and are only a fraction of the total. The hidden lines below the waterline determine real value and long-term success."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "title",
"v": "Licences & subscriptions",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "code"
},
{
"t": "title",
"v": "Token / usage charges",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "cloud"
},
{
"t": "title",
"v": "Cloud compute & storage",
"big": false
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Hidden costs — what is often overlooked"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Data & information"
},
{
"t": "small",
"v": "Data quality & remediation"
},
{
"t": "small",
"v": "Information architecture"
},
{
"t": "small",
"v": "Permissions & access"
},
{
"t": "small",
"v": "Metadata & cataloguing"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Integration & engineering"
},
{
"t": "small",
"v": "Connectors & APIs"
},
{
"t": "small",
"v": "Workflow redesign"
},
{
"t": "small",
"v": "Systems integration"
},
{
"t": "small",
"v": "Environment set-up"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Security & compliance"
},
{
"t": "small",
"v": "Security controls"
},
{
"t": "small",
"v": "Privacy & DPIAs"
},
{
"t": "small",
"v": "Compliance & legal"
},
{
"t": "small",
"v": "Pen testing & assurance"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Adoption & change"
},
{
"t": "small",
"v": "Training & enablement"
},
{
"t": "small",
"v": "Champions & local support"
},
{
"t": "small",
"v": "Communications"
},
{
"t": "small",
"v": "Workflow & culture change"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Assurance & evaluation"
},
{
"t": "small",
"v": "Model evaluation"
},
{
"t": "small",
"v": "Bias, fairness & robustness"
},
{
"t": "small",
"v": "Human review effort"
},
{
"t": "small",
"v": "Audit & red teaming"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Operations & support"
},
{
"t": "small",
"v": "Monitoring & observability"
},
{
"t": "small",
"v": "Incident response"
},
{
"t": "small",
"v": "Helpdesk & support"
},
{
"t": "small",
"v": "Continual improvement"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "FinOps & management"
},
{
"t": "small",
"v": "Cost monitoring & allocation"
},
{
"t": "small",
"v": "Forecasting & optimisation"
},
{
"t": "small",
"v": "Showback / chargeback"
},
{
"t": "small",
"v": "Demand management"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Governance & risk"
},
{
"t": "small",
"v": "Policies & standards"
},
{
"t": "small",
"v": "Risk management"
},
{
"t": "small",
"v": "Third-party risk"
},
{
"t": "small",
"v": "Records & retention"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Workforce & roles"
},
{
"t": "small",
"v": "Role evolution"
},
{
"t": "small",
"v": "Skills development"
},
{
"t": "small",
"v": "Capacity & backfill"
},
{
"t": "small",
"v": "HR & employee relations"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Sustainability"
},
{
"t": "small",
"v": "Energy & carbon impact"
},
{
"t": "small",
"v": "Efficient model usage"
},
{
"t": "small",
"v": "Sustainable operations"
},
{
"t": "small",
"v": "Reporting & metrics"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Transition & exit"
},
{
"t": "small",
"v": "Parallel running"
},
{
"t": "small",
"v": "Data portability"
},
{
"t": "small",
"v": "Supplier exit costs"
},
{
"t": "small",
"v": "Migration & decommission"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Value realisation"
},
{
"t": "small",
"v": "Benefit tracking"
},
{
"t": "small",
"v": "Outcome measurement"
},
{
"t": "small",
"v": "Realisation management"
},
{
"t": "small",
"v": "Continuous feedback"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "Total cost of AI = technology + data + people + process + assurance + operations + commercial. Budget for the full cost. Manage it actively. Realise the value."
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What drives hidden costs"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "**Complexity**Many moving parts across technology, data and people"
},
{
"t": "p",
"v": "**Change**Workflows, roles and ways of working must adapt"
},
{
"t": "p",
"v": "**Risk & assurance**Higher risk requires stronger controls and evidence"
},
{
"t": "p",
"v": "**Scale & usage**More users, data and interactions increase ongoing costs"
},
{
"t": "p",
"v": "**Governance**Policies, oversight and reporting add essential investment"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Make the hidden visible"
},
{
"t": "title",
"v": "Map all cost categories",
"big": false
},
{
"t": "title",
"v": "Track usage & value",
"big": false
},
{
"t": "title",
"v": "Allocate transparently",
"big": false
},
{
"t": "title",
"v": "Optimise continuously",
"big": false
},
{
"t": "title",
"v": "Review & adapt",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Beyond licences: adoption as an investment line"
},
{
"t": "p",
"v": "AI adoption is a long-term investment across technology, people, process, data and assurance. Sustainable value comes from funding the whole system, not the licence."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "1. Technology"
},
{
"t": "small",
"v": "Licences & subscriptions"
},
{
"t": "small",
"v": "Tokens / consumption"
},
{
"t": "small",
"v": "Models & platforms"
},
{
"t": "small",
"v": "Infrastructure & cloud"
},
{
"t": "small",
"v": "Tools & integrations"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "2. People & adoption"
},
{
"t": "small",
"v": "Training & skills"
},
{
"t": "small",
"v": "Champions & support"
},
{
"t": "small",
"v": "Communications"
},
{
"t": "small",
"v": "User research"
},
{
"t": "small",
"v": "Protected time"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "3. Process & change"
},
{
"t": "small",
"v": "Workflow redesign"
},
{
"t": "small",
"v": "New ways of working"
},
{
"t": "small",
"v": "Governance processes"
},
{
"t": "small",
"v": "Service redesign"
},
{
"t": "small",
"v": "Organisational change"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "4. Data & information"
},
{
"t": "small",
"v": "Data quality & remediation"
},
{
"t": "small",
"v": "Information management"
},
{
"t": "small",
"v": "Permissions & access"
},
{
"t": "small",
"v": "Knowledge management"
},
{
"t": "small",
"v": "Taxonomy & metadata"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "5. Assurance & risk"
},
{
"t": "small",
"v": "Security & privacy"
},
{
"t": "small",
"v": "Responsible AI"
},
{
"t": "small",
"v": "Testing & evaluation"
},
{
"t": "small",
"v": "Human review"
},
{
"t": "small",
"v": "Audit & compliance"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "6. Operations & continuous improvement"
},
{
"t": "small",
"v": "Monitoring & observability"
},
{
"t": "small",
"v": "Support & maintenance"
},
{
"t": "small",
"v": "Incident management"
},
{
"t": "small",
"v": "Benefit measurement"
},
{
"t": "small",
"v": "Iteration & enhancement"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Discover & plan** Identify needs, use cases and business value. Build the investment case."
},
{
"t": "p",
"v": "**2. Pilot & experiment** Test, learn and validate value, risk and cost. Measure and refine."
},
{
"t": "p",
"v": "**3. Scale & adopt** Roll out to teams and services. Build capability."
},
{
"t": "p",
"v": "**4. Operate & optimise** Monitor usage, value and cost. Optimise models and processes."
},
{
"t": "p",
"v": "**5. Sustain & evolve** Continually improve, adapt and reinvest for future value."
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Why invest"
},
{
"t": "small",
"v": "Improve public services and citizen outcomes"
},
{
"t": "small",
"v": "Increase productivity and efficiency"
},
{
"t": "small",
"v": "Enhance decision making with better information"
},
{
"t": "small",
"v": "Reduce risk and strengthen compliance"
},
{
"t": "small",
"v": "Drive innovation and long-term value"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Where value is realised"
},
{
"t": "small",
"v": "Faster, more consistent service delivery"
},
{
"t": "small",
"v": "Better experience for users and citizens"
},
{
"t": "small",
"v": "Lower cost to serve over time"
},
{
"t": "small",
"v": "Improved quality, accuracy and compliance"
},
{
"t": "small",
"v": "Stronger resilience and adaptability"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Funding approaches"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Central funding",
"big": false
},
{
"t": "title",
"v": "Departmental budgets",
"big": false
},
{
"t": "title",
"v": "Programme funds",
"big": false
},
{
"t": "title",
"v": "Shared services",
"big": false
},
{
"t": "title",
"v": "Outcome / value based funding",
"big": false
},
{
"t": "title",
"v": "Public-private partnerships",
"big": false
}
]
},
{
"t": "h",
"l": 3,
"v": "Measure what matters"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Adoption rate",
"big": false
},
{
"t": "title",
"v": "Usage & engagement",
"big": false
},
{
"t": "title",
"v": "Quality & accuracy",
"big": false
},
{
"t": "title",
"v": "Cost per transaction",
"big": false
},
{
"t": "title",
"v": "Time saved",
"big": false
},
{
"t": "title",
"v": "Outcome & impact",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Without full investment"
},
{
"t": "small",
"v": "Under-adoption and poor outcomes"
},
{
"t": "small",
"v": "Hidden costs and budget shocks"
},
{
"t": "small",
"v": "Security, compliance and quality risks"
},
{
"t": "small",
"v": "Duplication and technical debt"
},
{
"t": "small",
"v": "Lost trust and stalled programmes"
}
]
}
],
"box": false
}
]
},
{
"id": "finops",
"part": "fund-and-buy",
"title": "FinOps & models",
"icon": "payments",
"eyebrow": "FinOps & models",
"heading": "A builder's design choice sets a budget holder's cost",
"lede": "AI cost is a portfolio, not a licence line. You are billed per token, in and out — roughly four characters or three-quarters of a word each; an A4 page is 600–800 tokens. Micro-costs stack when an agent runs on repeat. Report by service, product, agent, owner, model and cost driver; showback before chargeback; no expenditure classification without finance-led assessment.",
"body": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Licences",
"big": false
},
{
"t": "title",
"v": "Credits",
"big": false
},
{
"t": "title",
"v": "Model calls",
"big": false
},
{
"t": "title",
"v": "Compute",
"big": false
},
{
"t": "title",
"v": "Storage",
"big": false
},
{
"t": "title",
"v": "Connectors",
"big": false
},
{
"t": "title",
"v": "Monitoring",
"big": false
},
{
"t": "title",
"v": "Evaluation",
"big": false
},
{
"t": "title",
"v": "Adoption & training",
"big": false
},
{
"t": "title",
"v": "Remediation",
"big": false
},
{
"t": "title",
"v": "Security",
"big": false
},
{
"t": "title",
"v": "Human review",
"big": false
},
{
"t": "title",
"v": "Support",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "How the two main vendors now charge, and where they process"
},
{
"t": "title",
"v": "Published list positions at 19 October 2026",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Microsoft Copilot",
"big": false
},
{
"t": "p",
"v": "**Cost:** a per-user licence ($30 per month list) covers Chat and Copilot in the Office apps. Cowork, Code, Autopilot and frontier models consume Copilot Credits at $0.01 each on pay-as-you-go. Agent security for custom agents needs Agent 365."
},
{
"t": "p",
"v": "**Residency:** follows the tenant's Microsoft 365 geography. Web grounding, subprocessor models (Claude, Grok) and Code apps on Managed Runtime need separate checks."
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Anthropic Claude",
"big": false
},
{
"t": "p",
"v": "**Cost:** a seat per user, plus usage at token rates by model (Opus 5.5 lists at $4 input / $20 output per million tokens). Model routing, effort settings and prompt caching are the main levers."
},
{
"t": "p",
"v": "**Residency:** set by deployment route. Anthropic-processed with a US-only option, or in-region through Bedrock, Google Cloud or Foundry. Connectors carry the content they fetch to the model."
}
]
}
],
"box": false
},
{
"t": "title",
"v": "See all 39 capabilities compared in the capability atlas →",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Cost lab — the same job, five model classes"
},
{
"t": "p",
"v": "Illustrative list prices from the \"Where your AI spend goes\" explainer (2025), GBP per 1,000 tokens — verify current pricing before any decision. Non-English output can cost more: the same words take more tokens."
},
{
"t": "row",
"items": [
{
"t": "p",
"v": "**GPT-4o class (OpenAI) **Balance of quality and tone"
},
{
"t": "p",
"v": "**£0.017****per run**"
},
{
"t": "p",
"v": "**£1.71****per month**"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Claude Haiku class (Anthropic) **Cost-efficient, high-volume work"
},
{
"t": "p",
"v": "**£0.0043****per run**"
},
{
"t": "p",
"v": "**£0.427****per month**"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Claude Opus class (Anthropic) **Deep, high-context threads"
},
{
"t": "p",
"v": "**£0.026****per run**"
},
{
"t": "p",
"v": "**£2.56****per month**"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Mixtral class (Mistral) · cheapest**Ultra-low cost; needs prompt tuning"
},
{
"t": "p",
"v": "**£0.0032****per run**"
},
{
"t": "p",
"v": "**£0.320****per month**"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Gemini Pro class (Google) **On par with GPT-4o by usage tier"
},
{
"t": "p",
"v": "**£0.017****per run**"
},
{
"t": "p",
"v": "**£1.71****per month**"
}
],
"box": true
},
{
"t": "p",
"v": "Design levers before model choice: reuse outputs, cache repeated answers, trim inputs, use plain logic before calling a model."
},
{
"t": "h",
"l": 2,
"v": "Four lenses of value — the estate case, measured and modelled"
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Reclaimed time",
"big": false
},
{
"t": "title",
"v": "~£0.5M–£1.6M / yr at ~680 staff",
"big": false
},
{
"t": "p",
"v": "A 10% findability gain on ~1.8 hrs/day spent searching. The big lever — and it can cover the licence bill 2.5–8× over (\"earned allocation\")."
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Storage cost",
"big": false
},
{
"t": "title",
"v": "~£1.5k / yr",
"big": false
},
{
"t": "small",
"v": "Effectively £0 marginal under bundled M365. Symbolic."
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Environmental",
"big": false
},
{
"t": "title",
"v": "~9 TB ≈ 55–360 kg CO₂e / yr",
"big": false
},
{
"t": "small",
"v": "Modest; aligns with Greening Government; curbs future growth."
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "Wellbeing",
"big": false
},
{
"t": "title",
"v": "117 emails/day · ~23 min to refocus",
"big": false
},
{
"t": "p",
"v": "Interrupted every ~2 minutes; 48% find work chaotic and fragmented. The human case."
}
],
"box": true
}
]
},
{
"t": "p",
"v": "**Rules that hold:** require total-cost and portability evidence before scaling any tool or agent · per-agent consumption caps and anomaly alerts · unit economics per production agent published within 12 months · value measured through the evidence chain: activity → output quality → workflow change → organisational outcome → public outcome."
}
]
},
{
"id": "commercial",
"part": "fund-and-buy",
"title": "Commercial & suppliers",
"icon": "handshake",
"eyebrow": "Commercial barriers · illustrative default scores",
"heading": "Barriers that slow adoption, increase cost and limit sustainability",
"lede": "Score each barrier by its impact and its likelihood in your context. Risk score is impact × likelihood; the band follows the published scale. The defaults are the sector-level assessment — change them and the bands move.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Critical",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "High",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Medium",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Low",
"big": false
}
]
}
],
"box": false
},
{
"t": "title",
"v": "4 barriers are critical and 8 high in this scoring. Address these before scale.",
"big": false
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "BARRIER",
"big": false
},
{
"t": "title",
"v": "IMPACT",
"big": false
},
{
"t": "title",
"v": "LIKELIHOOD",
"big": false
},
{
"t": "title",
"v": "I×L",
"big": false
},
{
"t": "title",
"v": "RISK",
"big": false
},
{
"t": "title",
"v": "TYPICAL CURRENT STATE",
"big": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Licensing clarity** Unclear boundaries between included features, credits, capacity and pay-as-you-go."
},
{
"t": "num",
"v": "20"
},
{
"t": "title",
"v": "Critical",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Moving boundaries"
},
{
"t": "small",
"v": "Preview versus general availability confusion"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Cost & usage transparency** Limited access to usage data, telemetry, forecasting and billing detail."
},
{
"t": "num",
"v": "20"
},
{
"t": "title",
"v": "Critical",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Trials paused"
},
{
"t": "small",
"v": "Inability to model total cost of ownership"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Price predictability** Usage-based pricing makes budgets volatile and hard to forecast."
},
{
"t": "num",
"v": "16"
},
{
"t": "title",
"v": "Critical",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Variable token costs"
},
{
"t": "small",
"v": "Capacity uncertainty"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Supplier / platform lock-in** High switching costs, proprietary connectors and formats."
},
{
"t": "num",
"v": "16"
},
{
"t": "title",
"v": "Critical",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Proprietary ecosystems"
},
{
"t": "small",
"v": "Data portability gaps"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Telemetry & audit rights** Lack of contractual rights to monitor, audit and export operational data."
},
{
"t": "num",
"v": "12"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Limited audit trails"
},
{
"t": "small",
"v": "Evidence gaps"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Data ownership & IP** Uncertainty over prompts, outputs, data use, training and generated code."
},
{
"t": "num",
"v": "12"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "IP ambiguity"
},
{
"t": "small",
"v": "Data use concerns"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Data residency & jurisdiction** Data and model processing across jurisdictions may create compliance risk."
},
{
"t": "num",
"v": "12"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Cross-border processing"
},
{
"t": "small",
"v": "Sovereignty concerns"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Service level & support** Unclear SLAs, responsibilities and support across multiple products and models."
},
{
"t": "num",
"v": "9"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Shared responsibility"
},
{
"t": "small",
"v": "Support fragmentation"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Interoperability & portability** Limited open standards, export options and easy integration with other systems."
},
{
"t": "num",
"v": "12"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Custom connectors"
},
{
"t": "small",
"v": "Integration complexity"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Commercial complexity** Complex contracts, overlapping costs, multiple components and change over time."
},
{
"t": "num",
"v": "9"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Complex terms"
},
{
"t": "small",
"v": "Change without notice"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Small body disadvantage** Smaller organisations lack commercial leverage and specialist expertise."
},
{
"t": "num",
"v": "12"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Limited negotiation power"
},
{
"t": "small",
"v": "Resource constraints"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Exit & transition risk** Uncertain exit costs, data extraction, migration and parallel running."
},
{
"t": "num",
"v": "12"
},
{
"t": "title",
"v": "High",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Exit complexity"
},
{
"t": "small",
"v": "Transition cost risk"
}
]
}
],
"box": false
}
]
},
{
"t": "p",
"v": "Bands are computed from the published impact × likelihood scale. Two rows on the source board carry a stated risk level that does not match its own band scale; the computed band is shown here."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Common examples"
},
{
"t": "small",
"v": "Trial paused due to lack of cost / usage data"
},
{
"t": "small",
"v": "Unexpected charges from token consumption"
},
{
"t": "small",
"v": "Unable to export data or prompts"
},
{
"t": "small",
"v": "Contract terms prevent model substitution"
},
{
"t": "small",
"v": "Data processed outside approved regions"
},
{
"t": "small",
"v": "No audit rights or limited visibility"
},
{
"t": "small",
"v": "SLA ambiguity across multiple services"
},
{
"t": "small",
"v": "Custom integrations create lock-in"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Actions to reduce risk"
},
{
"t": "small",
"v": "Consolidate demand and negotiate as one government"
},
{
"t": "small",
"v": "Demand transparency, telemetry and audit rights"
},
{
"t": "small",
"v": "Use open standards and ensure portability"
},
{
"t": "small",
"v": "Define clear data ownership and IP terms"
},
{
"t": "small",
"v": "Require model and data location assurance"
},
{
"t": "small",
"v": "Build modular architectures and exit plans"
},
{
"t": "small",
"v": "Share commercial intelligence across the community"
},
{
"t": "small",
"v": "Support smaller bodies with frameworks and guidance"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Build, buy, configure or share?"
},
{
"t": "p",
"v": "Match the approach to the need, risk, control, value and total cost. Decide wisely, start small, scale with confidence — and revisit the decision as needs, risk and technology evolve."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**1. Define the need**What problem are we solving?"
},
{
"t": "p",
"v": "**2. Assess context**Users, data, risk, regulation, criticality"
},
{
"t": "p",
"v": "**3. Evaluate options**Use the criteria to compare"
},
{
"t": "p",
"v": "**4. Decide & plan**Select approach, plan delivery"
},
{
"t": "p",
"v": "**5. Review & adapt**Measure outcomes, learn and adjust"
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Use platform / pre-built"
},
{
"t": "eyebrow",
"v": "Moderate complexity / risk"
}
],
"box": false
},
{
"t": "p",
"v": "Use existing platform features or pre-built agents with configuration."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "Best for"
},
{
"t": "small",
"v": "Common needs, moderate risk, fast value"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "You get"
},
{
"t": "small",
"v": "Faster delivery"
},
{
"t": "small",
"v": "Lower cost"
},
{
"t": "small",
"v": "Built-in updates & support"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "Considerations"
},
{
"t": "small",
"v": "Limited customisation"
},
{
"t": "small",
"v": "Platform roadmap dependency"
},
{
"t": "small",
"v": "Data & permission alignment"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 4,
"v": "Examples"
},
{
"t": "p",
"v": "Microsoft 365 Copilot, pre-built agents, Copilot Studio templates, AppSource solutions"
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Sample scoring matrix · 1 low to 5 high"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Strategic value",
"big": false
},
{
"t": "title",
"v": "Risk & assurance",
"big": false
},
{
"t": "title",
"v": "Control & flexibility",
"big": false
},
{
"t": "title",
"v": "Total cost (TCO)",
"big": false
},
{
"t": "title",
"v": "Time to value",
"big": false
},
{
"t": "title",
"v": "Reusability & scale",
"big": false
},
{
"t": "title",
"v": "AVG",
"big": false
},
{
"t": "title",
"v": "Personal configuration",
"big": false
},
{
"t": "num",
"v": "2"
},
{
"t": "num",
"v": "2"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "5"
},
{
"t": "num",
"v": "5"
},
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "3.2",
"big": false
},
{
"t": "title",
"v": "Use platform",
"big": false
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "3.3",
"big": false
},
{
"t": "title",
"v": "Configure / extend",
"big": false
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "3.5",
"big": false
},
{
"t": "title",
"v": "Build custom",
"big": false
},
{
"t": "num",
"v": "5"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "5"
},
{
"t": "num",
"v": "2"
},
{
"t": "num",
"v": "2"
},
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "3.5",
"big": false
},
{
"t": "title",
"v": "Share / buy",
"big": false
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "3"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "4"
},
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "4.0",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Very high (strong)"
},
{
"t": "small",
"v": "High (good)"
},
{
"t": "small",
"v": "Medium (acceptable)"
},
{
"t": "small",
"v": "Low (weak)"
},
{
"t": "small",
"v": "Very low (poor)"
}
]
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key questions to help decide"
},
{
"t": "small",
"v": "What is the impact if this goes wrong?"
},
{
"t": "small",
"v": "What data will it access or generate?"
},
{
"t": "small",
"v": "Do we need full control or can we share?"
},
{
"t": "small",
"v": "Is there an existing solution we can reuse?"
},
{
"t": "small",
"v": "What is the total cost over the lifecycle?"
},
{
"t": "small",
"v": "How quickly do we need value?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "The continuous review loop"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "1. Monitor usage, value and risk",
"big": false
},
{
"t": "title",
"v": "2. Evaluate outcomes and costs",
"big": false
},
{
"t": "title",
"v": "3. Learn and improve approach",
"big": false
},
{
"t": "title",
"v": "4. Adapt or reinvest",
"big": false
}
]
},
{
"t": "h",
"l": 3,
"v": "Common outcomes"
},
{
"t": "small",
"v": "Right solution for the need"
},
{
"t": "small",
"v": "Controlled risk and spend"
},
{
"t": "small",
"v": "Faster delivery and adoption"
},
{
"t": "small",
"v": "Reusable assets and knowledge"
},
{
"t": "small",
"v": "Better outcomes for citizens"
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "From community intelligence to better contracts"
},
{
"t": "p",
"v": "A continuous loop that turns real-world experience across government into stronger commercial engagement, better terms and improved public value. Open once, use many times."
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Inputs"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Departments & public bodies",
"big": false
},
{
"t": "title",
"v": "Devolved governments",
"big": false
},
{
"t": "title",
"v": "Local government & combined authorities",
"big": false
},
{
"t": "title",
"v": "Health & care",
"big": false
},
{
"t": "title",
"v": "Police, justice & emergency services",
"big": false
},
{
"t": "title",
"v": "Education & skills",
"big": false
},
{
"t": "title",
"v": "Regulators & public agencies",
"big": false
},
{
"t": "title",
"v": "Communities of practice & forums",
"big": false
},
{
"t": "title",
"v": "Audits, reviews & external reports",
"big": false
},
{
"t": "title",
"v": "Financial & FinOps data",
"big": false
}
]
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Community intelligence",
"big": false
},
{
"t": "small",
"v": "Live problems"
},
{
"t": "small",
"v": "Use cases & pilots"
},
{
"t": "small",
"v": "Costs & usage data"
},
{
"t": "small",
"v": "Risks & barriers"
},
{
"t": "small",
"v": "Lessons learned"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Consolidate & analyse",
"big": false
},
{
"t": "small",
"v": "Thematic analysis"
},
{
"t": "small",
"v": "Prioritise issues"
},
{
"t": "small",
"v": "Identify patterns"
},
{
"t": "small",
"v": "Evidence strength"
},
{
"t": "small",
"v": "Impact assessment"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Define requirements & questions",
"big": false
},
{
"t": "small",
"v": "Functional needs"
},
{
"t": "small",
"v": "Non-functional needs"
},
{
"t": "small",
"v": "Data & telemetry"
},
{
"t": "small",
"v": "Interoperability"
},
{
"t": "small",
"v": "Commercial terms"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Engage suppliers & market",
"big": false
},
{
"t": "small",
"v": "Share requirements"
},
{
"t": "small",
"v": "Test assumptions"
},
{
"t": "small",
"v": "Request evidence"
},
{
"t": "small",
"v": "Validate options"
},
{
"t": "small",
"v": "Explore innovation"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Refine terms & frameworks",
"big": false
},
{
"t": "small",
"v": "Contract clauses"
},
{
"t": "small",
"v": "Pricing & models"
},
{
"t": "small",
"v": "SLAs & commitments"
},
{
"t": "small",
"v": "Exit & portability"
},
{
"t": "small",
"v": "Risk & liability"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Implement & monitor",
"big": false
},
{
"t": "small",
"v": "Deploy solutions"
},
{
"t": "small",
"v": "Monitor performance"
},
{
"t": "small",
"v": "Track usage & cost"
},
{
"t": "small",
"v": "Manage risks"
},
{
"t": "small",
"v": "Capture issues"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "7"
},
{
"t": "title",
"v": "Feedback & share",
"big": false
},
{
"t": "small",
"v": "Share outcomes"
},
{
"t": "small",
"v": "Publish lessons"
},
{
"t": "small",
"v": "Update patterns"
},
{
"t": "small",
"v": "Support others"
},
{
"t": "small",
"v": "Improve guidance"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "8"
},
{
"t": "title",
"v": "Evolve & influence",
"big": false
},
{
"t": "small",
"v": "Refine priorities"
},
{
"t": "small",
"v": "Influence roadmaps"
},
{
"t": "small",
"v": "Shape market"
},
{
"t": "small",
"v": "Strengthen policy"
},
{
"t": "small",
"v": "Drive innovation"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Outputs"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Better contracts & terms",
"big": false
},
{
"t": "title",
"v": "Cost transparency & predictability",
"big": false
},
{
"t": "title",
"v": "Stronger controls & governance",
"big": false
},
{
"t": "title",
"v": "Interoperability & portability",
"big": false
},
{
"t": "title",
"v": "Improved supplier performance",
"big": false
},
{
"t": "title",
"v": "Shared patterns & reusable assets",
"big": false
},
{
"t": "title",
"v": "Evidence-based policy & standards",
"big": false
},
{
"t": "title",
"v": "Better outcomes for citizens",
"big": false
}
]
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Guiding principles"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Evidence led",
"big": false
},
{
"t": "small",
"v": "Real experience, not assertion"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Collaborative & inclusive",
"big": false
},
{
"t": "small",
"v": "Across departments, nations and sectors"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Fair & proportionate",
"big": false
},
{
"t": "small",
"v": "Terms matched to risk and scale"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Transparent & accountable",
"big": false
},
{
"t": "small",
"v": "Open about costs, terms and outcomes"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Open once, use many times",
"big": false
},
{
"t": "small",
"v": "Publish so others do not repeat the work"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Continuously improve",
"big": false
},
{
"t": "small",
"v": "The loop does not stop at signature"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers"
},
{
"t": "small",
"v": "Data & analytics"
},
{
"t": "small",
"v": "Skilled commercial & technical teams"
},
{
"t": "small",
"v": "Digital platforms & collaboration"
},
{
"t": "small",
"v": "Standards & guidance"
},
{
"t": "small",
"v": "Security & information assurance"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Outcomes"
},
{
"t": "small",
"v": "Faster, safer adoption"
},
{
"t": "small",
"v": "Better value for money"
},
{
"t": "small",
"v": "Reduced risk & duplication"
},
{
"t": "small",
"v": "Stronger innovation & competition"
},
{
"t": "small",
"v": "Continuous improvement"
},
{
"t": "small",
"v": "Trusted, sustainable public services"
}
]
}
],
"box": false
}
]
},
{
"id": "dependency",
"part": "fund-and-buy",
"title": "Dependency & exit",
"icon": "link",
"eyebrow": "Dependency, lock-in & exit",
"heading": "The ecosystem behind an AI-enabled public service",
"lede": "AI-enabled services rely on a chain of suppliers, platforms, models, data sources and infrastructure. Visibility of these dependencies is what makes risk management, resilience, portability and value possible.",
"body": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Public sector consumers"
},
{
"t": "title",
"v": "Central government departments",
"big": false
},
{
"t": "title",
"v": "Devolved administrations",
"big": false
},
{
"t": "title",
"v": "Local government",
"big": false
},
{
"t": "title",
"v": "Public bodies & arm's-length bodies",
"big": false
},
{
"t": "title",
"v": "Health & care",
"big": false
},
{
"t": "title",
"v": "Education",
"big": false
},
{
"t": "title",
"v": "Justice & policing",
"big": false
},
{
"t": "title",
"v": "Regulators",
"big": false
},
{
"t": "title",
"v": "Citizens & service users",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "AI-enabled public service / use case"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Knowledge assistant",
"big": false
},
{
"t": "title",
"v": "Case summary",
"big": false
},
{
"t": "title",
"v": "Policy drafting",
"big": false
},
{
"t": "title",
"v": "Meeting intelligence",
"big": false
},
{
"t": "title",
"v": "Citizen support",
"big": false
},
{
"t": "title",
"v": "Code assistant",
"big": false
},
{
"t": "title",
"v": "Workflow automation",
"big": false
},
{
"t": "title",
"v": "Analytics & insight",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Applications & platforms"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Microsoft 365 Copilot",
"big": false
},
{
"t": "title",
"v": "Copilot Studio / Agent Builder",
"big": false
},
{
"t": "title",
"v": "Teams, SharePoint, Outlook",
"big": false
},
{
"t": "title",
"v": "Power Platform (connectors, apps, automate)",
"big": false
},
{
"t": "title",
"v": "Foundry / Azure AI services",
"big": false
},
{
"t": "title",
"v": "Third-party applications",
"big": false
},
{
"t": "title",
"v": "Custom applications",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Models & AI capabilities"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Microsoft models",
"big": false
},
{
"t": "title",
"v": "OpenAI models",
"big": false
},
{
"t": "title",
"v": "Anthropic models",
"big": false
},
{
"t": "title",
"v": "Google models",
"big": false
},
{
"t": "title",
"v": "Open source models",
"big": false
},
{
"t": "title",
"v": "Specialist / domain models",
"big": false
},
{
"t": "title",
"v": "Future / emerging models",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Cloud & infrastructure"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Microsoft Azure (compute, storage, networking)",
"big": false
},
{
"t": "title",
"v": "Azure regions & availability zones",
"big": false
},
{
"t": "title",
"v": "GPU / compute capacity providers",
"big": false
},
{
"t": "title",
"v": "AI services (API management, safety, content filters)",
"big": false
},
{
"t": "title",
"v": "Identity & access (Entra ID)",
"big": false
},
{
"t": "title",
"v": "Monitoring & observability",
"big": false
},
{
"t": "title",
"v": "Backup & disaster recovery",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Data & information"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "SharePoint libraries",
"big": false
},
{
"t": "title",
"v": "Teams & collaboration content",
"big": false
},
{
"t": "title",
"v": "Line-of-business systems",
"big": false
},
{
"t": "title",
"v": "Open data & external sources",
"big": false
},
{
"t": "title",
"v": "Reference data & master data",
"big": false
},
{
"t": "title",
"v": "Vector stores & indexes",
"big": false
},
{
"t": "title",
"v": "Logs, telemetry & feedback",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Third parties & suppliers across the ecosystem"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Model providers",
"big": false
},
{
"t": "title",
"v": "Cloud provider",
"big": false
},
{
"t": "title",
"v": "Software vendors",
"big": false
},
{
"t": "title",
"v": "API / service providers",
"big": false
},
{
"t": "title",
"v": "Data providers",
"big": false
},
{
"t": "title",
"v": "Integration partners",
"big": false
},
{
"t": "title",
"v": "Support & operations",
"big": false
},
{
"t": "title",
"v": "Consulting & implementation",
"big": false
}
]
}
]
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key dependency categories"
},
{
"t": "p",
"v": "**Platform dependency**Reliance on specific platforms and ecosystems"
},
{
"t": "p",
"v": "**Model dependency**Reliance on model providers and availability"
},
{
"t": "p",
"v": "**Data dependency**Access to quality, timely and trusted data"
},
{
"t": "p",
"v": "**Infrastructure dependency**Compute, storage, network and region reliance"
},
{
"t": "p",
"v": "**Supplier & contractual dependency**Third-party services, SLAs, pricing and contractual terms"
},
{
"t": "p",
"v": "**Operational dependency**Skills, support, monitoring and continuity"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key questions to ask"
},
{
"t": "small",
"v": "What happens if a supplier or model is unavailable?"
},
{
"t": "small",
"v": "Can we switch models or platforms easily?"
},
{
"t": "small",
"v": "Where is our data stored and processed?"
},
{
"t": "small",
"v": "Do we have visibility of usage, cost and performance?"
},
{
"t": "small",
"v": "What contractual rights do we have — audit, data portability?"
},
{
"t": "small",
"v": "Are we building excessive lock-in or technical debt?"
},
{
"t": "small",
"v": "Do we have a clear exit and migration plan?"
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Managing dependencies"
},
{
"t": "small",
"v": "Map and document all dependencies"
},
{
"t": "small",
"v": "Assess risk and criticality"
},
{
"t": "small",
"v": "Ensure contractual protections"
},
{
"t": "small",
"v": "Design for portability"
},
{
"t": "small",
"v": "Monitor usage, cost and performance"
},
{
"t": "small",
"v": "Plan for resilience and exit"
},
{
"t": "small",
"v": "Review and adapt continuously"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Outcomes"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Better risk management",
"big": false
},
{
"t": "title",
"v": "Resilience & continuity",
"big": false
},
{
"t": "title",
"v": "Cost optimisation",
"big": false
},
{
"t": "title",
"v": "Informed supplier strategy",
"big": false
},
{
"t": "title",
"v": "Sustainable innovation",
"big": false
},
{
"t": "title",
"v": "Public value protection",
"big": false
}
]
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Design for choice today. Exit with confidence tomorrow."
},
{
"t": "p",
"v": "Avoid unintended dependency. Preserve portability across data, models, agents, integrations and knowledge, so you can adapt, switch or exit when you need to."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Portability**Keep your data, prompts, models and agents portable"
},
{
"t": "p",
"v": "**Transparency**Know what you use, where it is, and the terms"
},
{
"t": "p",
"v": "**Interoperability**Use open standards and avoid proprietary traps"
},
{
"t": "p",
"v": "**Plan ahead**Design exit and migration paths from day one"
},
{
"t": "p",
"v": "**Control costs**Understand true switching and transition costs"
},
{
"t": "p",
"v": "**Continuity**Maintain service and compliance through change"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Sources of lock-in"
},
{
"t": "small",
"v": "Proprietary platforms and ecosystems"
},
{
"t": "small",
"v": "Closed data formats and schemas"
},
{
"t": "small",
"v": "Proprietary connectors and APIs"
},
{
"t": "small",
"v": "Model dependency and exclusivity"
},
{
"t": "small",
"v": "Managed services and operations"
},
{
"t": "small",
"v": "Contractual restrictions and terms"
},
{
"t": "small",
"v": "Data residency and jurisdiction"
},
{
"t": "small",
"v": "Skills and tooling"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "The lock-in zone — where dependencies accumulate"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Technology dependencies"
},
{
"t": "small",
"v": "Proprietary features and roadmaps"
},
{
"t": "small",
"v": "Custom integrations"
},
{
"t": "small",
"v": "Closed standards"
},
{
"t": "small",
"v": "Single-tenant architectures"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Data & knowledge dependencies"
},
{
"t": "small",
"v": "Data locked in proprietary stores"
},
{
"t": "small",
"v": "Non-exportable metadata"
},
{
"t": "small",
"v": "Knowledge bases tied to platform"
},
{
"t": "small",
"v": "Prompt libraries not portable"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Model & AI dependencies"
},
{
"t": "small",
"v": "Model exclusivity clauses"
},
{
"t": "small",
"v": "No right to substitute"
},
{
"t": "small",
"v": "Fine-tuned models not exportable"
},
{
"t": "small",
"v": "Training data use restrictions"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Operational dependencies"
},
{
"t": "small",
"v": "Managed service entanglement"
},
{
"t": "small",
"v": "Monitoring and tool lock-in"
},
{
"t": "small",
"v": "Infrastructure coupling"
},
{
"t": "small",
"v": "Support and SLAs tied to stack"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Commercial dependencies"
},
{
"t": "small",
"v": "Long contracts with penalties"
},
{
"t": "small",
"v": "Limited termination rights"
},
{
"t": "small",
"v": "IP and data use restrictions"
},
{
"t": "small",
"v": "Unilateral price change rights"
}
]
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Organisational dependencies"
},
{
"t": "small",
"v": "Specialist skills tied to vendor"
},
{
"t": "small",
"v": "Processes built around tools"
},
{
"t": "small",
"v": "Change resistance"
},
{
"t": "small",
"v": "Limited market knowledge"
}
]
}
],
"box": false
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Exit & transition path — reduce risk, preserve value"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "1"
},
{
"t": "title",
"v": "Discover & inventory",
"big": false
},
{
"t": "small",
"v": "Map all assets, services and data"
},
{
"t": "small",
"v": "Identify critical dependencies"
},
{
"t": "small",
"v": "Assess complexity and effort"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "2"
},
{
"t": "title",
"v": "Define target state",
"big": false
},
{
"t": "small",
"v": "Choose target platform / model"
},
{
"t": "small",
"v": "Define data, security and compliance needs"
},
{
"t": "small",
"v": "Agree outcomes and constraints"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "3"
},
{
"t": "title",
"v": "Design migration strategy",
"big": false
},
{
"t": "small",
"v": "Prioritise workloads"
},
{
"t": "small",
"v": "Phased or parallel approach"
},
{
"t": "small",
"v": "Define data and application migration"
},
{
"t": "small",
"v": "Plan cutover and rollback"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "4"
},
{
"t": "title",
"v": "Prepare & extract",
"big": false
},
{
"t": "small",
"v": "Extract data, metadata and documents"
},
{
"t": "small",
"v": "Export prompts, agents and configurations"
},
{
"t": "small",
"v": "Validate completeness and quality"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "5"
},
{
"t": "title",
"v": "Build & test",
"big": false
},
{
"t": "small",
"v": "Rebuild integrations"
},
{
"t": "small",
"v": "Adapt to target platform / model"
},
{
"t": "small",
"v": "Test functionality, performance and security"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "6"
},
{
"t": "title",
"v": "Migrate & cutover",
"big": false
},
{
"t": "small",
"v": "Migrate data and configurations"
},
{
"t": "small",
"v": "Switch traffic and users"
},
{
"t": "small",
"v": "Monitor closely post-cutover"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "7"
},
{
"t": "title",
"v": "Optimise & decommission",
"big": false
},
{
"t": "small",
"v": "Optimise costs and performance"
},
{
"t": "small",
"v": "Retire old assets safely"
},
{
"t": "small",
"v": "Capture lessons and update runbooks"
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "h",
"l": 3,
"v": "Exit readiness check"
},
{
"t": "title",
"v": "Not assessed",
"big": false
},
{
"t": "p",
"v": "Rate how prepared you are to exit. The path above is the work; this is the readiness."
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What to preserve and make portable"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Data",
"big": false
},
{
"t": "title",
"v": "Metadata & schemas",
"big": false
},
{
"t": "title",
"v": "Documents & content",
"big": false
},
{
"t": "title",
"v": "Prompts & instructions",
"big": false
},
{
"t": "title",
"v": "Agents & workflows",
"big": false
},
{
"t": "title",
"v": "Configurations & settings",
"big": false
},
{
"t": "title",
"v": "Integrations & APIs",
"big": false
},
{
"t": "title",
"v": "Access & permissions",
"big": false
},
{
"t": "title",
"v": "Audit logs & evidence",
"big": false
},
{
"t": "title",
"v": "Knowledge & context",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "De-risking strategies"
},
{
"t": "small",
"v": "Use open standards and portable formats"
},
{
"t": "small",
"v": "Negotiate data and IP rights up front"
},
{
"t": "small",
"v": "Retain audit and telemetry access"
},
{
"t": "small",
"v": "Stay multi-cloud and multi-model ready"
},
{
"t": "small",
"v": "Modular architecture and loose coupling"
},
{
"t": "small",
"v": "Regular dependency reviews"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Red flags — high lock-in risk"
},
{
"t": "small",
"v": "No export or portability options"
},
{
"t": "small",
"v": "Data only accessible via proprietary APIs"
},
{
"t": "small",
"v": "High egress or extraction charges"
},
{
"t": "small",
"v": "Contract penalties for early exit or data retrieval"
},
{
"t": "small",
"v": "Complex custom integrations with no standards"
},
{
"t": "small",
"v": "Single model or provider dependency"
},
{
"t": "small",
"v": "Lack of roadmap visibility or support"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Exit cost components"
},
{
"t": "small",
"v": "Data extraction and transformation"
},
{
"t": "small",
"v": "Rebuild / re-engineering"
},
{
"t": "small",
"v": "Dual running (parallel operations)"
},
{
"t": "small",
"v": "Testing and assurance"
},
{
"t": "small",
"v": "Training and change"
},
{
"t": "small",
"v": "Contract exit and penalties"
},
{
"t": "small",
"v": "Downtime / disruption risk"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Exit success outcomes"
},
{
"t": "small",
"v": "Freedom to innovate and adapt"
},
{
"t": "small",
"v": "Lower total cost of ownership"
},
{
"t": "small",
"v": "Stronger resilience and continuity"
},
{
"t": "small",
"v": "Better negotiation power"
},
{
"t": "small",
"v": "Protected public data and trust"
},
{
"t": "small",
"v": "Sustainable, future-proof services"
}
]
}
],
"box": false
}
]
},
{
"id": "teams",
"part": "people-and-adoption",
"title": "Multidisciplinary teams",
"icon": "groups",
"eyebrow": "The team",
"heading": "Who needs to be in the room",
"lede": "The Service Manual's multidisciplinary team, extended for AI. Not every role is full-time, but every role is engaged from discovery, not summoned for sign-off. \"Critical at\" marks where each role's absence hurts most.",
"body": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Set the direction"
},
{
"t": "small",
"v": "Accountable and delivery leadership"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Senior Responsible Owner"
},
{
"t": "p",
"v": "Accountability, benefits, risk appetite. The name on the assurance forms."
},
{
"t": "eyebrow",
"v": "Critical at · Discover → Operate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Product manager"
},
{
"t": "small",
"v": "Problem framing, value, prioritisation of the backlog."
},
{
"t": "eyebrow",
"v": "Critical at · Discover · Design"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Delivery manager"
},
{
"t": "small",
"v": "Cadence, dependencies, unblocking the team."
},
{
"t": "eyebrow",
"v": "Critical at · Build · Deploy"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Programme / portfolio lead"
},
{
"t": "small",
"v": "Fit with the wider portfolio; stops duplicate AI pilots."
},
{
"t": "eyebrow",
"v": "Critical at · Discover · Deploy"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Build the capability"
},
{
"t": "small",
"v": "Data and AI engineering"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Data scientist"
},
{
"t": "small",
"v": "Modelling choices and honest evaluation design."
},
{
"t": "eyebrow",
"v": "Critical at · Design · Build"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI / ML engineer"
},
{
"t": "small",
"v": "Pipelines, integration, MLOps, guardrails in code."
},
{
"t": "eyebrow",
"v": "Critical at · Build · Operate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Data engineer"
},
{
"t": "small",
"v": "Data quality, provenance, pipelines that last."
},
{
"t": "eyebrow",
"v": "Critical at · Design · Build"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Technical architect"
},
{
"t": "small",
"v": "System design, interoperability, exit routes."
},
{
"t": "eyebrow",
"v": "Critical at · Design"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Assure and protect"
},
{
"t": "small",
"v": "Risk, rights and control"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cyber security specialist"
},
{
"t": "small",
"v": "Threat modelling and NCSC-aligned controls."
},
{
"t": "eyebrow",
"v": "Critical at · Design · Deploy"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "DPO / information governance"
},
{
"t": "small",
"v": "DPIA, lawful basis, records of processing."
},
{
"t": "eyebrow",
"v": "Critical at · Discover · Design"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Legal adviser"
},
{
"t": "small",
"v": "Public law, IP and copyright, contracts."
},
{
"t": "eyebrow",
"v": "Critical at · Design · Deploy"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Ethics adviser"
},
{
"t": "small",
"v": "Bias, participation, societal impact."
},
{
"t": "eyebrow",
"v": "Critical at · Discover → Operate"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Ground it in reality"
},
{
"t": "small",
"v": "Users, domain and money"
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "User researcher"
},
{
"t": "small",
"v": "Real user needs, and real harms, observed rather than assumed."
},
{
"t": "eyebrow",
"v": "Critical at · Discover · Operate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Service designer"
},
{
"t": "small",
"v": "The end-to-end journey, including the failure paths."
},
{
"t": "eyebrow",
"v": "Critical at · Design"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Domain expert / operational staff"
},
{
"t": "small",
"v": "Casework truth, edge cases, what the data actually means."
},
{
"t": "eyebrow",
"v": "Critical at · Discover → Operate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Commercial specialist"
},
{
"t": "small",
"v": "Procurement routes, vendor terms, ethical parity for third parties."
},
{
"t": "eyebrow",
"v": "Critical at · Discover · Deploy"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "**Skills pipeline.** Recruit and upskill through the Government Digital and Data profession's capability framework, Civil Service AI training and the cross-government AI community. Playbook principle 9 makes this the deploying organisation's responsibility, not the vendor's."
}
]
},
{
"id": "serviceteam",
"part": "people-and-adoption",
"title": "AI service team",
"icon": "hub",
"eyebrow": "Multidisciplinary AI service team",
"heading": "Twelve disciplines around one AI service",
"lede": "Collaborating to design, build, govern and operate trustworthy AI services that deliver public value — through design and plan, build and enable, deploy and operate, measure and improve. The service at the centre is defined by user needs, public value, trust, safety and outcomes.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Product management",
"big": false
},
{
"t": "small",
"v": "Strategy & roadmap"
},
{
"t": "small",
"v": "Prioritisation & value"
},
{
"t": "small",
"v": "Stakeholder alignment"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Service design",
"big": false
},
{
"t": "small",
"v": "End-to-end journeys"
},
{
"t": "small",
"v": "Inclusive & accessible"
},
{
"t": "small",
"v": "Experience design"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Solution architecture",
"big": false
},
{
"t": "small",
"v": "Architecture & patterns"
},
{
"t": "small",
"v": "Integration & interoperability"
},
{
"t": "small",
"v": "Scalability & resilience"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Data & AI",
"big": false
},
{
"t": "small",
"v": "Data strategy & governance"
},
{
"t": "small",
"v": "Data quality & lineage"
},
{
"t": "small",
"v": "Model design & evaluation"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Security",
"big": false
},
{
"t": "small",
"v": "Secure by design"
},
{
"t": "small",
"v": "Threat modelling"
},
{
"t": "small",
"v": "Security assurance"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Information governance",
"big": false
},
{
"t": "small",
"v": "Information management"
},
{
"t": "small",
"v": "Records & retention"
},
{
"t": "small",
"v": "Data classification"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Legal & policy",
"big": false
},
{
"t": "small",
"v": "Lawful & ethical use"
},
{
"t": "small",
"v": "Policy compliance"
},
{
"t": "small",
"v": "Contracts & liabilities"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Commercial",
"big": false
},
{
"t": "small",
"v": "Procurement strategy"
},
{
"t": "small",
"v": "Supplier management"
},
{
"t": "small",
"v": "Value for money"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Operations",
"big": false
},
{
"t": "small",
"v": "Run & support"
},
{
"t": "small",
"v": "Monitoring & reliability"
},
{
"t": "small",
"v": "Incident management"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Learning & development",
"big": false
},
{
"t": "small",
"v": "Capability & skills"
},
{
"t": "small",
"v": "Training & guidance"
},
{
"t": "small",
"v": "Adoption & culture"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Accessibility & inclusion",
"big": false
},
{
"t": "small",
"v": "WCAG & inclusive design"
},
{
"t": "small",
"v": "Equality impact"
},
{
"t": "small",
"v": "Usability for all"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "User research",
"big": false
},
{
"t": "small",
"v": "Understand users"
},
{
"t": "small",
"v": "Research & insights"
},
{
"t": "small",
"v": "Usability & validation"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Team principles"
},
{
"t": "p",
"v": "**User-centred**Design with and for users at every stage."
},
{
"t": "p",
"v": "**Inclusive**Accessible, equitable and easy to use."
},
{
"t": "p",
"v": "**Trust & safety**Build security, privacy and ethics in from the start."
},
{
"t": "p",
"v": "**Collaborative**Work together, share ideas and challenge constructively."
},
{
"t": "p",
"v": "**Agile & iterative**Learn quickly, adapt and improve continuously."
},
{
"t": "p",
"v": "**Outcome focused**Measure what matters and deliver public value."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Key inputs"
},
{
"t": "small",
"v": "User needs & insights"
},
{
"t": "small",
"v": "Policies, strategy & standards"
},
{
"t": "small",
"v": "Data & knowledge assets"
},
{
"t": "small",
"v": "Technology & platforms"
},
{
"t": "small",
"v": "Stakeholders & partners"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Key outputs"
},
{
"t": "small",
"v": "Trusted AI services that work for users"
},
{
"t": "small",
"v": "Better outcomes & public value"
},
{
"t": "small",
"v": "Evidence & insights for decisions"
},
{
"t": "small",
"v": "Assured, compliant & well governed"
},
{
"t": "small",
"v": "Skilled teams & continuous learning"
}
]
}
],
"box": false
}
]
},
{
"id": "roles",
"part": "people-and-adoption",
"title": "Roles & pathways",
"icon": "school",
"eyebrow": "Roles & pathways",
"heading": "Learn. Apply. Progress.",
"lede": "Role-specific learning journeys to build skills, confidence and impact at every stage of adoption. Pick the role you are supporting; the pathway shows what capability looks like at each of four levels.",
"body": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "YOUR JOURNEY"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "checklist"
},
{
"t": "title",
"v": "1. Assess current skills",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "signpost"
},
{
"t": "title",
"v": "2. Choose your pathway",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "title",
"v": "3. Learn & practise with support",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "settings"
},
{
"t": "title",
"v": "4. Apply in real work",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "insights"
},
{
"t": "title",
"v": "5. Measure impact & improve",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lightbulb"
},
{
"t": "h",
"l": 3,
"v": "1 · Aware"
},
{
"t": "small",
"v": "Understand AI basics and opportunities"
},
{
"t": "title",
"v": "AI fundamentals",
"big": false
},
{
"t": "title",
"v": "Opportunities & risks",
"big": false
},
{
"t": "title",
"v": "Strategic context",
"big": false
},
{
"t": "title",
"v": "Responsible AI",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "settings"
},
{
"t": "h",
"l": 3,
"v": "2 · Practitioner"
},
{
"t": "small",
"v": "Apply AI tools and techniques with guidance"
},
{
"t": "title",
"v": "Use cases & value",
"big": false
},
{
"t": "title",
"v": "Prioritisation",
"big": false
},
{
"t": "title",
"v": "Governance basics",
"big": false
},
{
"t": "title",
"v": "Supporting teams",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "h",
"l": 3,
"v": "3 · Advanced"
},
{
"t": "small",
"v": "Design, adapt and lead AI solutions"
},
{
"t": "title",
"v": "AI strategy alignment",
"big": false
},
{
"t": "title",
"v": "Portfolio oversight",
"big": false
},
{
"t": "title",
"v": "Investment decisions",
"big": false
},
{
"t": "title",
"v": "Risk appetite",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "star"
},
{
"t": "h",
"l": 3,
"v": "4 · Expert"
},
{
"t": "small",
"v": "Shape strategy, standards and innovation"
},
{
"t": "title",
"v": "AI vision & innovation",
"big": false
},
{
"t": "title",
"v": "Ecosystem leadership",
"big": false
},
{
"t": "title",
"v": "Policy influence",
"big": false
},
{
"t": "title",
"v": "Ethical leadership",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "What supports progress at every level"
},
{
"t": "icon",
"v": "menu_book"
},
{
"t": "p",
"v": "**Learning & development**Role-based curricula, microlearning, labs, communities of practice"
},
{
"t": "icon",
"v": "desktop_windows"
},
{
"t": "p",
"v": "**Tools & platforms**Approved tools, secure sandboxes, reusable components"
},
{
"t": "icon",
"v": "policy"
},
{
"t": "p",
"v": "**Guidance & standards**Policies, playbooks, architectures, guardrails and templates"
},
{
"t": "icon",
"v": "insights"
},
{
"t": "p",
"v": "**Measuring progress**Skills assessment, certification, impact and value metrics"
},
{
"t": "icon",
"v": "workspace_premium"
},
{
"t": "p",
"v": "**Recognition & career**Career pathways, communities, badges and incentives"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Principles behind the pathways"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Role relevant",
"big": false
},
{
"t": "small",
"v": "Right skills for the job"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Practical",
"big": false
},
{
"t": "small",
"v": "Learn by doing in real work"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Progressive",
"big": false
},
{
"t": "small",
"v": "Build from foundations to mastery"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Trusted",
"big": false
},
{
"t": "small",
"v": "Safe, secure and responsible AI use"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Measurable",
"big": false
},
{
"t": "small",
"v": "Evidence of competence and impact"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Inclusive",
"big": false
},
{
"t": "small",
"v": "Accessible and supportive for all"
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "From task execution to higher-value work"
},
{
"t": "p",
"v": "AI augments people, it does not replace them. Roles evolve to focus on judgement, creativity, oversight and continuous improvement."
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "ROLE",
"big": false
},
{
"t": "title",
"v": "TRADITIONAL FOCUS",
"big": false
},
{
"t": "title",
"v": "AI ENABLES",
"big": false
},
{
"t": "title",
"v": "AI-ENABLED FUTURE FOCUS",
"big": false
},
{
"t": "title",
"v": "IMPACT",
"big": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Policy adviser**"
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Research and information gathering"
},
{
"t": "small",
"v": "Drafting options and text"
},
{
"t": "small",
"v": "Compiling background papers"
}
]
},
{
"t": "p",
"v": "AI summarises evidence, suggests options, drafts content, finds patterns"
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Strategic analysis and judgement",
"big": false
},
{
"t": "title",
"v": "Evaluating trade-offs and impacts",
"big": false
},
{
"t": "title",
"v": "Engaging stakeholders and deciding",
"big": false
}
]
},
{
"t": "p",
"v": "**Better policy quality and impact**"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Contact centre adviser**"
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Answering routine queries"
},
{
"t": "small",
"v": "Searching knowledge bases"
},
{
"t": "small",
"v": "Logging and updating cases"
}
]
},
{
"t": "p",
"v": "AI handles routine queries, retrieves answers, suggests next best action"
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Complex case resolution",
"big": false
},
{
"t": "title",
"v": "Empathy and relationship building",
"big": false
},
{
"t": "title",
"v": "Handling exceptions and escalation",
"big": false
}
]
},
{
"t": "p",
"v": "**Improved citizen experience**"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Software engineer**"
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Writing and refactoring code"
},
{
"t": "small",
"v": "Testing and fixing defects"
},
{
"t": "small",
"v": "Reviewing changes"
}
]
},
{
"t": "p",
"v": "AI generates code, finds bugs, recommends tests, documents changes"
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Architecture and design decisions",
"big": false
},
{
"t": "title",
"v": "Code review and security assurance",
"big": false
},
{
"t": "title",
"v": "Solving novel and complex problems",
"big": false
}
]
},
{
"t": "p",
"v": "**Faster delivery, higher quality**"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Data analyst**"
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Data extraction and preparation"
},
{
"t": "small",
"v": "Running standard reports"
},
{
"t": "small",
"v": "Creating charts and dashboards"
}
]
},
{
"t": "p",
"v": "AI prepares data, generates insights, builds visuals, detects anomalies"
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Deeper analysis and insight",
"big": false
},
{
"t": "title",
"v": "Explaining drivers and implications",
"big": false
},
{
"t": "title",
"v": "Advising on data strategy",
"big": false
}
]
},
{
"t": "p",
"v": "**Stronger insight and decisions**"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**HR advisor**"
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Processing requests"
},
{
"t": "small",
"v": "Answering policy questions"
},
{
"t": "small",
"v": "Updating records"
}
]
},
{
"t": "p",
"v": "AI answers FAQs, guides processes, drafts documents, checks compliance"
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "People strategy and coaching",
"big": false
},
{
"t": "title",
"v": "Complex case management",
"big": false
},
{
"t": "title",
"v": "Culture and inclusion leadership",
"big": false
}
]
},
{
"t": "p",
"v": "**Better employee experience**"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Risk manager**"
},
{
"t": "group",
"kids": [
{
"t": "small",
"v": "Collecting risk data"
},
{
"t": "small",
"v": "Maintaining risk registers"
},
{
"t": "small",
"v": "Producing standard reports"
}
]
},
{
"t": "p",
"v": "AI identifies risks, analyses scenarios, monitors controls, suggests mitigations"
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Interpreting risk in context",
"big": false
},
{
"t": "title",
"v": "Prioritising actions and trade-offs",
"big": false
},
{
"t": "title",
"v": "Overseeing governance and assurance",
"big": false
}
]
},
{
"t": "p",
"v": "**Stronger risk management**"
}
],
"box": true
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "The shift"
},
{
"t": "title",
"v": "From doing tasks…",
"big": false
},
{
"t": "title",
"v": "…to orchestrating with AI",
"big": false
},
{
"t": "title",
"v": "…to focusing on judgement, creativity and impact",
"big": false
},
{
"t": "title",
"v": "Delivering better outcomes for citizens and society",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What people need"
},
{
"t": "small",
"v": "AI literacy and critical thinking"
},
{
"t": "small",
"v": "Change mindset and adaptability"
},
{
"t": "small",
"v": "Collaboration and communication"
},
{
"t": "small",
"v": "Ethics and responsible use"
},
{
"t": "p",
"v": "Success depends on people, process redesign, culture, leadership and trust — supported by responsible AI and strong governance."
}
]
}
],
"box": false
}
]
},
{
"id": "academy",
"part": "people-and-adoption",
"title": "Capability academy",
"icon": "cast_for_education",
"eyebrow": "AI workforce capability academy",
"heading": "Building the skills and confidence to design, develop, govern and operate trusted AI services",
"lede": "Five learning pathways across four stages. Build confidence. Apply skills. Drive impact. Lead change.",
"body": [
{
"t": "row",
"items": [
{
"t": "p",
"v": "**Foundation**Understand the basics"
},
{
"t": "p",
"v": "**Build**Develop practical skills"
},
{
"t": "p",
"v": "**Advance**Apply and lead"
},
{
"t": "p",
"v": "**Lead**Expertise and influence"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Leaders & decision makers"
},
{
"t": "small",
"v": "Set direction and provide oversight"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Foundation** **AI essentials for leaders** Opportunities, limitations · Risks and public value · Responsible AI principles"
},
{
"t": "p",
"v": "**Build** **AI governance & accountability** Oversight responsibilities · Risk appetite and decision rights · Human control and assurance"
},
{
"t": "p",
"v": "**Advance** **Strategic AI leadership** Portfolio management · Measuring outcomes and impact · Ethics, bias and inclusion"
},
{
"t": "p",
"v": "**Lead** **Shaping the AI agenda** Leading culture change · Stakeholder engagement · Policy influence and advocacy"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Product & service teams"
},
{
"t": "small",
"v": "Design and deliver AI-enabled services"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Foundation** **AI for product & service teams** User needs and problem framing · AI use cases and value · Ethical and inclusive design"
},
{
"t": "p",
"v": "**Build** **Designing AI-enabled services** Journey mapping with AI · Prototyping and evaluation · Managing suppliers"
},
{
"t": "p",
"v": "**Advance** **AI product management** Prioritisation and roadmapping · Data-informed outcomes · Scaling AI responsibly"
},
{
"t": "p",
"v": "**Lead** **Leading AI transformation** Cross-team leadership · Change management · Driving adoption and value"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Technical & data teams"
},
{
"t": "small",
"v": "Build and run secure, reliable AI agents"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Foundation** **AI & data fundamentals** Data types and quality · Intro to ML, LLMs and agents · Privacy and security basics"
},
{
"t": "p",
"v": "**Build** **Building AI agents** Prompt engineering · RAG and tool integration · Testing and evaluation"
},
{
"t": "p",
"v": "**Advance** **Advanced engineering** Agent orchestration · Model evaluation & guardrails · Red teaming and robustness"
},
{
"t": "p",
"v": "**Lead** **Technical excellence** Architecture leadership · Innovation and emerging tech · Open standards and reuse"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Governance & assurance teams"
},
{
"t": "small",
"v": "Assure, audit and manage risk"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Foundation** **AI governance basics** Policies, standards and roles · Risk types and controls · Transparency and ATRS"
},
{
"t": "p",
"v": "**Build** **Assessing and controlling risk** DPIA for AI systems · Bias, fairness and inclusion · Security threat modelling"
},
{
"t": "p",
"v": "**Advance** **AI assurance & audit** Assurance frameworks · Monitoring and evidence · Incident and issue management"
},
{
"t": "p",
"v": "**Lead** **Assurance leadership** Independent challenge · Audit strategy and reporting · Regulatory horizon scanning"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Operational & support teams"
},
{
"t": "small",
"v": "Operate, monitor and support users"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Foundation** **Working with AI tools** Everyday AI use · Human oversight basics · Data handling and security"
},
{
"t": "p",
"v": "**Build** **Operating AI services** Monitoring and logging · Incident reporting · User support and feedback"
},
{
"t": "p",
"v": "**Advance** **Optimising operations** Performance analysis · Continuous improvement · Knowledge sharing"
},
{
"t": "p",
"v": "**Lead** **Operational excellence** Service reliability leadership · Automation and productivity · Building a learning culture"
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Core capabilities for all roles"
},
{
"t": "p",
"v": "**Responsible AI**Ethics, fairness, transparency"
},
{
"t": "p",
"v": "**Security & privacy**Secure by design, data protection"
},
{
"t": "p",
"v": "**Human oversight**Accountability, challenge, escalation"
},
{
"t": "p",
"v": "**Data literacy**Quality, lineage, classification"
},
{
"t": "p",
"v": "**Accessibility & inclusion**Design for everyone, equitable outcomes"
},
{
"t": "p",
"v": "**Continuous learning**Iterate, adapt and improve"
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Learning methods"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Instructor-led workshops",
"big": false
},
{
"t": "title",
"v": "E-learning modules",
"big": false
},
{
"t": "title",
"v": "Hands-on labs & sandboxes",
"big": false
},
{
"t": "title",
"v": "Communities of practice",
"big": false
},
{
"t": "title",
"v": "Certification & badging",
"big": false
},
{
"t": "title",
"v": "Playbooks & job aids",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Capability progression"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Aware",
"big": false
},
{
"t": "title",
"v": "Practitioner",
"big": false
},
{
"t": "title",
"v": "Expert",
"big": false
},
{
"t": "title",
"v": "Leader",
"big": false
}
]
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "capability",
"part": "people-and-adoption",
"title": "Capability & maturity",
"icon": "stairs",
"eyebrow": "Capability & maturity · illustrative frameworks",
"heading": "Maturity is a journey, not a destination",
"lede": "Seven levels, from ad-hoc use to adaptive capability. Choose a level to see what it looks like across the dimensions and the twelve capability domains — then record the evidence you could actually show.",
"body": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Level 3 · Governed"
},
{
"t": "title",
"v": "Next: 3 → 4 — Integrate and embed",
"big": false
},
{
"t": "p",
"v": "Defined governance and controls. Repeatable patterns, registered use cases and measured pilots."
},
{
"t": "table",
"rows": [
[
{
"h": false,
"v": "**Leadership** **Accountable owners**"
},
{
"h": false,
"v": "**Governance** **Controls, risk tiers, assurance gates**"
},
{
"h": false,
"v": "**People & skills** **Role-based skills**"
}
],
[
{
"h": false,
"v": "**Data & knowledge** **Catalogued, governed access**"
},
{
"h": false,
"v": "**Technology** **Reference architecture, model choice**"
},
{
"h": false,
"v": "**Value & outcomes** **Measured value from pilots**"
}
]
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Evidence over opinion"
},
{
"t": "title",
"v": "Record the evidence you can actually show for each domain.",
"big": false
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Strong",
"big": false
},
{
"t": "small",
"v": "Independently validated"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Moderate",
"big": false
},
{
"t": "small",
"v": "Documented internal evidence"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Emerging",
"big": false
},
{
"t": "small",
"v": "In pilot or partially proven"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Weak",
"big": false
},
{
"t": "small",
"v": "Anecdotal or not evidenced"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Not assessed",
"big": false
},
{
"t": "small",
"v": "No evidence available"
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "explore"
},
{
"t": "h",
"l": 3,
"v": "Strategy & leadership"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "AI strategy aligned to business objectives and risk appetite"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "groups"
},
{
"t": "h",
"l": 3,
"v": "People & skills"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Role-based skills framework and capability plans"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "target"
},
{
"t": "h",
"l": 3,
"v": "Use cases & value"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Use case portfolio managed with benefits tracking"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "account_tree"
},
{
"t": "h",
"l": 3,
"v": "Technology & architecture"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Target architecture, standards and integration patterns"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "database"
},
{
"t": "h",
"l": 3,
"v": "Data & information"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Governed data assets, metadata and lineage"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "shield"
},
{
"t": "h",
"l": 3,
"v": "Security & privacy"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Security-by-design; identity, access and monitoring in place"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "balance"
},
{
"t": "h",
"l": 3,
"v": "Responsible AI & ethics"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Responsible AI framework and assurance process"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "handshake"
},
{
"t": "h",
"l": 3,
"v": "Commercial & suppliers"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Managed suppliers, contracts and risk assessments"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "payments"
},
{
"t": "h",
"l": 3,
"v": "Finance & FinOps"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Cost visibility, allocation and showback"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "support_agent"
},
{
"t": "h",
"l": 3,
"v": "Operations & service management"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Defined operating model and SLAs"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "insights"
},
{
"t": "h",
"l": 3,
"v": "Measurement & outcomes"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Benefits tracking; outcomes measured"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "icon",
"v": "forum"
},
{
"t": "h",
"l": 3,
"v": "Community & knowledge"
},
{
"t": "title",
"v": "Not recorded",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Communities of practice and curated knowledge"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How to use this"
},
{
"t": "small",
"v": "**1** Assess current evidence against each domain"
},
{
"t": "small",
"v": "**2** Identify priority gaps — emerging and weak ratings"
},
{
"t": "small",
"v": "**3** Take actions to build evidence and capability"
},
{
"t": "small",
"v": "**4** Review and update regularly"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence that counts"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Policies & strategies",
"big": false
},
{
"t": "title",
"v": "Architectures & standards",
"big": false
},
{
"t": "title",
"v": "Training records",
"big": false
},
{
"t": "title",
"v": "Use case business cases",
"big": false
},
{
"t": "title",
"v": "KPIs & dashboards",
"big": false
},
{
"t": "title",
"v": "Audit & assurance reports",
"big": false
},
{
"t": "title",
"v": "Lessons learned & patterns",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What good looks like"
},
{
"t": "small",
"v": "Clear strategy and governance"
},
{
"t": "small",
"v": "Skilled and empowered people"
},
{
"t": "small",
"v": "Trusted data and knowledge"
},
{
"t": "small",
"v": "Secure and responsible by design"
},
{
"t": "small",
"v": "Smart commercial and supplier management"
},
{
"t": "small",
"v": "Efficient and sustainable spend"
},
{
"t": "small",
"v": "Measurable outcomes and public value"
},
{
"t": "small",
"v": "Strong communities and shared learning"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Where the capability gaps are, by role"
},
{
"t": "p",
"v": "Current against required capability, scored 5 (critical gap) to 1 (strong, sustain). Illustrative sector-level assessment — replace with your own to prioritise investment. Filter to one role group to read a single column."
},
{
"t": "card",
"kids": [
{
"t": "table",
"rows": [
[
{
"h": false,
"v": "CAPABILITY DOMAIN"
},
{
"h": false,
"v": "Executives & leaders"
},
{
"h": false,
"v": "Business & service owners"
},
{
"h": false,
"v": "Product & delivery managers"
},
{
"h": false,
"v": "Data & info professionals"
},
{
"h": false,
"v": "Engineers & technical teams"
},
{
"h": false,
"v": "Security & risk professionals"
},
{
"h": false,
"v": "Commercial & procurement"
},
{
"h": false,
"v": "Finance & FinOps"
},
{
"h": false,
"v": "Frontline staff & practitioners"
}
],
[
{
"h": false,
"v": "**Strategy & leadership** Vision, governance, value"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**People & change** Adoption, culture, engagement"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**AI & data literacy** Understanding AI, data and information"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "5"
}
],
[
{
"h": false,
"v": "**Technology & engineering** Tools, platforms, integration"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**Data & information** Quality, access, governance"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**Security, privacy & risk** Threats, controls, compliance"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**Responsible AI & ethics** Fairness, transparency, human oversight"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**Commercial & procurement** Contracts, suppliers, value"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "4"
}
],
[
{
"h": false,
"v": "**Finance & FinOps** Cost management, allocation, value"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "2"
},
{
"h": false,
"v": "3"
}
],
[
{
"h": false,
"v": "**Operations & service management** Monitoring, support, continuity"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "4"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
},
{
"h": false,
"v": "3"
}
],
[
{
"h": false,
"v": "**Measurement & outcomes** KPIs, impact, benefits — not scored on the source board"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
},
{
"h": false,
"v": "—"
}
]
]
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Critical gap",
"big": false
},
{
"t": "small",
"v": "High priority"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Major gap",
"big": false
},
{
"t": "small",
"v": "Act soon"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Moderate gap",
"big": false
},
{
"t": "small",
"v": "Plan to address"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Minor gap",
"big": false
},
{
"t": "small",
"v": "Monitor"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Strong / no gap",
"big": false
},
{
"t": "small",
"v": "Sustain"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Top priority gaps"
},
{
"t": "small",
"v": "AI & data literacy for frontline staff"
},
{
"t": "p",
"v": "Data & information for executives, business owners and product managers"
},
{
"t": "small",
"v": "People & change for technical and security roles"
},
{
"t": "small",
"v": "Commercial & procurement for product managers and finance"
},
{
"t": "small",
"v": "Finance & FinOps for product managers"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Key insights"
},
{
"t": "small",
"v": "Frontline staff need stronger AI literacy and confidence"
},
{
"t": "small",
"v": "Data and information capabilities are a cross-role enabler"
},
{
"t": "small",
"v": "Product managers and finance need deeper commercial and FinOps skills"
},
{
"t": "p",
"v": "Technical teams are relatively strong but need continued investment in security by design"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "How to use the heatmap"
},
{
"t": "small",
"v": "**1** Assess current capability levels by role"
},
{
"t": "small",
"v": "**2** Compare to required capability levels"
},
{
"t": "small",
"v": "**3** Prioritise high-impact gaps"
},
{
"t": "small",
"v": "**4** Invest in learning, support and resources"
},
{
"t": "small",
"v": "**5** Measure progress and outcomes"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Actions to close gaps"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Role-based learning pathways",
"big": false
},
{
"t": "title",
"v": "Communities of practice and peer learning",
"big": false
},
{
"t": "title",
"v": "Targeted coaching and enablement",
"big": false
},
{
"t": "title",
"v": "Tools, templates and reusable assets",
"big": false
},
{
"t": "title",
"v": "Time, budget and leadership support",
"big": false
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "inclusion",
"part": "people-and-adoption",
"title": "Inclusive adoption",
"icon": "diversity_3",
"eyebrow": "Inclusive & accessible adoption",
"heading": "Different people. Different needs. Shared success.",
"lede": "AI tools should empower everyone by offering choice, clarity and support. The spectrum below runs from empowered and confident to apprehensive — and inclusion is not one size for all: listening, choice and support help everyone succeed with AI.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enthusiastic & confident"
},
{
"t": "small",
"v": "Sees AI as a productivity booster and enjoys exploring new tools."
},
{
"t": "h",
"l": 4,
"v": "May typically"
},
{
"t": "small",
"v": "Experiment and customise"
},
{
"t": "small",
"v": "Share tips and best practice"
},
{
"t": "small",
"v": "Integrate AI into daily work"
},
{
"t": "h",
"l": 4,
"v": "Key needs"
},
{
"t": "title",
"v": "Advanced features",
"big": false
},
{
"t": "title",
"v": "Experimentation space",
"big": false
},
{
"t": "title",
"v": "Recognition & influence",
"big": false
},
{
"t": "title",
"v": "Ways to contribute back",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cautious but willing"
},
{
"t": "small",
"v": "Interested and willing to try, but wants reassurance and guidance."
},
{
"t": "h",
"l": 4,
"v": "May typically"
},
{
"t": "small",
"v": "Use AI for simple, low-risk tasks"
},
{
"t": "small",
"v": "Seek examples and best practice"
},
{
"t": "small",
"v": "Prefers structured learning"
},
{
"t": "h",
"l": 4,
"v": "Key needs"
},
{
"t": "title",
"v": "Clear guidance & examples",
"big": false
},
{
"t": "title",
"v": "Safe practice environments",
"big": false
},
{
"t": "title",
"v": "Step-by-step learning",
"big": false
},
{
"t": "title",
"v": "Trusted peer support",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Needs support & clarity"
},
{
"t": "small",
"v": "Wants to benefit from AI but needs help understanding how and when."
},
{
"t": "h",
"l": 4,
"v": "May typically"
},
{
"t": "small",
"v": "Uses guided features"
},
{
"t": "small",
"v": "Needs help in context"
},
{
"t": "small",
"v": "Prefers practical, relevant training"
},
{
"t": "h",
"l": 4,
"v": "Key needs"
},
{
"t": "title",
"v": "Plain language & relevance",
"big": false
},
{
"t": "title",
"v": "Role-based training",
"big": false
},
{
"t": "title",
"v": "Hand-holding & coaching",
"big": false
},
{
"t": "title",
"v": "Time to learn & practise",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Anxious & overwhelmed"
},
{
"t": "small",
"v": "Feels unsure or overwhelmed by change, complexity or risk."
},
{
"t": "h",
"l": 4,
"v": "May typically"
},
{
"t": "small",
"v": "Avoids or minimises use"
},
{
"t": "small",
"v": "Worried about errors or job impact"
},
{
"t": "small",
"v": "Prefers human interaction"
},
{
"t": "h",
"l": 4,
"v": "Key needs"
},
{
"t": "title",
"v": "Reassurance & transparency",
"big": false
},
{
"t": "title",
"v": "Simple, limited choices",
"big": false
},
{
"t": "title",
"v": "Human support always available",
"big": false
},
{
"t": "title",
"v": "Safe, low-pressure options",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Resistant / disengaged"
},
{
"t": "small",
"v": "Does not see value or may distrust AI or the organisation's intent."
},
{
"t": "h",
"l": 4,
"v": "May typically"
},
{
"t": "small",
"v": "Declines to use"
},
{
"t": "small",
"v": "Questions fairness or reliability"
},
{
"t": "small",
"v": "Prefers current ways of working"
},
{
"t": "h",
"l": 4,
"v": "Key needs"
},
{
"t": "title",
"v": "Listen & understand concerns",
"big": false
},
{
"t": "title",
"v": "Involve in decisions",
"big": false
},
{
"t": "title",
"v": "Address trust & fairness",
"big": false
},
{
"t": "title",
"v": "Show tangible benefits",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Design for inclusion"
},
{
"t": "p",
"v": "**Choice & control**Let users choose how AI helps, when and in what way"
},
{
"t": "p",
"v": "**Clear & simple**Use plain language, short explanations and examples"
},
{
"t": "p",
"v": "**Flexible interaction**Support different workflows, speeds and preferences"
},
{
"t": "p",
"v": "**Neuroinclusive design**Reduce cognitive load, provide structure, minimise distractions"
},
{
"t": "p",
"v": "**Accessible by default**WCAG compliant, multimodal, and assistive-tech friendly"
},
{
"t": "p",
"v": "**Safe & trusted**Protect data, explain limits, and make AI behaviour predictable"
},
{
"t": "p",
"v": "**Human support**Easy access to help, feedback channels and champions"
}
]
},
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers & supports"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Leadership sponsorship",
"big": false
},
{
"t": "title",
"v": "Digital & AI champions",
"big": false
},
{
"t": "title",
"v": "Inclusive communication",
"big": false
},
{
"t": "title",
"v": "Role-based learning paths",
"big": false
},
{
"t": "title",
"v": "Protected learning time",
"big": false
},
{
"t": "title",
"v": "Local support networks",
"big": false
},
{
"t": "title",
"v": "Feedback & continuous listening",
"big": false
},
{
"t": "title",
"v": "Measure & adapt for all users",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Remember"
},
{
"t": "p",
"v": "Inclusion is not one size fits all. Listening, choice and support help everyone succeed with AI — and inclusive adoption empowers people, strengthens trust and delivers better public services."
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The cross-government community constellation"
},
{
"t": "p",
"v": "Working together across communities to share knowledge, avoid duplication and build trusted, reusable AI and Copilot capabilities. The weekly cross-government call is the open forum at the centre."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Practice and engineering"
},
{
"t": "small",
"v": "Hands-on development, experimentation and engineering excellence."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "AI Engineering Lab",
"big": false
},
{
"t": "title",
"v": "Data Science & AI Community",
"big": false
},
{
"t": "title",
"v": "Low Code Community",
"big": false
},
{
"t": "title",
"v": "Microsoft 365 Technical Community",
"big": false
},
{
"t": "title",
"v": "Prototyping and Sandboxes",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Adoption and digital workplace"
},
{
"t": "small",
"v": "User enablement, organisational change and the digital workplace."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Digital Workplace Community",
"big": false
},
{
"t": "title",
"v": "Change & Adoption Network",
"big": false
},
{
"t": "title",
"v": "AI Champions Network",
"big": false
},
{
"t": "title",
"v": "Learning & Capability Network",
"big": false
},
{
"t": "title",
"v": "Comms & Engagement Network",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Data management communities"
},
{
"t": "small",
"v": "Data governance, catalogues and information management excellence."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Data Management Network",
"big": false
},
{
"t": "title",
"v": "Data Catalogue Community",
"big": false
},
{
"t": "title",
"v": "Information Governance Network",
"big": false
},
{
"t": "title",
"v": "Metadata & Standards Community",
"big": false
},
{
"t": "title",
"v": "Knowledge Management Community",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Ethics, safety and assurance"
},
{
"t": "small",
"v": "Ethics, AI governance and assurance across the public sector."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Cross-Government Data & AI Ethics Community",
"big": false
},
{
"t": "title",
"v": "AI Assurance Network",
"big": false
},
{
"t": "title",
"v": "Security & Privacy Community",
"big": false
},
{
"t": "title",
"v": "Accessibility & Inclusion Network",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Governance and policy network"
},
{
"t": "small",
"v": "Policy, standards, guidance and compliance."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Government Digital Service (GDS)",
"big": false
},
{
"t": "title",
"v": "DSIT AI Policy & Standards",
"big": false
},
{
"t": "title",
"v": "Government Analysis Function",
"big": false
},
{
"t": "title",
"v": "Identity & Access Community",
"big": false
},
{
"t": "title",
"v": "Service Design Network",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Commercial, finance and suppliers"
},
{
"t": "small",
"v": "Commercial strategy, vendor engagement and cost insight."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Commercial Community",
"big": false
},
{
"t": "title",
"v": "FinOps Community",
"big": false
},
{
"t": "title",
"v": "Supplier Engagement Community",
"big": false
},
{
"t": "title",
"v": "Contract & Licensing Community",
"big": false
},
{
"t": "title",
"v": "Procurement Policy Community",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cyber and risk network"
},
{
"t": "small",
"v": "Cyber security, risk management and resilience."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Cyber Security Network",
"big": false
},
{
"t": "title",
"v": "Risk & Assurance Community",
"big": false
},
{
"t": "title",
"v": "Resilience Community",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Digital workplace communities"
},
{
"t": "small",
"v": "Adoption, change, champions and enablement."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Copilot Weekly XGov Call",
"big": false
},
{
"t": "title",
"v": "Community of practice sessions",
"big": false
},
{
"t": "title",
"v": "Show and tell forums",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cross-sector and place-based connectors"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Central government departments & ALBs",
"big": false
},
{
"t": "title",
"v": "Devolved administrations",
"big": false
},
{
"t": "title",
"v": "Local government & combined authorities",
"big": false
},
{
"t": "title",
"v": "Health & care network",
"big": false
},
{
"t": "title",
"v": "Education network",
"big": false
},
{
"t": "title",
"v": "Policing & justice network",
"big": false
},
{
"t": "title",
"v": "International & supranational links",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Why the constellation matters"
},
{
"t": "p",
"v": "**Share once, use many times**Publish once so others do not repeat the work"
},
{
"t": "p",
"v": "**Avoid duplication and rework**Find what exists before building it again"
},
{
"t": "p",
"v": "**Consistent standards, safer adoption**Common expectations across organisations"
},
{
"t": "p",
"v": "**Better value and lower total cost**Aggregate demand and reuse assets"
},
{
"t": "p",
"v": "**Faster innovation with confidence**Learn from those who have already tried"
},
{
"t": "p",
"v": "**Better outcomes for citizens**The point of all of it"
}
]
}
],
"box": false
}
]
},
{
"id": "journeys",
"part": "people-and-adoption",
"title": "Adoption journey",
"icon": "route",
"eyebrow": "Journeys · adoption operating model",
"heading": "Adoption is a journey we take together",
"lede": "A system of people, processes and support, not a rollout. Leadership sets direction. Business owns outcomes. Communities enable change. Data and evidence drive continuous improvement.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "title",
"v": "People first",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "verified_user"
},
{
"t": "title",
"v": "Trust & responsibility",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "target"
},
{
"t": "title",
"v": "Outcome focused",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "autorenew"
},
{
"t": "title",
"v": "Iterate & improve",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "handshake"
},
{
"t": "title",
"v": "Collaborate & share",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "insights"
},
{
"t": "title",
"v": "Evidence & value",
"big": false
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Inputs"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Strategy & priorities",
"big": false
},
{
"t": "title",
"v": "User needs & frontline insights",
"big": false
},
{
"t": "title",
"v": "Data, content & tools",
"big": false
},
{
"t": "title",
"v": "Policy & regulatory requirements",
"big": false
},
{
"t": "title",
"v": "Evidence & external intelligence",
"big": false
},
{
"t": "title",
"v": "Budget & investment",
"big": false
}
]
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "num",
"v": "1"
},
{
"t": "h",
"l": 3,
"v": "Lead & sponsor"
}
],
"box": false
},
{
"t": "small",
"v": "Executive sponsorship and vision"
},
{
"t": "small",
"v": "Align to strategy and public value"
},
{
"t": "small",
"v": "Set priorities and risk appetite"
},
{
"t": "small",
"v": "Allocate investment and authority"
},
{
"t": "small",
"v": "Communicate the why"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "num",
"v": "2"
},
{
"t": "h",
"l": 3,
"v": "Engage & align"
}
],
"box": false
},
{
"t": "small",
"v": "Business ownership and champions"
},
{
"t": "small",
"v": "Stakeholder mapping and engagement"
},
{
"t": "small",
"v": "Co-design with users"
},
{
"t": "small",
"v": "Change impact assessment"
},
{
"t": "small",
"v": "Inclusive and accessible by design"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "num",
"v": "3"
},
{
"t": "h",
"l": 3,
"v": "Prepare & enable"
}
],
"box": false
},
{
"t": "small",
"v": "Skills, training and role-based learning"
},
{
"t": "small",
"v": "Guidance, guardrails and policies"
},
{
"t": "small",
"v": "Tools, environments and data readiness"
},
{
"t": "small",
"v": "Community of practice and peer support"
},
{
"t": "small",
"v": "Address accessibility and neurodiversity"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "num",
"v": "4"
},
{
"t": "h",
"l": 3,
"v": "Adopt & use"
}
],
"box": false
},
{
"t": "small",
"v": "Start small, iterate and learn"
},
{
"t": "small",
"v": "Embed in workflows and processes"
},
{
"t": "small",
"v": "Measure early value and experience"
},
{
"t": "small",
"v": "Support and coaching at point of need"
},
{
"t": "small",
"v": "Promote quick wins and success stories"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "num",
"v": "5"
},
{
"t": "h",
"l": 3,
"v": "Measure & improve"
}
],
"box": false
},
{
"t": "small",
"v": "Track usage, quality and outcomes"
},
{
"t": "small",
"v": "Measure value, cost and productivity"
},
{
"t": "small",
"v": "Gather feedback and identify issues"
},
{
"t": "small",
"v": "Improve models, processes and data"
},
{
"t": "small",
"v": "Report transparently and share learnings"
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "num",
"v": "6"
},
{
"t": "h",
"l": 3,
"v": "Scale & sustain"
}
],
"box": false
},
{
"t": "small",
"v": "Scale proven use cases and patterns"
},
{
"t": "small",
"v": "Standardise where appropriate"
},
{
"t": "small",
"v": "Manage risk, compliance and assurance"
},
{
"t": "small",
"v": "Optimise cost and resource use"
},
{
"t": "small",
"v": "Sustain, adapt and reinvest for impact"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Outcomes"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Better public services",
"big": false
},
{
"t": "title",
"v": "Improved user experience",
"big": false
},
{
"t": "title",
"v": "Lower cost, higher productivity",
"big": false
},
{
"t": "title",
"v": "Stronger compliance and trust",
"big": false
},
{
"t": "title",
"v": "Sustainable and adaptable",
"big": false
},
{
"t": "title",
"v": "Measurable public value",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Enabling foundations"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "database"
},
{
"t": "title",
"v": "Data & information foundations",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "lock"
},
{
"t": "title",
"v": "Security, privacy & safety",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "balance"
},
{
"t": "title",
"v": "Responsible AI, ethics & inclusion",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "settings"
},
{
"t": "title",
"v": "Technology & platforms",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "policy"
},
{
"t": "title",
"v": "Policies, standards & governance",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "payments"
},
{
"t": "title",
"v": "Commercial & FinOps",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "forum"
},
{
"t": "title",
"v": "Communities & knowledge sharing",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "support_agent"
},
{
"t": "title",
"v": "Operations & support",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Measurement & feedback loop"
},
{
"t": "row",
"items": [
{
"t": "p",
"v": "**1. Observe** usage, quality, cost, risk"
},
{
"t": "p",
"v": "**2. Analyse** results, feedback, insights"
},
{
"t": "p",
"v": "**3. Learn** what works, what doesn't"
},
{
"t": "p",
"v": "**4. Decide** prioritise actions and changes"
},
{
"t": "p",
"v": "**5. Act** improve, enable, remove barriers"
},
{
"t": "p",
"v": "**6. Repeat** continuous improvement"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Key roles"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Leadership",
"big": false
},
{
"t": "small",
"v": "Set direction, sponsor change and unblock"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Business owner",
"big": false
},
{
"t": "small",
"v": "Define outcomes, prioritise use cases and realise value"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Product / service manager",
"big": false
},
{
"t": "small",
"v": "Manage lifecycle, measure value and iterate"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Champion network",
"big": false
},
{
"t": "small",
"v": "Advocate, share learning and support peers"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Data & information professional",
"big": false
},
{
"t": "small",
"v": "Ensure quality, access, lineage and governance"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Security & privacy",
"big": false
},
{
"t": "small",
"v": "Manage risk, protect data and users"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Commercial & FinOps",
"big": false
},
{
"t": "small",
"v": "Manage contracts, licensing, cost and value"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Change & adoption team",
"big": false
},
{
"t": "small",
"v": "Drive engagement, training and communications"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Technical team",
"big": false
},
{
"t": "small",
"v": "Build, integrate and operate solutions"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Frontline users",
"big": false
},
{
"t": "small",
"v": "Use, feed back and help shape improvements"
}
]
}
],
"box": false
}
]
},
{
"t": "eyebrow",
"v": "The person-level journey"
},
{
"t": "h",
"l": 2,
"v": "From training to organisational capability"
},
{
"t": "p",
"v": "Six stages from awareness to community contribution — a progression, not a course. The continuous improvement loop runs across all six."
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Stage 1 of 6"
},
{
"t": "title",
"v": "All staff",
"big": false
}
]
},
{
"t": "h",
"l": 3,
"v": "Awareness"
},
{
"t": "title",
"v": "Understand the opportunity",
"big": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "Goal"
},
{
"t": "small",
"v": "Build awareness of AI basics and potential in public services."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "Key activities"
},
{
"t": "small",
"v": "Introductory sessions"
},
{
"t": "small",
"v": "AI fundamentals"
},
{
"t": "small",
"v": "Real-world examples"
},
{
"t": "small",
"v": "Responsible AI awareness"
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 4,
"v": "Outcome"
},
{
"t": "title",
"v": "Informed and curious workforce",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Support model"
},
{
"t": "small",
"v": "Central resources and guidance"
},
{
"t": "small",
"v": "Local support and champions"
},
{
"t": "small",
"v": "Communities of practice"
},
{
"t": "small",
"v": "Expert clinics and labs"
},
{
"t": "small",
"v": "Feedback and improvement"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Learning modes"
},
{
"t": "small",
"v": "Self-paced learning"
},
{
"t": "small",
"v": "Instructor-led training"
},
{
"t": "small",
"v": "On-the-job practice"
},
{
"t": "small",
"v": "Peer learning and sharing"
},
{
"t": "small",
"v": "Microlearning and nudges"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Measuring progress"
},
{
"t": "small",
"v": "Adoption rates"
},
{
"t": "small",
"v": "Capability assessments"
},
{
"t": "small",
"v": "Business outcomes"
},
{
"t": "small",
"v": "User satisfaction"
},
{
"t": "small",
"v": "Value realised"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers throughout the journey"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Leadership and sponsorship",
"big": false
},
{
"t": "title",
"v": "Clear guidance and guardrails",
"big": false
},
{
"t": "title",
"v": "Accessible tools and environments",
"big": false
},
{
"t": "title",
"v": "Role-based learning and support",
"big": false
},
{
"t": "title",
"v": "Time to learn and experiment",
"big": false
},
{
"t": "title",
"v": "Recognition and career pathways",
"big": false
},
{
"t": "title",
"v": "Feedback and user research",
"big": false
},
{
"t": "title",
"v": "Measurement and value realisation",
"big": false
}
]
}
]
}
]
},
{
"id": "rollout",
"part": "people-and-adoption",
"title": "Agent rollout",
"icon": "rocket_launch",
"eyebrow": "Copilot programme · agent rollout",
"heading": "Why we decided not to just switch on first-party and Agent Builder agents",
"lede": "The opportunity is significant, but ownership, risk and consistency matter.",
"body": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Agents can influence real work"
},
{
"t": "p",
"v": "More than simple prompts, built well, they can support processes, provide advice, surface information and shape decisions."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Creation creates responsibility"
},
{
"t": "p",
"v": "Every agent needs a clear purpose, an accountable owner and active lifecycle management. Risk management is critical."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Scale can amplify inconsistency or provide opportunity"
},
{
"t": "p",
"v": "Through a culture of safe experimentation, coupled with close monitoring, we will find opportunities to deliver the best agents, while avoiding “agent-aggedon”."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Success is confident adoption and organisational maturity — not simply the speed of feature enablement.",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Finding the sweet spot — a “test & learn” approach"
},
{
"t": "small",
"v": "Innovate confidently. Manage risk. Deliver value. Together."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Experimentation at the core"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "rocket_launch"
},
{
"t": "title",
"v": "Start small",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "science"
},
{
"t": "title",
"v": "Test & learn",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "autorenew"
},
{
"t": "title",
"v": "Iterate quickly",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "lightbulb"
},
{
"t": "title",
"v": "Learn & adapt",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "eyebrow",
"v": "Delivering"
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Outcomes that matter"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "bar_chart"
},
{
"t": "title",
"v": "Business value",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "title",
"v": "User adoption",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "autorenew"
},
{
"t": "title",
"v": "Continuous improvement",
"big": false
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "verified_user"
},
{
"t": "title",
"v": "Sustainable scale",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Start small. Learn safely. Scale with confidence."
},
{
"t": "p",
"v": "Public Beta and measured communications ramp up gives the organisation space to learn about the technology, ownership and responsibility before broader scale."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "search"
},
{
"t": "small",
"v": "Copilot Agent Builder is one of many business tools"
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "search"
},
{
"t": "small",
"v": "Assurance requirements vary by risk and operational impact"
}
]
},
{
"t": "group",
"kids": [
{
"t": "icon",
"v": "search"
},
{
"t": "small",
"v": "Business areas tailor guidance to their environment"
}
]
}
],
"box": true
}
]
},
{
"t": "h",
"l": 2,
"v": "Governance that enables, not blocks"
},
{
"t": "small",
"v": "Proportionate guardrails create confidence to experiment."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Proactive, not reactive"
},
{
"t": "small",
"v": "Put standards and guidance in place early."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Central guardrails, local accountability"
},
{
"t": "small",
"v": "Set common expectations; keep responsibility with agent owners."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Proportionate and risk-based"
},
{
"t": "small",
"v": "Apply controls according to use case and data sensitivity."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Test & learn"
},
{
"t": "small",
"v": "Evolve the model using real scenarios and evidence."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What good looks like"
},
{
"t": "small",
"v": "Agents are visible, owned and managed through their lifecycle."
},
{
"t": "small",
"v": "Responsible-use standards are embedded into governance and guidance."
},
{
"t": "small",
"v": "Oversight covers usage, value and risk signals."
},
{
"t": "small",
"v": "High-value use cases can progress without unnecessary friction."
},
{
"t": "small",
"v": "Experimentation remains welcome within clear guardrails."
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Governance is the licence to innovate confidently.",
"big": false
}
]
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Support around every colleague"
},
{
"t": "p",
"v": "Technology enablement is paired with readiness, guidance and local ownership."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Clear guidance"
},
{
"t": "small",
"v": "What colleagues can use, when and how to work within the guardrails."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Communications"
},
{
"t": "p",
"v": "Initial soft launch, then expand messages around use cases and value. Align to OBT."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Peer support"
},
{
"t": "p",
"v": "Digital Champions, Sponsors and business representatives help locally."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Learning"
},
{
"t": "p",
"v": "Options from “getting started” through to building agents, focused on real use cases."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Adoption support"
},
{
"t": "p",
"v": "Tailored help to embed useful agents into team workflows. May extend to Tiger Teams."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Programme sets the foundation for safe, consistent and responsible use",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Business areas adapts to local needs and drives adoption in their context",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Measure what matters"
},
{
"t": "p",
"v": "Reporting is not the outcome. It is the feedback loop that helps us decide what to support, change, scale or retire."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Builders"
},
{
"t": "small",
"v": "Who is creating agents?"
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Ownership"
},
{
"t": "small",
"v": "Are agents visible and actively owned?"
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Usage"
},
{
"t": "small",
"v": "Which agents are being used?"
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Enablement"
},
{
"t": "small",
"v": "Are learning and communications changing behaviour?"
}
]
},
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Value"
},
{
"t": "small",
"v": "What outcomes and benefits are emerging?"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Challenges we are facing into"
},
{
"t": "small",
"v": "Reporting will mature"
},
{
"t": "small",
"v": "Identify duplication — spot the best agents"
},
{
"t": "small",
"v": "Alert when agents breach a usage or share threshold"
},
{
"t": "p",
"v": "Business takes ownership of their use cases and determines the best tool for the job"
},
{
"t": "small",
"v": "Local resources to take action on the measures"
}
]
}
],
"box": false
}
]
},
{
"id": "community",
"part": "people-and-adoption",
"title": "Community engagement",
"icon": "forum",
"eyebrow": "People and adoption · community",
"heading": "From isolated questions to collective capability",
"lede": "How a cross-government community matures: the engagement arc, the flywheel that turns experience into shared assets, and the honest exchange of what members give and get. Capability without community repeats mistakes; community without operationalisation produces discussion.",
"body": [
{
"t": "h",
"l": 2,
"v": "The engagement arc"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Information access & orientation"
},
{
"t": "title",
"v": "Establishing the forum and enabling participation",
"big": false
},
{
"t": "small",
"v": "Join meetings"
},
{
"t": "small",
"v": "Access chat and resources"
},
{
"t": "small",
"v": "Understand purpose"
},
{
"t": "small",
"v": "Share initial concerns"
},
{
"t": "small",
"v": "Learn community norms"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Exploration & safe experimentation"
},
{
"t": "title",
"v": "Understanding tools and testing possibilities",
"big": false
},
{
"t": "small",
"v": "Product changes & updates"
},
{
"t": "small",
"v": "Prompting and capabilities"
},
{
"t": "small",
"v": "Early pilots and proofs"
},
{
"t": "small",
"v": "Controls and guardrails"
},
{
"t": "small",
"v": "Licensing questions"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Use cases & practical learning"
},
{
"t": "title",
"v": "Sharing examples and learning what works",
"big": false
},
{
"t": "small",
"v": "Show & tell sessions"
},
{
"t": "small",
"v": "Success stories"
},
{
"t": "small",
"v": "Failures and lessons"
},
{
"t": "small",
"v": "Templates and patterns"
},
{
"t": "small",
"v": "Emerging use cases"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Governance & assurance emerge"
},
{
"t": "title",
"v": "Defining guardrails and ways of working",
"big": false
},
{
"t": "small",
"v": "Ownership and accountability"
},
{
"t": "small",
"v": "Data, security & privacy"
},
{
"t": "small",
"v": "Information management"
},
{
"t": "small",
"v": "Human oversight"
},
{
"t": "small",
"v": "Assurance approaches"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Adoption & scaling together"
},
{
"t": "title",
"v": "Enabling confident organisational adoption",
"big": false
},
{
"t": "small",
"v": "Adoption strategies"
},
{
"t": "small",
"v": "Change & comms"
},
{
"t": "small",
"v": "Training & champions"
},
{
"t": "small",
"v": "FinOps & cost visibility"
},
{
"t": "small",
"v": "Scaled rollouts"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Capability & continuous improvement"
},
{
"t": "title",
"v": "Embedding capability and driving improvement",
"big": false
},
{
"t": "small",
"v": "Measuring outcomes"
},
{
"t": "small",
"v": "Community feedback loop"
},
{
"t": "small",
"v": "Reusable assets"
},
{
"t": "small",
"v": "Influencing suppliers"
},
{
"t": "small",
"v": "Evolving standards"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "How we engage"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "groups"
},
{
"t": "title",
"v": "Inclusive participation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "forum"
},
{
"t": "title",
"v": "Open questions & discussion",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "share"
},
{
"t": "title",
"v": "Share knowledge & experience",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "psychology"
},
{
"t": "title",
"v": "Challenge & constructively critique",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "assignment_turned_in"
},
{
"t": "title",
"v": "Capture decisions & actions",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "autorenew"
},
{
"t": "title",
"v": "Feedback & iterate",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers"
},
{
"t": "small",
"v": "Safe space"
},
{
"t": "small",
"v": "Regular cadence"
},
{
"t": "small",
"v": "Diverse perspectives"
},
{
"t": "small",
"v": "Community sponsorship"
},
{
"t": "small",
"v": "Accessible resources"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Outcomes"
},
{
"t": "p",
"v": "**Greater clarity & confidence** Less ambiguity about products, controls and next steps."
},
{
"t": "p",
"v": "**Reduced risk & duplication** Shared lessons prevent repeated mistakes."
},
{
"t": "p",
"v": "**Reusable knowledge** Patterns, templates and guidance that can be adapted."
},
{
"t": "p",
"v": "**Faster & safer implementation** Less rework, clearer pathways, better preparation."
},
{
"t": "p",
"v": "**Stronger supplier engagement** Consolidated questions and influence based on community insight."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Value created"
},
{
"t": "title",
"v": "Better public services",
"big": false
},
{
"t": "title",
"v": "Better use of public money",
"big": false
},
{
"t": "title",
"v": "Increased trust & assurance",
"big": false
},
{
"t": "title",
"v": "Continuous improvement of AI capability",
"big": false
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Continuous learning loop"
},
{
"t": "title",
"v": "Experience",
"big": false
},
{
"t": "title",
"v": "Insight",
"big": false
},
{
"t": "title",
"v": "Guidance",
"big": false
},
{
"t": "title",
"v": "Action",
"big": false
},
{
"t": "title",
"v": "Impact",
"big": false
},
{
"t": "title",
"v": "Feedback",
"big": false
},
{
"t": "title",
"v": "Improved practice",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "The collective learning flywheel"
},
{
"t": "small",
"v": "The more we share, the more we all improve."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "num",
"v": "01"
},
{
"t": "h",
"l": 3,
"v": "Bring questions & challenges"
},
{
"t": "small",
"v": "Surface real needs and uncertainties"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "02"
},
{
"t": "h",
"l": 3,
"v": "Share evidence & experience"
},
{
"t": "small",
"v": "Contribute what you have seen and tried"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "03"
},
{
"t": "h",
"l": 3,
"v": "Compare, challenge & analyse"
},
{
"t": "small",
"v": "Learn together and test ideas"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "04"
},
{
"t": "h",
"l": 3,
"v": "Create insights & patterns"
},
{
"t": "small",
"v": "Turn learning into shared assets"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "05"
},
{
"t": "h",
"l": 3,
"v": "Apply, adapt & implement"
},
{
"t": "small",
"v": "Put learning into practice locally"
}
]
},
{
"t": "card",
"kids": [
{
"t": "num",
"v": "06"
},
{
"t": "h",
"l": 3,
"v": "Report outcomes & lessons"
},
{
"t": "small",
"v": "Share results and what is next"
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Inputs from the community"
},
{
"t": "small",
"v": "Real problems & open questions"
},
{
"t": "small",
"v": "Use cases & demonstrations"
},
{
"t": "small",
"v": "Challenges, risks & failures"
},
{
"t": "small",
"v": "Policies, approaches & processes"
},
{
"t": "small",
"v": "Cost, licensing & FinOps data"
},
{
"t": "small",
"v": "Security, privacy & assurance issues"
},
{
"t": "small",
"v": "Ideas, suggestions & future needs"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Outcomes for all"
},
{
"t": "small",
"v": "Practical guidance & templates"
},
{
"t": "small",
"v": "Stronger guardrails & assurance"
},
{
"t": "small",
"v": "Reusable patterns & components"
},
{
"t": "small",
"v": "Skills, confidence & capability"
},
{
"t": "small",
"v": "Smarter investment & cost control"
},
{
"t": "small",
"v": "Influence with suppliers & partners"
},
{
"t": "small",
"v": "Better services & public value"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Enablers of the flywheel"
},
{
"t": "small",
"v": "Inclusive & welcoming culture"
},
{
"t": "small",
"v": "Safe spaces for openness"
},
{
"t": "small",
"v": "Accessible knowledge & tools"
},
{
"t": "small",
"v": "Clear communication & feedback"
},
{
"t": "small",
"v": "Recognition of contribution"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The value exchange"
},
{
"t": "p",
"v": "What members contribute, what the community turns it into, and what everyone gains."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What we contribute"
},
{
"t": "p",
"v": "**Our questions & challenges** Share real problems, uncertainties and barriers"
},
{
"t": "p",
"v": "**Use cases & demos** Show what we are building, testing and learning"
},
{
"t": "p",
"v": "**Successes & lessons** Share what works, what did not and why"
},
{
"t": "p",
"v": "**Policies & approaches** Contribute guidance, guardrails and governance ideas"
},
{
"t": "p",
"v": "**Tools & templates** Share artefacts, patterns, checklists and assets"
},
{
"t": "p",
"v": "**Commercial & cost insight** Raise licensing, FinOps and procurement experiences"
},
{
"t": "p",
"v": "**Security, privacy & assurance** Share risk findings, controls and assurance practice"
},
{
"t": "p",
"v": "**Our time & expertise** Offer help, review, connect others and support each other"
}
]
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "**Open discussions & peer challenge** Honest conversations that surface real issues"
},
{
"t": "p",
"v": "**Shared evidence & patterns** Collective intelligence from across government"
},
{
"t": "p",
"v": "**Expert input & supplier access** Access to Microsoft, experts and the wider network"
},
{
"t": "p",
"v": "**Connections & collaboration** Find partners, avoid duplication and work together"
},
{
"t": "p",
"v": "**Influence & impact** Consolidate our voice to shape products, policy and standards"
},
{
"t": "p",
"v": "**Capability uplift & confidence** Improve skills, maturity and organisational readiness"
},
{
"t": "p",
"v": "**Early awareness & reducing risk** Spot changes, risks and opportunities earlier"
},
{
"t": "p",
"v": "**Reusable outputs & guidance** Turn learning into practical guidance and tools"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "What we gain"
},
{
"t": "p",
"v": "**Practical answers** Faster solutions to real problems"
},
{
"t": "p",
"v": "**Time saved** Learn from others' experience, avoid repeating mistakes"
},
{
"t": "p",
"v": "**Safer adoption** Better decisions with shared assurance and guardrails"
},
{
"t": "p",
"v": "**Better quality outcomes** Access to proven patterns and worked examples"
},
{
"t": "p",
"v": "**Cost confidence** Understand licensing, usage and FinOps more clearly"
},
{
"t": "p",
"v": "**Stronger relationships** Collaborate across departments, sectors and professions"
},
{
"t": "p",
"v": "**Measurable progress** Track maturity, impact and value together"
},
{
"t": "p",
"v": "**Greater public value** Better services, better use of public money, more trust"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "Shared principle: be open, be respectful, challenge constructively and focus on learning together to deliver safe, inclusive and effective AI in the public sector."
}
]
}
]
},
{
"id": "operating",
"part": "people-and-adoption",
"title": "Operating model",
"icon": "account_tree",
"eyebrow": "Operating model & programme",
"heading": "One operating model. Everyone aboard.",
"lede": "One AI capability operating model — not separate Copilot, agent, data, assurance and FinOps programmes. Cognitive load down, recognition up, DEIA by default rather than afterthought.",
"body": [
{
"t": "h",
"l": 2,
"v": "Four defaults, baked in"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Privacy by default",
"big": false
},
{
"t": "small",
"v": "Every product, every build. DPIA in, not post."
},
{
"t": "eyebrow",
"v": "Existing"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Secure by default",
"big": false
},
{
"t": "small",
"v": "Open, auditable, no vendor lock-in."
},
{
"t": "eyebrow",
"v": "Existing"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "DEIA by default",
"big": false
},
{
"t": "small",
"v": "Inclusion, equity, access — in every spec."
},
{
"t": "eyebrow",
"v": "Adding now"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Recognition by default",
"big": false
},
{
"t": "small",
"v": "Named people, visible work, earned respect."
},
{
"t": "eyebrow",
"v": "Adding now"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Reducing cognitive load"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Less red tape",
"big": false
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Decision logs replace approval chains"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "AI drafts briefings; you review"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Single source of truth per project"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Open by default, restrict by exception"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Cross-pollination",
"big": false
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Weekly cross-team show-and-tells"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Shared prompt and tool library"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "SME pairing across directorates"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Secondments and shadow programmes"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Leadership engagement",
"big": false
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Monthly programme board updates"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Outcome dashboards, not status reports"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Senior sponsors own a use-case"
},
{
"t": "icon",
"v": "check_circle"
},
{
"t": "small",
"v": "Ministers briefed on citizen impact"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "DEIA by default — who this serves, and the checklist"
},
{
"t": "p",
"v": "Every product must actively serve those experiencing the greatest adversity: economic hardship, disabled and neurodivergent users, underrepresented communities, vulnerable households and carers, digital exclusion and low literacy, and everyone algorithmic unfairness would otherwise reach first."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Who is most harmed if this fails or excludes?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Does the design assume internet access or literacy?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Has it been tested with people from affected groups?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Are legal obligations (Equality Act, PSED) documented?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Is there a non-digital fallback route?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Does the data we collect reflect the full population?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "Are there language and accessibility alternatives?"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "check"
},
{
"t": "small",
"v": "How will we measure equitable outcomes, not just usage?"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Bias will be reproduced unless actively constrained"
},
{
"t": "p",
"v": "A model trained on historical outcomes learns the discrimination in the record, not the potential of a well-supported person. Four pre-specified bias domains to audit — and abstention (\"insufficient data — human decision only\") is always acceptable, and must be possible."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Allocation",
"big": false
},
{
"t": "p",
"v": "Systematically fewer or lesser recommendations for protected groups with equivalent profiles."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Calibration",
"big": false
},
{
"t": "p",
"v": "Predictions accurate against a biased historical record, wrong about supported potential."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Representation gap",
"big": false
},
{
"t": "small",
"v": "Too few cases for a group; the model fills gaps with majority priors."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Feedback loop",
"big": false
},
{
"t": "p",
"v": "Biased outputs produce worse outcomes that retrain the model — break the loop by design."
}
]
}
],
"box": false
},
{
"t": "p",
"v": "Training-data standards: exclude invalidated-treatment-era and sanctioned-employer records and proxy features; require minimum cases per group, representation audits, labelled synthetic augmentation for regularisation only, and temporal weighting toward modern practice. Lived-experience panels review real outputs; community bodies sign off the test criteria; explanations reference functional needs, never the diagnosis."
}
]
},
{
"t": "h",
"l": 2,
"v": "The capability programme — four phases"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "**Phase 0 · Foundation**Weeks 1–4"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Starter capability, safely",
"big": false
},
{
"t": "small",
"v": "Private, data-safe assistant; no lock-in; GDPR by design"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Data protection baseline",
"big": false
},
{
"t": "small",
"v": "DPIA template, handling principles, red lines"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Use-case registry",
"big": false
},
{
"t": "small",
"v": "Crowdsource the top jobs-to-be-done; impact × feasibility"
}
]
}
],
"box": false
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "**Phase 1 · Pilot**Months 2–4"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Learning pathway",
"big": false
},
{
"t": "small",
"v": "Prompting fundamentals, hands-on sprints, peer cohorts"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Open-tool pilots",
"big": false
},
{
"t": "small",
"v": "Open-weight models on gov cloud; avoid lock-in from day one"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Impact measurement",
"big": false
},
{
"t": "small",
"v": "Hours saved, quality uplift, citizen outcomes — not vanity metrics"
}
]
}
],
"box": false
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "**Phase 2 · Scale**Months 5–9"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Shared platform",
"big": false
},
{
"t": "small",
"v": "Modular, containerised infrastructure across directorates"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Champions network",
"big": false
},
{
"t": "small",
"v": "One champion per team; shared prompt library; show-and-tells"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Legacy reduction",
"big": false
},
{
"t": "small",
"v": "AI-assisted modernisation, documentation, automation"
}
]
}
],
"box": false
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "**Phase 3 · Embed**Months 10–18"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Public trust framework",
"big": false
},
{
"t": "small",
"v": "Transparent registers, explainability standards, citizen comms"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Recognition & visibility",
"big": false
},
{
"t": "small",
"v": "Awards, case-study library, cross-government showcases"
}
]
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Continuous learning",
"big": false
},
{
"t": "small",
"v": "Evergreen curriculum, AI-native onboarding, horizon scans"
}
]
}
],
"box": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "100%"
},
{
"t": "small",
"v": "team with baseline AI literacy"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "≥10",
"big": false
},
{
"t": "small",
"v": "measurable use-cases live"
}
]
},
{
"t": "card",
"kids": [
{
"t": "stat",
"v": "0"
},
{
"t": "small",
"v": "proprietary lock-in"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "DPIA",
"big": false
},
{
"t": "small",
"v": "embedded in every build"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Named",
"big": false
},
{
"t": "small",
"v": "civil servants celebrated"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "The sequence — three, six, twelve, twenty-four months"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "3 months"
},
{
"t": "h",
"l": 3,
"v": "Establish preconditions"
},
{
"t": "small",
"v": "Reachability audit and permissions remediation"
},
{
"t": "small",
"v": "Publish agent-register schema and intake form"
},
{
"t": "small",
"v": "Consumption dashboards and per-agent caps"
},
{
"t": "small",
"v": "Agree retention and FOI position"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "6 months"
},
{
"t": "h",
"l": 3,
"v": "Govern"
},
{
"t": "small",
"v": "Operate the register; classify every agent by tier"
},
{
"t": "small",
"v": "Apply assurance tiers; define the orphaned-agent process"
},
{
"t": "small",
"v": "Secure exit, portability and audit rights"
},
{
"t": "small",
"v": "Publish an evaluation method standard"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "12 months"
},
{
"t": "h",
"l": 3,
"v": "Integrate"
},
{
"t": "small",
"v": "Fund adoption as a standing operating line"
},
{
"t": "small",
"v": "Publish unit economics per production agent"
},
{
"t": "small",
"v": "Unify information and agent governance"
},
{
"t": "small",
"v": "Publish one outcome measure, not an activity"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "24 months"
},
{
"t": "h",
"l": 3,
"v": "Federate"
},
{
"t": "small",
"v": "Share registered agent patterns across bodies"
},
{
"t": "small",
"v": "Aggregate demand for small public bodies"
},
{
"t": "small",
"v": "Prove exit is real via a live substitution"
},
{
"t": "small",
"v": "Publish a community value evaluation"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "Sequence matters more than speed: permissions before scale, register before governance claims, cost instrumentation before volume. Recognition runs throughout — named people, visible work, credited contributions, minister-level visibility on citizen wins."
}
]
},
{
"id": "raiadoption",
"part": "people-and-adoption",
"title": "Adoption & change model",
"icon": "diversity_1",
"eyebrow": "Responsible AI · people and adoption",
"heading": "People at the centre: the adoption, capability and change model",
"lede": "Eight segments that turn responsible AI into lasting value, with accessibility, guest access and ways-of-working impact. Teams, roles, the academy and the adoption journey in this part go deeper on each.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Build capability"
},
{
"t": "title",
"v": "Leadership and sponsorship",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Clear vision and strategic alignment",
"Senior sponsor and accountable owners",
"Cross-government collaboration",
"Adequate funding and resources",
"Visible role modelling"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Build capability"
},
{
"t": "title",
"v": "Skills, learning and enablement",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Role-based learning pathways",
"Practical, hands-on training",
"Prompts, skills and agent use",
"Protected learning time",
"Ongoing refresher content"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Engage and empower"
},
{
"t": "title",
"v": "Communication and engagement",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Clear, consistent messaging",
"What is approved, and why",
"How to get help and give feedback",
"Real examples and case studies",
"Address concerns and misinformation"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Engage and empower"
},
{
"t": "title",
"v": "Champions and communities",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Local AI champions and networks",
"Communities of practice such as xGov",
"Share prompts, skills and patterns",
"Connect pilot teams and early adopters",
"Recognise success"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Support and sustain"
},
{
"t": "title",
"v": "Support and service management",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Clear support model and service levels",
"Guidance, FAQs and how-to resources",
"Escalation for issues and incidents",
"Access to specialist expertise",
"Integration with IT and business support"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Support and sustain"
},
{
"t": "title",
"v": "Measurement and value",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Track adoption and meaningful use",
"Measure quality, productivity and outcomes",
"Include accessibility and inclusion benefits",
"Monitor cost and environmental impact",
"Report value to stakeholders"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Learn and improve"
},
{
"t": "title",
"v": "Continuous improvement",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Capture feedback and lessons",
"Share good practice across government",
"Update guidance, training and controls",
"Respond to product and model changes",
"Reassess risks and opportunities"
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Learn and improve"
},
{
"t": "title",
"v": "Workforce, ways of working and inclusion",
"big": false
},
{
"t": "list",
"o": false,
"items": [
"Understand role impacts and job changes",
"Support new ways of working",
"Engage staff and representative groups",
"Promote inclusive and accessible design",
"Address capability gaps and change fatigue"
]
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Accessibility and workplace adjustments",
"big": false
},
{
"t": "p",
"v": "Two questions, kept separate: is the AI tool itself accessible, and can AI support someone as a workplace adjustment? Request either through your line manager or the adjustments route. Every tool needs a non-AI alternative."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Guest and cross-tenant access",
"big": false
},
{
"t": "p",
"v": "Do not rely on meeting recaps for guests. Store resources where external participants can reach them, set permissions deliberately and provide an accessible copy."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Transcription dictionary",
"big": false
},
{
"t": "p",
"v": "Maintain a controlled glossary of cross-government terms and acronyms so that transcription and summaries get names right."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Employment and workplace effects",
"big": false
},
{
"t": "p",
"v": "Include workload, role, delegation, autonomy, training and workplace-adjustment impact in the adoption assessment, and monitor them after rollout."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Enablers across the model"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Policy and governance",
"big": false
},
{
"t": "title",
"v": "Approved tools and data",
"big": false
},
{
"t": "title",
"v": "Security and privacy",
"big": false
},
{
"t": "title",
"v": "Accessible by design",
"big": false
},
{
"t": "title",
"v": "Sustainability by design",
"big": false
},
{
"t": "title",
"v": "Cost control and FinOps",
"big": false
},
{
"t": "title",
"v": "Supplier and model management",
"big": false
},
{
"t": "title",
"v": "Change management",
"big": false
},
{
"t": "title",
"v": "Cross-tenant and external working",
"big": false
},
{
"t": "title",
"v": "Legal, ethical and regulatory compliance",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Desired outcomes"
},
{
"t": "icon",
"v": "sentiment_satisfied"
},
{
"t": "title",
"v": "Greater confidence",
"big": false
},
{
"t": "icon",
"v": "star"
},
{
"t": "title",
"v": "Higher productivity and quality",
"big": false
},
{
"t": "icon",
"v": "diversity_3"
},
{
"t": "title",
"v": "More inclusive services",
"big": false
},
{
"t": "icon",
"v": "eco"
},
{
"t": "title",
"v": "Reduced environmental impact",
"big": false
},
{
"t": "icon",
"v": "flag"
},
{
"t": "title",
"v": "Better outcomes for citizens",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Ways-of-working impact questions"
},
{
"t": "list",
"o": false,
"items": [
"Could it replace developmental delegation or professional review?",
"Could it create meeting, message or document overload?",
"How will roles and responsibilities change?",
"What training and protected time are required?",
"How will non-users be affected?",
"What workforce evidence will be monitored?"
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Worked example"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Worked example"
},
{
"t": "small",
"v": "Portfolio reporting grounded in the Graph."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Worked example"
},
{
"t": "small",
"v": "The same day run on Claude Projects and scheduled tasks."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Worked example"
},
{
"t": "small",
"v": "Whole-portfolio reading in one context."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Training film"
},
{
"t": "small",
"v": "The Microsoft journey, end to end."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Training film"
},
{
"t": "title",
"v": "Twin journeys film →",
"big": false
},
{
"t": "small",
"v": "Both stacks, the same functions."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Comparison"
},
{
"t": "title",
"v": "Twin Capability Atlas →",
"big": false
},
{
"t": "small",
"v": "39 capabilities with residency and FinOps layers."
}
]
}
],
"box": false
}
]
},
{
"id": "pilotview",
"part": "live-pilot",
"title": "The pilot in one view",
"icon": "play_circle",
"eyebrow": "Live pilot handbook · prototype pilot pack, 25 August 2026",
"heading": "Testing this guide through a real deployment",
"lede": "We are not reviewing a document. We are using the playbook during a live AI deployment to determine whether it improves decisions, delivery, assurance and public-service outcomes, and whether there is sufficient evidence and appetite to establish a permanent, interactive toolkit.",
"body": [
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**Evidence status.** The pilot model, thresholds and templates are proposed practice. They must be agreed for the selected live deployment. No pilot outcome is claimed in advance."
}
]
},
{
"t": "h",
"l": 2,
"v": "Seven stages, seven decisions"
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Mobilise"
},
{
"t": "small",
"v": "Is the pilot worth initiating?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Mandate, sponsor, scope and evaluation plan."
},
{
"t": "title",
"v": "Proceed or stop.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Select"
},
{
"t": "small",
"v": "Is the deployment suitable and safe for a pilot?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Candidate assessment and baseline."
},
{
"t": "title",
"v": "Select, condition or reject.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Prepare"
},
{
"t": "small",
"v": "Are people, controls, evidence and fallbacks ready?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Readiness and assurance plan."
},
{
"t": "title",
"v": "Authorise controlled deployment.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Deploy"
},
{
"t": "small",
"v": "Can the playbook be used during real delivery?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Operational evidence and decision trail."
},
{
"t": "title",
"v": "Continue, restrict, pause or stop.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Challenge"
},
{
"t": "small",
"v": "Did governance and assurance improve decisions?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Independent findings and participant evidence."
},
{
"t": "title",
"v": "Accept, remediate or retest.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Evaluate"
},
{
"t": "small",
"v": "Was the playbook useful, proportionate and valuable?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Evaluation and options appraisal."
},
{
"t": "title",
"v": "Stop, retain, simplify or evolve.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Decide"
},
{
"t": "small",
"v": "Is there appetite and evidence for a permanent capability?"
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Investment recommendation and roadmap."
},
{
"t": "title",
"v": "Toolkit decision.",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Who this is for, and their first action"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Sponsor or SRO"
},
{
"t": "small",
"v": "Decision points, risk boundaries, evidence and investment gate."
},
{
"t": "title",
"v": "First action: Confirm outcome, authority and resourcing.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Candidate team"
},
{
"t": "p",
"v": "Eligibility, readiness criteria, expected contribution and safeguards."
},
{
"t": "title",
"v": "First action: Complete the candidate screen.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Pilot delivery team"
},
{
"t": "small",
"v": "Step-by-step lifecycle, artefacts, ceremonies and feedback routes."
},
{
"t": "title",
"v": "First action: Attend induction and baseline assessment.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance reviewer"
},
{
"t": "small",
"v": "Review lenses, evidence standards and independence rules."
},
{
"t": "title",
"v": "First action: Agree assurance plan and conflicts.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Interested party"
},
{
"t": "p",
"v": "Ways to observe, challenge and contribute without disrupting delivery."
},
{
"t": "title",
"v": "First action: Choose an engagement route.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Toolkit product team"
},
{
"t": "p",
"v": "Evidence needed to decide whether to stop, retain a static guide or build a living toolkit."
},
{
"t": "title",
"v": "First action: Establish evaluation repository.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"id": "pilottest",
"part": "live-pilot",
"title": "What it tests",
"icon": "biotech",
"eyebrow": "Live pilot · hypothesis and test dimensions",
"heading": "What the pilot is testing",
"lede": "Primary hypothesis: using the playbook during a live deployment will improve the clarity, consistency, traceability and proportionality of governance and assurance without imposing unjustified delivery burden.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Usability",
"big": false
},
{
"t": "p",
"v": "People can find and apply the right guidance without the author present."
},
{
"t": "small",
"v": "Task observation, completion, help requests and interviews."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Proportionality",
"big": false
},
{
"t": "p",
"v": "Controls reflect risk, autonomy, data, decision influence and public effect."
},
{
"t": "small",
"v": "Exceptions, duplicated work, effort diary and reviewer judgement."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Decision quality",
"big": false
},
{
"t": "p",
"v": "Material decisions have clear owners, evidence, uncertainty and conditions."
},
{
"t": "small",
"v": "Decision records and assurance sampling."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Delivery value",
"big": false
},
{
"t": "small",
"v": "The playbook prevents rework or reveals issues early enough to act."
},
{
"t": "small",
"v": "Before and after comparison, avoided rework and issue chronology."
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Assurance quality",
"big": false
},
{
"t": "small",
"v": "Claims are tested against meaningful evidence."
},
{
"t": "small",
"v": "Findings, closure evidence and independent opinion."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Learning value",
"big": false
},
{
"t": "small",
"v": "Participants gain usable capability and confidence."
},
{
"t": "small",
"v": "Applied assessments, job-aid use and follow-up interviews."
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Future demand",
"big": false
},
{
"t": "p",
"v": "Teams and leaders prefer a maintained toolkit over no framework or a static document."
},
{
"t": "small",
"v": "Investment survey, sponsor decision and committed use cases."
}
],
"box": true
},
{
"t": "h",
"l": 2,
"v": "Non-objectives"
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "block"
},
{
"t": "small",
"v": "Proving that every AI deployment should use identical controls."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "block"
},
{
"t": "small",
"v": "Treating participant satisfaction as proof of safety or value."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "block"
},
{
"t": "small",
"v": "Using a successful pilot to claim cross-government impact."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "block"
},
{
"t": "p",
"v": "Replacing formal legal, security, privacy, commercial or service-assessment obligations."
}
]
}
]
},
{
"id": "pilotselect",
"part": "live-pilot",
"title": "Candidate selection",
"icon": "how_to_reg",
"eyebrow": "Live pilot · candidate selection",
"heading": "Choosing a deployment worth learning from",
"lede": "Seven criteria, the evidence each needs, and the condition that turns a candidate into a rejection or an escalation.",
"body": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Real deployment"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Named service, users, owner and intended live or limited-live use."
},
{
"t": "small",
"v": "Reject or escalate when: There is only a hypothetical demo."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Bounded purpose"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Defined tasks, prohibited uses and measurable outcomes."
},
{
"t": "small",
"v": "Reject or escalate when: Purpose is open-ended or unowned."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Safe test route"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Controlled environment, data route, fallback and stop mechanism."
},
{
"t": "small",
"v": "Reject or escalate when: Critical safeguards are unavailable."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Representative learning"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "The candidate exercises meaningful parts of the playbook."
},
{
"t": "p",
"v": "Reject or escalate when: The use case is too trivial to generate evidence."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Team capacity"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Product, operational, technical and governance participation."
},
{
"t": "p",
"v": "Reject or escalate when: Delivery relies on one person or unavailable specialists."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence access"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Permission to examine decisions, tests, incidents and effort."
},
{
"t": "small",
"v": "Reject or escalate when: Evaluation cannot access material evidence."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Change tolerance"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Team accepts challenge and can alter or pause deployment."
},
{
"t": "p",
"v": "Reject or escalate when: Go-live is predetermined regardless of evidence."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Recommended cohort"
},
{
"t": "p",
"v": "Use three contrasting deployments where feasible: a bounded low-risk assistant, a reusable internal knowledge or workflow agent, and a higher-impact agent influencing operational judgement. This is a suggested design, not a mandated sample."
}
]
}
]
},
{
"id": "pilotroles",
"part": "live-pilot",
"title": "Pilot roles",
"icon": "badge",
"eyebrow": "Live pilot · participants and responsibilities",
"heading": "Seventeen roles, one accountable line",
"lede": "The deployment SRO stays accountable for the live service throughout. The pilot adds evaluation and independent challenge; it does not take accountability away from delivery.",
"body": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Executive sponsor",
"big": false
},
{
"t": "p",
"v": "Protects the purpose, secures capacity and receives the investment recommendation."
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Pilot SRO",
"big": false
},
{
"t": "small",
"v": "Accountable for pilot outcomes, scope and residual risk."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Deployment SRO",
"big": false
},
{
"t": "small",
"v": "Remains accountable for the live AI-enabled service."
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Pilot manager",
"big": false
},
{
"t": "small",
"v": "Runs plan, ceremonies, dependencies, evidence and communications."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Product manager",
"big": false
},
{
"t": "small",
"v": "Applies the playbook to needs, outcomes and backlog decisions."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Service owner",
"big": false
},
{
"t": "small",
"v": "Owns operational performance, support, fallback and continuity."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "User researcher",
"big": false
},
{
"t": "small",
"v": "Tests usability and captures staff and affected-user evidence."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Service designer",
"big": false
},
{
"t": "small",
"v": "Maps whole service, human controls, channels and failure demand."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Technical owner / AI engineer",
"big": false
},
{
"t": "p",
"v": "Builds bounded capability and supplies evaluation and operational evidence."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Data and information lead",
"big": false
},
{
"t": "small",
"v": "Assures sources, provenance, metadata, retention and records."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Security and privacy leads",
"big": false
},
{
"t": "small",
"v": "Assess threats, access, personal data, incidents and mitigations."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Responsible AI and accessibility leads",
"big": false
},
{
"t": "small",
"v": "Challenge fairness, human control, inclusion and public impact."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Learning and change lead",
"big": false
},
{
"t": "p",
"v": "Delivers induction, job aids, competence assessment and adoption support."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Independent assurance lead",
"big": false
},
{
"t": "small",
"v": "Plans and conducts evidence-based challenge independent of delivery."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Evaluation lead",
"big": false
},
{
"t": "small",
"v": "Owns baseline, methods, synthesis, limitations and options appraisal."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Community reviewers",
"big": false
},
{
"t": "small",
"v": "Provide targeted architecture, service, audit or policy challenge."
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Toolkit product owner",
"big": false
},
{
"t": "p",
"v": "Tests appetite and prepares the permanent-toolkit option if warranted."
}
],
"box": true
}
],
"box": false
}
]
},
{
"id": "pilotassurance",
"part": "live-pilot",
"title": "Assurance model",
"icon": "verified_user",
"eyebrow": "Live pilot · formal assurance model",
"heading": "Five assurance stages, five kinds of opinion",
"lede": "Each stage states what it is for, what evidence it reviews, and the opinions it can legitimately reach.",
"body": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Readiness"
},
{
"t": "small",
"v": "Determine whether controlled deployment can begin."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "Evidence: Scope, owners, risk, authority, data, tests, fallback, training."
},
{
"t": "title",
"v": "Opinion: Ready; ready with conditions; not ready.",
"big": false
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "In-flight"
},
{
"t": "small",
"v": "Test whether controls operate and the playbook supports decisions."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "Evidence: Logs, samples, incidents, exceptions, user evidence, decisions."
},
{
"t": "title",
"v": "Opinion: Effective; partially effective; ineffective.",
"big": false
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Change"
},
{
"t": "small",
"v": "Assess material changes to model, tools, data, purpose or service."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "Evidence: Impact assessment, regression tests, permissions and updated record."
},
{
"t": "title",
"v": "Opinion: Authorise; condition; retest; reject.",
"big": false
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Close-out"
},
{
"t": "small",
"v": "Judge whether claims and conclusions are supported."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "Evidence: Outcome evidence, costs, findings, limitations and participant testimony."
},
{
"t": "title",
"v": "Opinion: Reasonable; qualified; insufficient assurance.",
"big": false
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Investment"
},
{
"t": "small",
"v": "Assess whether permanence is justified."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "Evidence: Need, repeat demand, value, maintainability, ownership and funding."
},
{
"t": "title",
"v": "Opinion: Stop; static guide; simplify; interactive toolkit.",
"big": false
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Independence rule"
},
{
"t": "p",
"v": "Delivery teams own controls and evidence. Independent assurance challenges claims and reports limitations. The SRO makes the accountable decision; assurance does not transfer accountability."
}
]
}
]
},
{
"id": "pilotjourney",
"part": "live-pilot",
"title": "Journey & ceremonies",
"icon": "timeline",
"eyebrow": "Live pilot · journey and rhythm",
"heading": "End-to-end pilot journey",
"lede": "Ten steps from mobilise to share, each with its activities and the artefact it must produce.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Mobilise"
},
{
"t": "p",
"v": "Approve charter; appoint roles; agree scope, ethics, data and communications; define hypotheses and baselines."
},
{
"t": "title",
"v": "Pilot charter, RACI, evaluation protocol, communication calendar.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Select"
},
{
"t": "p",
"v": "Screen candidates; interview owners; classify risk and learning value; confirm evidence access."
},
{
"t": "title",
"v": "Candidate scorecard, selection decision and conditions.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Induct"
},
{
"t": "p",
"v": "Brief all roles; assess learning needs; issue playbook routes and job aids; rehearse escalation."
},
{
"t": "title",
"v": "Induction record and competence baseline.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Baseline"
},
{
"t": "p",
"v": "Capture current governance journey, effort, uncertainty, rework, readiness and expected outcomes."
},
{
"t": "title",
"v": "Baseline map and measure definitions.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Apply"
},
{
"t": "p",
"v": "Use playbook lifecycle, authority model, internal ATRS record, evidence pack and decision gates."
},
{
"t": "title",
"v": "Completed artefacts and traceable decisions.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Deploy"
},
{
"t": "p",
"v": "Begin bounded live or limited-live operation; monitor; sample outputs; record exceptions and effort."
},
{
"t": "title",
"v": "Operational evidence, incidents, feedback and change log.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Challenge"
},
{
"t": "p",
"v": "Run architecture, service, security, privacy, responsible-AI, accessibility and audit reviews."
},
{
"t": "title",
"v": "Formal findings and closure plan.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evaluate"
},
{
"t": "p",
"v": "Compare baseline; triangulate metrics, observation and interviews; document limitations."
},
{
"t": "title",
"v": "Evaluation report and playbook backlog.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Decide"
},
{
"t": "p",
"v": "Options appraisal and sponsor decision: stop, static, simplify or toolkit."
},
{
"t": "title",
"v": "Decision record, investment recommendation and roadmap.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Share"
},
{
"t": "p",
"v": "Return findings to participants and communities; publish only what is approved and evidenced."
},
{
"t": "title",
"v": "Lessons pack and controlled reusable assets.",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Ceremonies and evidence rhythm"
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Candidate interview"
},
{
"t": "small",
"v": "Pilot, deployment and assurance leads."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Test eligibility, constraints and evidence access."
},
{
"t": "title",
"v": "Selection record.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Kick-off and induction"
},
{
"t": "small",
"v": "All core roles."
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Align purpose, boundaries, responsibilities and stop routes."
},
{
"t": "title",
"v": "Signed charter and readiness actions.",
"big": false
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Weekly evidence clinic"
},
{
"t": "small",
"v": "Pilot manager, product, operations, evaluation."
}
],
"box": false
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Review decisions, friction, incidents, metrics and learning."
},
{
"t": "title",
"v": "Evidence log and backlog.",
"big": false
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Fortnightly assurance checkpoint"
},
{
"t": "small",
"v": "SRO, service owner, assurance specialists."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Test controls and conditions; decide remedial action."
},
{
"t": "title",
"v": "Assurance note and decision.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Participant pulse"
},
{
"t": "small",
"v": "Users and delivery team."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Capture usability and burden close to the work."
},
{
"t": "title",
"v": "Short structured feedback.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Mid-point challenge"
},
{
"t": "small",
"v": "Independent reviewers and community representatives."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Identify blind spots before conclusions harden."
},
{
"t": "title",
"v": "Challenge findings.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Close-out review"
},
{
"t": "small",
"v": "All accountable and assurance roles."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Agree evidence, limitations and recommendations."
},
{
"t": "title",
"v": "Final assurance opinion.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Investment gate"
},
{
"t": "small",
"v": "Sponsor, SRO, toolkit owner, finance and governance."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Choose future option."
},
{
"t": "title",
"v": "Investment decision.",
"big": false
}
]
}
]
}
]
},
{
"id": "pilotcomms",
"part": "live-pilot",
"title": "Communications",
"icon": "campaign",
"eyebrow": "Live pilot · communications plan",
"heading": "Nine moments, one honest message",
"lede": "Each moment names its audience, message, channel and owner. The framework is being assessed — not the teams taking part.",
"body": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Pre-launch"
},
{
"t": "title",
"v": "Sponsors and governance",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Why this is a live deployment test; decision sought; safeguards."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Executive brief and sponsor conversation."
},
{
"t": "title",
"v": "Pilot sponsor.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Recruitment"
},
{
"t": "title",
"v": "Candidate teams",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "The framework is being assessed, not the team; eligibility and contribution."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Teams post, briefing session and candidate pack."
},
{
"t": "title",
"v": "Pilot manager.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Selection"
},
{
"t": "title",
"v": "Applicants",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Transparent criteria, result, conditions and alternative engagement."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Decision note and direct conversation."
},
{
"t": "title",
"v": "Pilot SRO.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Induction"
},
{
"t": "title",
"v": "Selected participants",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Process, roles, boundaries, evidence, learning and escalation."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Workshop, handbook and role cards."
},
{
"t": "title",
"v": "Learning lead.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "In-flight"
},
{
"t": "title",
"v": "Participants and sponsors",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Progress, decisions, risks, changes and support."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Weekly digest and monthly dashboard."
},
{
"t": "title",
"v": "Pilot manager.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Challenge"
},
{
"t": "title",
"v": "Assurance and community",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Focused review question, evidence boundaries and response route."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Review pack and facilitated session."
},
{
"t": "title",
"v": "Assurance lead.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Close-out"
},
{
"t": "title",
"v": "All participants",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "What evidence shows, limitations and what will change."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Playback and findings report."
},
{
"t": "title",
"v": "Evaluation lead.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Investment"
},
{
"t": "title",
"v": "Senior decision-makers",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Options, value, demand, costs, ownership and recommendation."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Decision paper and deck."
},
{
"t": "title",
"v": "Pilot SRO.",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Wider sharing"
},
{
"t": "title",
"v": "Interested communities",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Verified lessons, reusable patterns and next decision."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Community session and controlled publication."
},
{
"t": "title",
"v": "Communications owner.",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Core messages"
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "campaign"
},
{
"t": "small",
"v": "This is a controlled test of the framework in real delivery."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "campaign"
},
{
"t": "small",
"v": "Participation is not an audit of team performance."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "campaign"
},
{
"t": "p",
"v": "Safety, service obligations and accountable decisions override pilot learning."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "campaign"
},
{
"t": "p",
"v": "Evidence includes what fails, adds burden or duplicates existing controls."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "campaign"
},
{
"t": "p",
"v": "No decision to establish a permanent toolkit has been made in advance."
}
]
}
]
},
{
"id": "pilotsuccess",
"part": "live-pilot",
"title": "Success & feedback",
"icon": "insights",
"eyebrow": "Live pilot · success framework and feedback",
"heading": "Twelve measures, no invented thresholds",
"lede": "Each measure names its evidence, owner and cadence. Thresholds are agreed after the baseline is captured: a pre-set percentage without context would create false precision.",
"body": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Independent use"
},
{
"t": "p",
"v": "Participants navigate lifecycle and required artefacts without author dependency."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Observed task completion; help requests; navigation errors."
},
{
"t": "title",
"v": "Evaluation lead.",
"big": false
},
{
"t": "small",
"v": "Baseline and end."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Proportionality"
},
{
"t": "p",
"v": "Teams and reviewers agree controls match material risk and consequence."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "small",
"v": "Effort diary; duplicated controls; exceptions; reviewer judgement."
},
{
"t": "title",
"v": "Assurance lead.",
"big": false
},
{
"t": "small",
"v": "Weekly and close-out."
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Decision traceability"
},
{
"t": "p",
"v": "Material decisions identify owner, evidence, uncertainty, conditions and review."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Decision-record sample."
},
{
"t": "title",
"v": "Pilot manager.",
"big": false
},
{
"t": "small",
"v": "Fortnightly."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Early issue detection"
},
{
"t": "small",
"v": "Material gaps are discovered while action remains possible."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Issue chronology and avoided or reduced rework evidence."
},
{
"t": "title",
"v": "Service owner.",
"big": false
},
{
"t": "small",
"v": "Monthly."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance quality"
},
{
"t": "p",
"v": "Findings distinguish intent, design, implementation, operation and outcome."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Independent assurance quality review."
},
{
"t": "title",
"v": "Assurance lead.",
"big": false
},
{
"t": "small",
"v": "Mid-point and close."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Operational readiness"
},
{
"t": "p",
"v": "Support, monitoring, fallback and incident routes work in exercises and operation."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Readiness checks, exercises and operational records."
},
{
"t": "title",
"v": "Operations lead.",
"big": false
},
{
"t": "small",
"v": "Pre-live and live."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Learning efficacy"
},
{
"t": "small",
"v": "Participants demonstrate relevant competence in realistic tasks."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Applied assessment and coaching records."
},
{
"t": "title",
"v": "Learning lead.",
"big": false
},
{
"t": "small",
"v": "Before and after."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "User and service value"
},
{
"t": "p",
"v": "Observed service outcomes improve or credible learning justifies continuation."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Service-specific baseline and evaluation."
},
{
"t": "title",
"v": "Product manager.",
"big": false
},
{
"t": "small",
"v": "Agreed cadence."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Burden and cost"
},
{
"t": "p",
"v": "Implementation effort and recurring cost are visible and acceptable relative to value."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Time, support and consumption data."
},
{
"t": "title",
"v": "Evaluation and finance.",
"big": false
},
{
"t": "small",
"v": "Monthly."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Demand and appetite"
},
{
"t": "p",
"v": "Leaders and teams prefer a maintained option and commit use cases, ownership or resources."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Structured option choice and commitments."
},
{
"t": "title",
"v": "Toolkit product owner.",
"big": false
},
{
"t": "small",
"v": "Close-out."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence integrity"
},
{
"t": "p",
"v": "Claims separate verified results, internal evidence, assumptions and expected outcomes."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Evidence audit."
},
{
"t": "title",
"v": "Independent assurance.",
"big": false
},
{
"t": "small",
"v": "Close-out."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Inclusion"
},
{
"t": "p",
"v": "Research and controls address accessibility and affected-user differences."
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Accessibility and inclusion evidence."
},
{
"t": "title",
"v": "Accessibility lead.",
"big": false
},
{
"t": "small",
"v": "Design and close-out."
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Feedback as formal evidence"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Observed use",
"big": false
},
{
"t": "small",
"v": "Whether participants can apply the guide in context."
},
{
"t": "small",
"v": "At key lifecycle tasks."
},
{
"t": "small",
"v": "Observe the framework, not individual performance."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Decision sampling",
"big": false
},
{
"t": "small",
"v": "Whether governance claims are traceable."
},
{
"t": "small",
"v": "Throughout deployment."
},
{
"t": "small",
"v": "Risk-based sample agreed in advance."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Effort diary",
"big": false
},
{
"t": "small",
"v": "Where burden, duplication and rework occur."
},
{
"t": "small",
"v": "Weekly."
},
{
"t": "small",
"v": "Record estimates honestly and label them."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Structured interviews",
"big": false
},
{
"t": "small",
"v": "What helped, hindered or remained missing."
},
{
"t": "small",
"v": "Baseline, mid-point, close."
},
{
"t": "small",
"v": "Separate experience from factual claims."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Tabletop scenarios",
"big": false
},
{
"t": "small",
"v": "Whether controls work under failure and uncertainty."
},
{
"t": "small",
"v": "Pre-live and mid-point."
},
{
"t": "small",
"v": "Use synthetic scenarios where appropriate."
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Assassin review",
"big": false
},
{
"t": "small",
"v": "Where the model could fail, be gamed or become bureaucratic."
},
{
"t": "small",
"v": "Mid-point."
},
{
"t": "small",
"v": "Facilitated, blameless challenge."
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Independent evidence review",
"big": false
},
{
"t": "small",
"v": "Whether conclusions follow from evidence."
},
{
"t": "small",
"v": "Close-out."
},
{
"t": "small",
"v": "Declare conflicts and evidence limitations."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Option-preference survey",
"big": false
},
{
"t": "small",
"v": "Whether appetite exists for future forms."
},
{
"t": "small",
"v": "Close-out."
},
{
"t": "small",
"v": "Require rationale and commitment, not preference alone."
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Community consultation",
"big": false
},
{
"t": "small",
"v": "Whether patterns transfer beyond the pilot."
},
{
"t": "small",
"v": "After initial evidence."
},
{
"t": "small",
"v": "Do not generalise from advocacy alone."
}
]
},
{
"t": "h",
"l": 2,
"v": "Feedback taxonomy"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "KEEP"
},
{
"t": "small",
"v": "Works and should remain."
},
{
"t": "title",
"v": "Protect and evidence.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "IMPROVE"
},
{
"t": "small",
"v": "Useful but unclear or incomplete."
},
{
"t": "title",
"v": "Refine and retest.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "REMOVE"
},
{
"t": "small",
"v": "Adds unjustified burden or duplicates controls."
},
{
"t": "title",
"v": "Remove or explain exception.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "MISSING"
},
{
"t": "small",
"v": "A necessary need or control is absent."
},
{
"t": "title",
"v": "Add, assign owner, validate.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "CONFLICT"
},
{
"t": "small",
"v": "Contradicts policy, service or assurance practice."
},
{
"t": "title",
"v": "Escalate and resolve.",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "LOCALISE"
},
{
"t": "small",
"v": "Valid but needs context-specific adaptation."
},
{
"t": "title",
"v": "Create configurable pattern.",
"big": false
}
]
}
],
"box": false
}
]
},
{
"id": "pilotdecision",
"part": "live-pilot",
"title": "Decision & toolkit",
"icon": "gavel",
"eyebrow": "Live pilot · decision and future state",
"heading": "What happens after the pilot",
"lede": "Four options, the conditions that select each, and what each commits the organisation to. No option is pre-selected.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Stop"
},
{
"t": "p",
"v": "Value is weak, burden is excessive, risks are unresolved or demand is absent."
},
{
"t": "small",
"v": "Archive evidence and retain lessons; do not operationalise."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Retain static playbook"
},
{
"t": "p",
"v": "The guide is useful as reference but repeat operational demand is limited."
},
{
"t": "small",
"v": "Assign light-touch ownership and scheduled review."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Simplify and retest"
},
{
"t": "small",
"v": "Value exists but usability or proportionality is inadequate."
},
{
"t": "small",
"v": "Prioritise critical changes and conduct focused retest."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Build living toolkit"
},
{
"t": "p",
"v": "Evidence shows repeat need, practical value, appetite and sustainable ownership."
},
{
"t": "small",
"v": "Fund discovery and product development for an ongoing service."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Investment evidence required for option D"
},
{
"t": "p",
"v": "Multiple committed user groups or deployments, not general interest alone."
},
{
"t": "p",
"v": "A named product or service owner and multidisciplinary maintenance capacity."
},
{
"t": "small",
"v": "A sustainable source of policy, standards and pattern updates."
},
{
"t": "p",
"v": "Clear content governance, versioning, supersession and retirement rules."
},
{
"t": "p",
"v": "Evidence that dynamic interaction adds value beyond a downloadable guide."
},
{
"t": "p",
"v": "Security, privacy, accessibility, information-management and operating-cost assessment."
},
{
"t": "small",
"v": "A measurable service outcome and roadmap, not only a feature list."
}
]
},
{
"t": "h",
"l": 2,
"v": "Future-state toolkit concept"
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Flagship implementation guide"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Understand the end-to-end route."
},
{
"t": "small",
"v": "Adaptive journey through lifecycle, roles, gates and evidence."
},
{
"t": "small",
"v": "Evidence: Navigation, completion and repeat-use data."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Governance navigator"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Know which controls and forums apply."
},
{
"t": "small",
"v": "Context-led route based on purpose, risk, autonomy and public effect."
},
{
"t": "small",
"v": "Evidence: Routing errors, exceptions and decision consistency."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Internal ATRS workspace"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Create and maintain a transparency record."
},
{
"t": "small",
"v": "Guided record, evidence links, review and publication decision."
},
{
"t": "small",
"v": "Evidence: Field usability, maintenance burden and scope decisions."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance workspace"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Build and review an evidence-based assurance case."
},
{
"t": "small",
"v": "Claims, evidence, findings, conditions and closure."
},
{
"t": "small",
"v": "Evidence: Evidence quality and review efficiency."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Learning academy"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Build competence by responsibility."
},
{
"t": "small",
"v": "Role pathways, scenarios, assessment and job aids."
},
{
"t": "small",
"v": "Evidence: Competence outcomes and support demand."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Pattern library"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Reuse proven controls and artefacts."
},
{
"t": "small",
"v": "Versioned patterns with evidence, owner and applicability."
},
{
"t": "small",
"v": "Evidence: Verified reuse and reduced duplication."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Portfolio control tower"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Oversee value, risk, cost and lifecycle."
},
{
"t": "small",
"v": "Inventory, status, evidence freshness and decision queues."
},
{
"t": "small",
"v": "Evidence: Decision usefulness and data quality."
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Community feedback loop"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "small",
"v": "Keep the toolkit relevant."
},
{
"t": "small",
"v": "Structured proposals, challenge, supersession and release notes."
},
{
"t": "small",
"v": "Evidence: Contribution quality and maintenance capacity."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "Product principle: the implementation guide remains the flagship resource. Interactive functions should reduce cognitive load and help people act safely; they must not hide accountability, evidence or applicable authoritative guidance."
}
]
},
{
"t": "h",
"l": 2,
"v": "Participant checklist"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Before deployment"
},
{
"t": "small",
"v": "Read assigned playbook route."
},
{
"t": "small",
"v": "Complete induction and competence tasks."
},
{
"t": "small",
"v": "Confirm purpose, authority and prohibited use."
},
{
"t": "small",
"v": "Validate data, tools and owners."
},
{
"t": "small",
"v": "Practise fallback and escalation."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "During deployment"
},
{
"t": "small",
"v": "Use approved process and boundaries."
},
{
"t": "small",
"v": "Record material decisions and uncertainty."
},
{
"t": "small",
"v": "Report incidents and friction promptly."
},
{
"t": "small",
"v": "Maintain evidence and effort diary."
},
{
"t": "small",
"v": "Attend evidence clinics and assurance reviews."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "At close-out"
},
{
"t": "small",
"v": "Complete interview and option survey."
},
{
"t": "small",
"v": "Review evidence and factual accuracy."
},
{
"t": "small",
"v": "Declare unresolved concerns."
},
{
"t": "small",
"v": "Identify KEEP, IMPROVE, REMOVE or MISSING."
},
{
"t": "small",
"v": "Confirm future interest and commitment."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Pilot safety stop rules"
},
{
"t": "small",
"v": "Immediate or plausible material harm to users or the public."
},
{
"t": "small",
"v": "Loss of control over permissions, data or state-changing actions."
},
{
"t": "p",
"v": "Critical security, privacy, legal or accessibility condition breached."
},
{
"t": "small",
"v": "Required human oversight or operational fallback unavailable."
},
{
"t": "p",
"v": "Evidence integrity compromised in a way that prevents accountable decision-making."
},
{
"t": "small",
"v": "Deployment moves outside approved purpose or environment."
}
]
},
{
"t": "h",
"l": 2,
"v": "Quick templates"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Candidate screen"
},
{
"t": "small",
"v": "Service and user need"
},
{
"t": "small",
"v": "Agent purpose and prohibited use"
},
{
"t": "small",
"v": "Users and affected parties"
},
{
"t": "small",
"v": "Data, tools and authority"
},
{
"t": "small",
"v": "Expected public-service influence"
},
{
"t": "small",
"v": "Named SRO, service and technical owners"
},
{
"t": "small",
"v": "Safe test environment and fallback"
},
{
"t": "small",
"v": "Evidence access and team capacity"
},
{
"t": "small",
"v": "Why this candidate creates transferable learning"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Weekly evidence log"
},
{
"t": "small",
"v": "Date"
},
{
"t": "small",
"v": "Decision or event"
},
{
"t": "small",
"v": "Evidence"
},
{
"t": "small",
"v": "Effect on delivery"
},
{
"t": "small",
"v": "Playbook feedback"
},
{
"t": "small",
"v": "Owner and action"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Close-out option record"
},
{
"t": "small",
"v": "Option"
},
{
"t": "small",
"v": "Evidence for"
},
{
"t": "small",
"v": "Evidence against"
},
{
"t": "small",
"v": "Conditions"
},
{
"t": "small",
"v": "Decision owner"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Source and status map"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "AI Playbook for the UK Government",
"big": false
},
{
"t": "p",
"v": "Human control, lifecycle management, governance, skills, quality assurance and risk."
},
{
"t": "title",
"v": "Authoritative external grounding.",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Service Standard and Service Manual",
"big": false
},
{
"t": "p",
"v": "User needs, multidisciplinary delivery, security, measures, reliability and improvement."
},
{
"t": "title",
"v": "Applied proportionately as a proposed service lens.",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Technology Code of Practice",
"big": false
},
{
"t": "p",
"v": "Criteria for government technology, including user needs, accessibility, security, privacy, data and Service Standard alignment."
},
{
"t": "title",
"v": "Authoritative external grounding.",
"big": false
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "ATRS guidance and scope policy",
"big": false
},
{
"t": "small",
"v": "Transparency purpose, scope tests and publication route."
},
{
"t": "title",
"v": "Official process; internal registry extension is proposed.",
"big": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Responsible AI Toolkit and AI Knowledge Hub",
"big": false
},
{
"t": "p",
"v": "Responsible deployment, assurance, governance and practical resources."
},
{
"t": "title",
"v": "Authoritative and current resource grounding.",
"big": false
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "AI Agents as Digital Workers Playbook",
"big": false
},
{
"t": "p",
"v": "Policy, lifecycle, roles, internal registry, learning and roadmap prototype."
},
{
"t": "title",
"v": "Primary prototype under live test.",
"big": false
}
],
"box": true
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**Final discipline.** Pilot conclusions must distinguish observed and verified results, internally evidenced results, participant views, assumptions, expected outcomes and unresolved unknowns."
}
]
}
]
},
{
"id": "docprd",
"part": "product-documentation",
"title": "Product requirements",
"icon": "assignment",
"eyebrow": "Delivery documentation · PRD",
"heading": "Product requirements",
"lede": "What this guide is for, who it serves, what it must do and what it deliberately does not do. Derived from published guidance and community discussion; a proposal, not an adopted product mandate.",
"body": [
{
"t": "h",
"l": 2,
"v": "Problems worth solving"
},
{
"t": "small",
"v": "Each problem is visible in the source material rather than assumed."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Guidance is real but scattered"
},
{
"t": "p",
"v": "The AI Playbook, Service Standard, TCoP, ATRS, NCSC secure-AI guidelines, ethics frameworks and FinOps practice exist separately. Teams reassemble them from scratch, inconsistently, each time."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Experimentation outruns governance"
},
{
"t": "p",
"v": "Personal assistants become shared agents and then business-critical dependencies without registration, named ownership, assurance or funded support."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Cost is discussed as licences"
},
{
"t": "p",
"v": "Token and licence spend is visible; data remediation, workflow redesign, assurance, human review, monitoring, support and exit are not costed or owned."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence is claimed, not demonstrated"
},
{
"t": "p",
"v": "The same product yields different results under different evaluation methods, so headline productivity figures travel further than the method that produced them."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Community knowledge decays"
},
{
"t": "p",
"v": "Sessions, transcripts, repositories and personal networks hold the learning; there is no shared discovery layer, taxonomy or evidence pack, so lessons are relearned."
}
]
},
{
"t": "h",
"l": 2,
"v": "Goals and how each is judged"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "One route from mandate to governed operation"
},
{
"t": "p",
"v": "A reader can go from no policy position to a registered, assured, monitored agent using only this guide."
},
{
"t": "title",
"v": "All eight start-here stages have a chapter and a tool",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Make the invisible costs visible"
},
{
"t": "small",
"v": "Every cost class in the investment line is named, owned and modelled."
},
{
"t": "title",
"v": "Cost lab produces a per-run and monthly figure with review effort included",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Proportionate assurance, not blanket control"
},
{
"t": "small",
"v": "Criticality drives the gate; low-risk work stays fast."
},
{
"t": "title",
"v": "Five tiers with stated evidence and review frequency",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence before assertion"
},
{
"t": "p",
"v": "Claims carry method, publisher and date; illustrative material is labelled."
},
{
"t": "title",
"v": "Every figure in the guide is attributed or marked illustrative",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Reusable artefacts, not prose"
},
{
"t": "p",
"v": "Register schema, ATRS record, instruction block and evidence card can be copied out and used."
},
{
"t": "title",
"v": "Four artefacts exportable or copyable",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Accessible to every role"
},
{
"t": "p",
"v": "Leaders, engineers, assurance, commercial, finance and frontline staff each have a path."
},
{
"t": "title",
"v": "Role pathways cover the twelve priority roles",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "In scope"
},
{
"t": "small",
"v": "Guidance synthesis and chapter navigation"
},
{
"t": "p",
"v": "Assessment tools (readiness, criticality, tiers, maturity, quality gate)"
},
{
"t": "small",
"v": "Register and ATRS record schemas"
},
{
"t": "small",
"v": "Cost and unit-economics modelling"
},
{
"t": "small",
"v": "Skill instruction builder"
},
{
"t": "small",
"v": "Evidence, use-case and international comparison libraries"
},
{
"t": "small",
"v": "Illustrative portfolio dashboard"
},
{
"t": "small",
"v": "Delivery documentation set (this part)"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Out of scope"
},
{
"t": "small",
"v": "Holding the authoritative record of any live deployment"
},
{
"t": "small",
"v": "Storing personal or OFFICIAL-SENSITIVE data"
},
{
"t": "small",
"v": "Connecting to tenant telemetry, licensing or billing systems"
},
{
"t": "small",
"v": "Issuing approvals or acting as the assurance authority"
},
{
"t": "small",
"v": "Publishing ATRS records on behalf of an organisation"
},
{
"t": "small",
"v": "Replacing or overriding central or departmental guidance"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Non-functional requirements"
},
{
"t": "table",
"rows": [
[
{
"h": false,
"v": "**Accessibility** WCAG 2.2 AA: contrast, focus visibility, keyboard operation, reduced-motion, semantic headings and labels."
},
{
"h": false,
"v": "**Information handling** OFFICIAL working content only. No personal data, no case data, no credentials. Assessment answers stay on the reader's device."
},
{
"h": false,
"v": "**Portability** Runs as a single file in a browser with no build step, no backend and no network dependency beyond fonts and cited links."
}
],
[
{
"h": false,
"v": "**Performance** First paint before the whole document streams; interactions respond within one frame budget on a standard civil-service laptop."
},
{
"h": false,
"v": "**Provenance** Crown-copyright content reused under the Open Government Licence v3.0, with publisher and date on every figure."
},
{
"h": false,
"v": "**Print and export** Chapters print legibly; the register schema and instruction block copy to the clipboard as plain text."
}
]
]
},
{
"t": "h",
"l": 2,
"v": "Release phases"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Now"
},
{
"t": "h",
"l": 3,
"v": "Phase 1 — proposal"
},
{
"t": "p",
"v": "This guide, reviewed by the community, with illustrative figures clearly labelled."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Next"
},
{
"t": "h",
"l": 3,
"v": "Phase 2 — validated"
},
{
"t": "p",
"v": "Illustrative figures replaced with measured departmental data; international benchmarking executed."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Later"
},
{
"t": "h",
"l": 3,
"v": "Phase 3 — adopted"
},
{
"t": "p",
"v": "Referenced in departmental AI policy, with the register schema and assurance gates in operational use."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Risks to the product itself"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Illustrative figures read as measured High",
"big": false
},
{
"t": "p",
"v": "Label every illustrative frame; keep publisher and date on evidence; publication checkpoint before circulation."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Guide treated as authority Medium",
"big": false
},
{
"t": "p",
"v": "Proposal status on the landing chapter and in the hierarchy; explicit statement that it does not override guidance."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Content drifts from source guidance High",
"big": false
},
{
"t": "p",
"v": "Chapter-to-source map; review date per chapter; ticker tracks guidance in force."
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Assessment answers mistaken for assurance Medium",
"big": false
},
{
"t": "p",
"v": "Device-only storage, no scoring authority claimed, evidence-based maturity rather than self-scored levels."
}
],
"box": true
}
]
},
{
"id": "docftir",
"part": "product-documentation",
"title": "Functional & technical spec",
"icon": "checklist",
"eyebrow": "Delivery documentation · FTIR specification",
"heading": "Functional, technical and integration requirements",
"lede": "Numbered requirements with priority, traceable to the chapters that satisfy them. Must, should and could follow the usual prioritisation convention.",
"body": [
{
"t": "h",
"l": 2,
"v": "Functional"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-01",
"big": false
},
{
"t": "p",
"v": "Navigate 69 chapters in eleven numbered parts with contents, previous, next and sibling links"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-02",
"big": false
},
{
"t": "p",
"v": "Search every chapter, principle, gate, register field, role and source from one field"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "FR-03",
"big": false
},
{
"t": "small",
"v": "Record which chapters have been opened and show progress"
},
{
"t": "eyebrow",
"v": "Should"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-04",
"big": false
},
{
"t": "small",
"v": "Rate readiness across eight dimensions and return a banded barometer"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "FR-05",
"big": false
},
{
"t": "small",
"v": "Determine an assurance tier from a two-question intake"
},
{
"t": "eyebrow",
"v": "Must"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-06",
"big": false
},
{
"t": "small",
"v": "Score agent criticality across ten dimensions and derive the gate"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "FR-07",
"big": false
},
{
"t": "small",
"v": "Score barriers by impact and likelihood and band the result"
},
{
"t": "eyebrow",
"v": "Should"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-08",
"big": false
},
{
"t": "small",
"v": "Model cost per run and per month across models, languages and volumes"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-09",
"big": false
},
{
"t": "small",
"v": "Model unit economics: value per interaction, net value and return"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "FR-10",
"big": false
},
{
"t": "small",
"v": "Build a seven-part instruction block and copy it out"
},
{
"t": "eyebrow",
"v": "Must"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-11",
"big": false
},
{
"t": "p",
"v": "Assess evidence strength per capability domain across seven maturity levels"
},
{
"t": "eyebrow",
"v": "Should"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "FR-12",
"big": false
},
{
"t": "small",
"v": "Step a failure and recovery loop, marking stages complete"
},
{
"t": "eyebrow",
"v": "Should"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-13",
"big": false
},
{
"t": "p",
"v": "Present the eighteen-field register schema and the eight-section ATRS record"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "FR-14",
"big": false
},
{
"t": "p",
"v": "Show an illustrative portfolio dashboard with lifecycle, assurance, risk and domain views"
},
{
"t": "eyebrow",
"v": "Could"
}
],
"box": true
},
{
"t": "h",
"l": 2,
"v": "Technical"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "TR-01",
"big": false
},
{
"t": "p",
"v": "Single-file browser document; no build step, bundler or server runtime"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "TR-02",
"big": false
},
{
"t": "p",
"v": "All state in memory; only the readiness barometer persists, in localStorage on the reader's device"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "TR-03",
"big": false
},
{
"t": "small",
"v": "Inline styling so content paints as it streams"
},
{
"t": "eyebrow",
"v": "Must"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "TR-04",
"big": false
},
{
"t": "p",
"v": "Keyboard access for every control; visible focus; reduced-motion honoured"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "TR-05",
"big": false
},
{
"t": "small",
"v": "Contrast at or above 4.5:1 for text and 3:1 for meaningful non-text"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "TR-06",
"big": false
},
{
"t": "small",
"v": "No third-party analytics, cookies or tracking"
},
{
"t": "eyebrow",
"v": "Must"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "TR-07",
"big": false
},
{
"t": "p",
"v": "External links open in a new tab with noopener and carry publisher and date"
},
{
"t": "eyebrow",
"v": "Should"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "TR-08",
"big": false
},
{
"t": "p",
"v": "Deterministic calculations — same inputs, same outputs, no model calls"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "h",
"l": 2,
"v": "Integration"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "IR-01",
"big": false
},
{
"t": "small",
"v": "Register schema aligns field-for-field with the ATRS record sections"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "IR-02",
"big": false
},
{
"t": "p",
"v": "Register, information asset register and data catalogue treated as three integrated registers, not duplicates"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "IR-03",
"big": false
},
{
"t": "p",
"v": "Criticality score maps to the five assurance gates and their evidence requirements"
},
{
"t": "eyebrow",
"v": "Must"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "IR-04",
"big": false
},
{
"t": "p",
"v": "Cost model outputs feed the investment line and FinOps allocation layers"
},
{
"t": "eyebrow",
"v": "Should"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "IR-05",
"big": false
},
{
"t": "small",
"v": "Instruction block output is platform-neutral plain text"
},
{
"t": "eyebrow",
"v": "Must"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "IR-06",
"big": false
},
{
"t": "small",
"v": "Chapters cite the guidance in force and carry a review date"
},
{
"t": "eyebrow",
"v": "Must"
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "IR-07",
"big": false
},
{
"t": "p",
"v": "Links out to the AI Engineering Lab, AI Knowledge Hub and departmental policy rather than copying them"
},
{
"t": "eyebrow",
"v": "Should"
}
],
"box": true
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "title",
"v": "IR-08",
"big": false
},
{
"t": "p",
"v": "Sits alongside the govukOS twin front door and portals as one app in a shared shell"
},
{
"t": "eyebrow",
"v": "Should"
}
],
"box": true
}
]
},
{
"id": "docsad",
"part": "product-documentation",
"title": "Solution architecture",
"icon": "architecture",
"eyebrow": "Delivery documentation · SAD",
"heading": "Solution architecture",
"lede": "Principles, context, layers and the decisions that shaped them. The architecture is deliberately small: the guide holds interpretation, never the record of a live deployment.",
"body": [
{
"t": "h",
"l": 2,
"v": "Architectural principles"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Guidance is referenced, not copied"
},
{
"t": "small",
"v": "Link to the governed source; hold interpretation and sequence here."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "No shadow record"
},
{
"t": "small",
"v": "The guide never becomes the register of record for a live deployment."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Client-side by default"
},
{
"t": "p",
"v": "No backend means no new data controller, no new attack surface, no new service to run."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Accessible before decorative"
},
{
"t": "p",
"v": "Contrast, focus and keyboard behaviour are acceptance criteria, not polish."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence carries method"
},
{
"t": "p",
"v": "Every figure states publisher, date and method, or is labelled illustrative."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Open by default"
},
{
"t": "p",
"v": "Crown-copyright content under OGL v3.0; the guide itself is reusable and adaptable."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Context — actors and their part"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Reader",
"big": false
},
{
"t": "small",
"v": "Civil servant in one of the twelve priority roles"
},
{
"t": "small",
"v": "Reads chapters, runs assessments, copies artefacts"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owning organisation",
"big": false
},
{
"t": "small",
"v": "Department, agency, council or health body"
},
{
"t": "small",
"v": "Holds the register of record, policy and approvals"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Central guidance",
"big": false
},
{
"t": "small",
"v": "GDS, DSIT, NCSC, CDDO successor functions"
},
{
"t": "small",
"v": "Publishes the guidance this guide interprets"
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Community",
"big": false
},
{
"t": "p",
"v": "Cross-government Copilot forum, AI Engineering Lab, ethics and data communities"
},
{
"t": "small",
"v": "Supplies patterns, failures and supplier intelligence"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Suppliers",
"big": false
},
{
"t": "small",
"v": "Platform, model and integration providers"
},
{
"t": "small",
"v": "Subject of dependency, exit and commercial chapters"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "govukOS shell",
"big": false
},
{
"t": "small",
"v": "Twin front door and sibling portals"
},
{
"t": "small",
"v": "Hosts the guide as one app among several"
}
]
},
{
"t": "h",
"l": 2,
"v": "Logical layers"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Experience",
"big": false
},
{
"t": "p",
"v": "Masthead, ticker, drawer, chapter rail, contents, chapter body, footer navigation"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Chapter registry",
"big": false
},
{
"t": "small",
"v": "Parts, chapters, numbering, accents, icons, read state"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Content",
"big": false
},
{
"t": "p",
"v": "Source-derived data: principles, gates, domains, roles, tiers, schemas, dashboards"
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Assessment logic",
"big": false
},
{
"t": "p",
"v": "Barometer, criticality, tiers, maturity evidence, quality gate, barrier scoring"
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Economics",
"big": false
},
{
"t": "small",
"v": "Token maths, language multipliers, unit economics, allocation layers"
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Local state",
"big": false
},
{
"t": "p",
"v": "In-memory interaction state plus one persisted barometer record on the device"
}
],
"box": true
},
{
"t": "h",
"l": 2,
"v": "Decisions"
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "ADR-01"
},
{
"t": "h",
"l": 3,
"v": "Single-file client-side document"
},
{
"t": "p",
"v": "No hosting, no data processing agreement, no service to operate; trades server-side features for portability."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "ADR-02"
},
{
"t": "h",
"l": 3,
"v": "No personal or case data anywhere"
},
{
"t": "p",
"v": "Removes the need for a DPIA on the guide itself; assessments stay on the device."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "ADR-03"
},
{
"t": "h",
"l": 3,
"v": "One artefact, many chapters"
},
{
"t": "p",
"v": "Keeps navigation, search and cross-references coherent; avoids a fragmented document set."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "ADR-04"
},
{
"t": "h",
"l": 3,
"v": "Illustrative data labelled, never silently synthetic"
},
{
"t": "small",
"v": "Protects the evidence boundary this playbook sets out."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "ADR-05"
},
{
"t": "h",
"l": 3,
"v": "Accent colour per part"
},
{
"t": "p",
"v": "Gives readers a positional cue; all accents are contrast-tested as text."
}
]
}
]
},
{
"id": "dochld",
"part": "product-documentation",
"title": "High-level design",
"icon": "schema",
"eyebrow": "Delivery documentation · HLD",
"heading": "High-level design",
"lede": "Components, the entities they work on, and the four flows that carry most of the value.",
"body": [
{
"t": "h",
"l": 2,
"v": "Components"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Shell"
},
{
"t": "small",
"v": "Masthead, guidance ticker, launchpad, search field"
},
{
"t": "title",
"v": "Holds identity and cross-app links",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Navigation"
},
{
"t": "p",
"v": "Drawer with parts and chapters, chapter rail, contents overlay, prev/next"
},
{
"t": "title",
"v": "Owns chapter numbering and read state",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Chapter renderer"
},
{
"t": "small",
"v": "One section per chapter, streamed in order"
},
{
"t": "title",
"v": "Eyebrow, accent title, lede, body blocks",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assessment engine"
},
{
"t": "p",
"v": "Barometer, criticality radar, tier finder, maturity evidence, quality gate"
},
{
"t": "title",
"v": "Pure functions over local state",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Economics engine"
},
{
"t": "small",
"v": "Token cost, language multiplier, unit economics, allocation layers"
},
{
"t": "title",
"v": "Deterministic arithmetic, GBP formatting",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Artefact builders"
},
{
"t": "small",
"v": "Instruction block, register schema, ATRS record, evidence card"
},
{
"t": "title",
"v": "Copy-to-clipboard output",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Search index"
},
{
"t": "small",
"v": "Flat index of chapters, principles, gates, fields, roles, sources"
},
{
"t": "title",
"v": "Substring match with type and destination",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Dashboard"
},
{
"t": "small",
"v": "Illustrative portfolio control tower"
},
{
"t": "title",
"v": "Read-only, clearly labelled illustrative",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Entities"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Chapter",
"big": false
},
{
"t": "small",
"v": "id, part, number, title, icon, accent, read"
},
{
"t": "small",
"v": "Navigation and progress"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Blueprint",
"big": false
},
{
"t": "small",
"v": "title, source, ingest date, chapters it feeds"
},
{
"t": "small",
"v": "Provenance and coverage"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Assessment",
"big": false
},
{
"t": "small",
"v": "dimension, score, band, timestamp"
},
{
"t": "small",
"v": "Barometer and maturity evidence"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Register record",
"big": false
},
{
"t": "small",
"v": "18 fields from identity to retirement"
},
{
"t": "small",
"v": "The record an organisation holds"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "ATRS record",
"big": false
},
{
"t": "small",
"v": "8 sections, each with its own fields"
},
{
"t": "small",
"v": "Transparency record per agent"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Cost scenario",
"big": false
},
{
"t": "small",
"v": "tokens in/out, prices, tools, infra, review effort, volume"
},
{
"t": "small",
"v": "Cost and unit economics"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Instruction block",
"big": false
},
{
"t": "small",
"v": "goal, context, sources, task, format, guardrails, output"
},
{
"t": "small",
"v": "Skill injection"
}
]
},
{
"t": "h",
"l": 2,
"v": "Key flows"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "p",
"v": "**Read a chapter** Drawer or contents → chapter renders → read state recorded → next chapter offered"
},
{
"t": "p",
"v": "**Assess readiness** Rate eight dimensions → band computed → stored on device → shown in the drawer workspace"
},
{
"t": "p",
"v": "**Tier an agent** Two intake questions → tier → gate, controls, evidence and review frequency shown"
},
{
"t": "p",
"v": "**Price a job** Enter words in and out, model, language, volume → per-run and monthly cost → unit economics"
}
],
"box": false
}
]
},
{
"id": "doclld",
"part": "product-documentation",
"title": "Low-level design",
"icon": "code",
"eyebrow": "Delivery documentation · LLD",
"heading": "Low-level design",
"lede": "Module responsibilities, the arithmetic behind every score, the validation rules and the accessibility implementation. Enough for someone else to rebuild the behaviour exactly.",
"body": [
{
"t": "h",
"l": 2,
"v": "Modules and their state"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Navigation module",
"big": false
},
{
"t": "p",
"v": "groupDefs → flat chapter list → numbering, accents, prev/next, contents"
},
{
"t": "small",
"v": "section, groups, read, toc"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Barometer module",
"big": false
},
{
"t": "small",
"v": "eight dimension scores → mean → band with colour and label"
},
{
"t": "small",
"v": "scores (persisted)"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Criticality module",
"big": false
},
{
"t": "small",
"v": "ten dimension scores → total → tier and gate mapping"
},
{
"t": "small",
"v": "crit, gateTier"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Barrier module",
"big": false
},
{
"t": "small",
"v": "impact × likelihood → 1–25 → four bands"
},
{
"t": "small",
"v": "bar"
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Economics module",
"big": false
},
{
"t": "p",
"v": "tokens = words ÷ 0.75; cost = in×priceIn + out×priceOut + tools + infra + review"
},
{
"t": "small",
"v": "lab, ue"
}
],
"box": true
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Instruction module",
"big": false
},
{
"t": "small",
"v": "seven fields → composed block → clipboard"
},
{
"t": "small",
"v": "blk, copied"
}
]
},
{
"t": "h",
"l": 2,
"v": "Algorithms"
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Barometer banding"
},
{
"t": "p",
"v": "mean of rated dimensions; 0–0.9 Nascent, 1.0–1.9 Emerging, 2.0–2.4 Establishing, 2.5–3.0 Leading; unrated dimensions excluded from the mean and reported as unrated."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Criticality to gate"
},
{
"t": "p",
"v": "sum of ten dimensions scored 1–5; 10–20 personal, 21–30 team, 31–38 departmental, 39–44 operational service, 45–50 citizen impact."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Barrier banding"
},
{
"t": "p",
"v": "impact × likelihood on 1–5 scales; 1–3 low, 4–8 medium, 9–15 high, 16–25 critical."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Unit economics"
},
{
"t": "p",
"v": "value per interaction = (minutes saved ÷ 60 × hourly rate) × quality factor; net = value − cost; return = net ÷ cost expressed as a percentage."
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "Validation rules"
},
{
"t": "p",
"v": "**Numeric inputs** Clamp to declared minimum and maximum; reject non-numeric silently rather than throwing."
},
{
"t": "p",
"v": "**Unrated dimensions** Show an em dash, never a zero, and exclude from means."
},
{
"t": "p",
"v": "**Currency** Two decimal places, pound sign, thousands separated; never rounded away in totals."
},
{
"t": "p",
"v": "**Clipboard** Plain text only; confirm with a transient state, no alert."
},
{
"t": "p",
"v": "**Persistence** One key, versioned; parse defensively and fall back to blank on any error."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "Accessibility implementation"
},
{
"t": "p",
"v": "**Colour** Every text and meaningful non-text pair contrast-tested; accents have separate text-safe variants where a fill carries text."
},
{
"t": "p",
"v": "**Keyboard** All controls are buttons or inputs; aria-pressed and aria-current reflect state; focus ring never suppressed."
},
{
"t": "p",
"v": "**Motion** The guidance ticker pauses on hover and respects prefers-reduced-motion."
},
{
"t": "p",
"v": "**Structure** One h1 per chapter, sequential headings, aria-labels on sections, skip link to content."
},
{
"t": "p",
"v": "**Language** British English, plain terms, expanded acronyms on first use in a chapter."
}
]
}
],
"box": false
}
]
},
{
"id": "doctoc",
"part": "product-documentation",
"title": "Theory of change",
"icon": "account_tree",
"eyebrow": "Delivery documentation · theory of change",
"heading": "Theory of change",
"lede": "From inputs to impact, with the assumptions each step rests on. Written so the chain can be falsified: if an assumption fails, the link it supports fails with it.",
"body": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Inputs"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Central guidance in force",
"big": false
},
{
"t": "title",
"v": "Community evidence and patterns",
"big": false
},
{
"t": "title",
"v": "Evaluation studies",
"big": false
},
{
"t": "title",
"v": "Departmental policy and delegated authority",
"big": false
},
{
"t": "title",
"v": "Funded time from named roles",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Activities"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Synthesise guidance into sequenced chapters",
"big": false
},
{
"t": "title",
"v": "Build assessment and costing tools",
"big": false
},
{
"t": "title",
"v": "Publish register and ATRS schemas",
"big": false
},
{
"t": "title",
"v": "Run role-based capability pathways",
"big": false
},
{
"t": "title",
"v": "Return findings to the community",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Outputs"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "One navigable implementation guide",
"big": false
},
{
"t": "title",
"v": "Completed readiness and criticality assessments",
"big": false
},
{
"t": "title",
"v": "Registered agents with named owners",
"big": false
},
{
"t": "title",
"v": "Costed business cases including hidden costs",
"big": false
},
{
"t": "title",
"v": "Reusable evidence cards and patterns",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Short-term outcomes"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Teams find the applicable guidance quickly",
"big": false
},
{
"t": "title",
"v": "Proportionate assurance replaces blanket caution",
"big": false
},
{
"t": "title",
"v": "Agents have owners, tiers and review dates",
"big": false
},
{
"t": "title",
"v": "Cost conversations include people and exit",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Medium-term outcomes"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Fewer orphaned and unregistered agents",
"big": false
},
{
"t": "title",
"v": "Duplicated procurement and assurance effort falls",
"big": false
},
{
"t": "title",
"v": "Evidence is reused across organisations",
"big": false
},
{
"t": "title",
"v": "Oversight is continuous, not one-off",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Impact"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Governed AI-enabled services that create demonstrable public value",
"big": false
},
{
"t": "title",
"v": "Sustained public trust in how government uses AI",
"big": false
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Each row depends on the one above it holding. Outcomes are claimed only where evidence exists; everything below outputs is currently a proposal.",
"big": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "Assumptions"
},
{
"t": "p",
"v": "**Guidance stays reasonably stable** Chapters carry review dates so drift is visible rather than silent."
},
{
"t": "p",
"v": "**Teams have protected time** Without funded time, the guide is read but not applied."
},
{
"t": "p",
"v": "**Leaders accept proportionality** Blanket caution defeats the tiering model as surely as no control at all."
},
{
"t": "p",
"v": "**Community keeps sharing** Reuse depends on organisations publishing failures as well as successes."
},
{
"t": "p",
"v": "**Illustrative figures get replaced** Measured departmental data must supersede the labelled placeholders."
}
]
},
{
"t": "group",
"kids": [
{
"t": "h",
"l": 2,
"v": "Indicators to track"
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "small",
"v": "Chapters opened per role pathway"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "small",
"v": "Agents registered with a named owner and review date"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "small",
"v": "Proportion of agents with an ATRS record"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "small",
"v": "Assurance decisions made at the tier they belong to"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "small",
"v": "Business cases including the full investment line"
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "small",
"v": "Patterns and evidence cards reused across organisations"
}
]
}
]
}
],
"box": false
}
]
},
{
"id": "docresearch",
"part": "product-documentation",
"title": "Personas & journeys",
"icon": "groups",
"eyebrow": "Delivery documentation · user research bundle",
"heading": "Personas, stories, journeys and service blueprints",
"lede": "Drawn from the roles, frictions and lifecycle material in published guidance and community discussion. These are evidence-derived working personas, not research participants — they need validating with real users before they carry design decisions.",
"body": [
{
"t": "h",
"l": 2,
"v": "Six personas"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Accountable leader"
},
{
"t": "title",
"v": "Director or SRO, accounting-officer line",
"big": false
},
{
"t": "p",
"v": "Decide where AI is worth the risk and the money, and be able to answer for it."
},
{
"t": "eyebrow",
"v": "Needs"
},
{
"t": "small",
"v": "Evidence that a claim is measured, not asserted"
},
{
"t": "small",
"v": "A defensible risk appetite"
},
{
"t": "small",
"v": "One page that survives a select-committee question"
},
{
"t": "eyebrow",
"v": "Frustrations"
},
{
"t": "small",
"v": "Productivity figures with no method attached"
},
{
"t": "small",
"v": "Being asked to approve what nobody owns"
},
{
"t": "small",
"v": "Cost framed as licences only"
},
{
"t": "title",
"v": "Reads: Assurance tiers, evidence chapter, investment line, control tower",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Service owner"
},
{
"t": "title",
"v": "Owns an operational service with real users",
"big": false
},
{
"t": "small",
"v": "Adopt agents without destabilising a service people depend on."
},
{
"t": "eyebrow",
"v": "Needs"
},
{
"t": "small",
"v": "Clear promotion gates"
},
{
"t": "small",
"v": "Named ownership and support model"
},
{
"t": "small",
"v": "A rollback and exit route"
},
{
"t": "eyebrow",
"v": "Frustrations"
},
{
"t": "small",
"v": "Agents appearing in the service without registration"
},
{
"t": "small",
"v": "Unclear escalation when output is wrong"
},
{
"t": "small",
"v": "Support cost landing on the service unfunded"
},
{
"t": "title",
"v": "Reads: Agent journey, criticality and gates, failure and recovery, dependency and exit",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI engineer"
},
{
"t": "title",
"v": "Builds and evaluates agents",
"big": false
},
{
"t": "small",
"v": "Build well-grounded agents and prove they work."
},
{
"t": "eyebrow",
"v": "Needs"
},
{
"t": "small",
"v": "Evaluation and red-team patterns"
},
{
"t": "small",
"v": "Grounding decision tree"
},
{
"t": "small",
"v": "Instruction blocks that transfer between platforms"
},
{
"t": "eyebrow",
"v": "Frustrations"
},
{
"t": "small",
"v": "Ungoverned copies of source data made for grounding"
},
{
"t": "small",
"v": "No baseline to evaluate drift against"
},
{
"t": "small",
"v": "Prototype standards applied to production, and the reverse"
},
{
"t": "title",
"v": "Reads: Skill studio, architecture and quality, knowledge foundation, AI Engineering Lab",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Assurance and ethics specialist"
},
{
"t": "title",
"v": "Reviews, challenges and signs off",
"big": false
},
{
"t": "small",
"v": "Apply proportionate control and see the evidence behind it."
},
{
"t": "eyebrow",
"v": "Needs"
},
{
"t": "small",
"v": "Criticality scoring that distinguishes usage from consequence"
},
{
"t": "small",
"v": "ATRS record completeness"
},
{
"t": "small",
"v": "A residual-risk statement per agent"
},
{
"t": "eyebrow",
"v": "Frustrations"
},
{
"t": "small",
"v": "Assurance as a gate at the end rather than a thread throughout"
},
{
"t": "small",
"v": "Self-scored maturity with no evidence"
},
{
"t": "small",
"v": "Incidents surfacing without a route to redress"
},
{
"t": "title",
"v": "Reads: Risks and mitigations, internal ATRS registry, maturity evidence, autonomy boundaries",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Commercial and FinOps lead"
},
{
"t": "title",
"v": "Buys, allocates and forecasts",
"big": false
},
{
"t": "small",
"v": "Fund adoption honestly and keep exit possible."
},
{
"t": "eyebrow",
"v": "Needs"
},
{
"t": "small",
"v": "Unit economics per use case"
},
{
"t": "small",
"v": "Allocation model across six cost layers"
},
{
"t": "small",
"v": "Supplier requirements for telemetry, portability and exit"
},
{
"t": "eyebrow",
"v": "Frustrations"
},
{
"t": "small",
"v": "Token cost mistaken for total cost"
},
{
"t": "small",
"v": "Lock-in discovered at renewal"
},
{
"t": "small",
"v": "Shared costs with no agreed allocation"
},
{
"t": "title",
"v": "Reads: Cost value and investment, commercial and suppliers, FinOps, dependency and exit",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Frontline practitioner"
},
{
"t": "title",
"v": "Uses agents in daily casework",
"big": false
},
{
"t": "small",
"v": "Do the work faster without carrying unowned risk."
},
{
"t": "eyebrow",
"v": "Needs"
},
{
"t": "small",
"v": "Plain guidance on what is allowed"
},
{
"t": "small",
"v": "A way to challenge an output"
},
{
"t": "small",
"v": "Training in the flow of work"
},
{
"t": "eyebrow",
"v": "Frustrations"
},
{
"t": "small",
"v": "Guidance written for architects"
},
{
"t": "small",
"v": "No clear escalation when the agent is wrong"
},
{
"t": "small",
"v": "Being blamed for a tool decision made elsewhere"
},
{
"t": "title",
"v": "Reads: Support around every colleague, inclusive adoption, roles and pathways",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "User stories with acceptance criteria"
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Accountable leader**, I need to see the method behind any productivity claim, so that I do not repeat an unevidenced figure in public."
},
{
"t": "p",
"v": "Accepted when: Every figure shows publisher, date and method, or is labelled illustrative"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Accountable leader**, I need to know which agents are high risk and who owns them, so that I can direct attention where consequence is highest."
},
{
"t": "p",
"v": "Accepted when: Control tower lists risk level, owner and next review per agent"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Service owner**, I need to know the gate an agent must pass before it touches my service, so that I can plan assurance rather than discover it."
},
{
"t": "p",
"v": "Accepted when: Criticality score returns a tier with named evidence and review frequency"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Service owner**, I need to have an exit path for a platform or model, so that a supplier change does not stop my service."
},
{
"t": "p",
"v": "Accepted when: Seven-step exit path with the dependencies it addresses"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a AI engineer**, I need to decide where grounding information should live, so that I do not create ungoverned copies."
},
{
"t": "p",
"v": "Accepted when: Decision tree returns one of four options in priority order"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a AI engineer**, I need to compose an instruction block once and reuse it, so that behaviour is consistent across platforms."
},
{
"t": "small",
"v": "Accepted when: Seven-part block copies out as plain text"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Assurance specialist**, I need to score criticality across ten dimensions, so that control is proportionate to consequence."
},
{
"t": "small",
"v": "Accepted when: Score maps to one of five gates with its controls"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Assurance specialist**, I need to record transparency information for every internal agent, so that oversight and audit have something to inspect."
},
{
"t": "small",
"v": "Accepted when: Eight-section ATRS record with all fields listed"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a FinOps lead**, I need to model cost per run and per month, so that a business case includes consumption, not guesses."
},
{
"t": "p",
"v": "Accepted when: Cost lab returns per-run and monthly figures with a language multiplier"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a FinOps lead**, I need to see value against cost per interaction, so that I can show whether the agent earns its keep."
},
{
"t": "small",
"v": "Accepted when: Unit economics returns value, net and return"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Frontline practitioner**, I need to know what I may use and when to escalate, so that I can work confidently within the guardrails."
},
{
"t": "p",
"v": "Accepted when: Support chapter states allowed use, escalation and where help comes from"
}
]
},
{
"t": "card",
"kids": [
{
"t": "p",
"v": "**As a Community manager**, I need to turn a session into a reusable evidence pack, so that learning survives the meeting."
},
{
"t": "p",
"v": "Accepted when: Evidence pack structure with owner, taxonomy and review date"
}
]
},
{
"t": "h",
"l": 2,
"v": "Journeys"
},
{
"t": "h",
"l": 3,
"v": "First-time reader to first decision"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Arrives with a question** Lands on the overview, sees proposal status “Is this authoritative?” **Proposal flag and hierarchy answer it in one screen**"
},
{
"t": "p",
"v": "**Finds the applicable chapter** Uses search or contents “Where does my case sit?” **Chapter numbering and part accents orient quickly**"
},
{
"t": "p",
"v": "**Runs an assessment** Tier finder or criticality radar “Am I over- or under-controlling?” **Tier arrives with evidence and review expectations**"
},
{
"t": "p",
"v": "**Takes an artefact away** Copies the register schema or instruction block “What do I do on Monday?” **Copyable artefacts, not prose**"
},
{
"t": "p",
"v": "**Returns for oversight** Re-opens the control tower and maturity evidence “Has anything drifted?” **Read state and review cadence bring them back**"
}
],
"box": false
},
{
"t": "h",
"l": 3,
"v": "Agent from idea to retirement"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Idea** Someone builds a helpful agent “This saves me an hour a day” **Register at creation, not at scale**"
},
{
"t": "p",
"v": "**Share** Colleagues start relying on it “It works, so why formalise it?” **Promotion gate between personal and team scope**"
},
{
"t": "p",
"v": "**Assure** Criticality scored, tier applied “How much control is proportionate?” **Gate names the evidence needed**"
},
{
"t": "p",
"v": "**Operate** Monitoring, cost and quality watched “Who notices if it degrades?” **Named owner and continuous duties**"
},
{
"t": "p",
"v": "**Review** Re-authorise, improve, restrict or retire “Is it still earning its authority?” **Review date forces the decision**"
},
{
"t": "p",
"v": "**Retire** Access revoked, records archived “What happens to the evidence?” **Archive and audit stage in the lifecycle**"
}
],
"box": false
},
{
"t": "h",
"l": 3,
"v": "Community session to reused pattern"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "p",
"v": "**Session happens** Weekly call, recorded from 17 August 2026 “Will I find this again?” **Automatic recording closes part of the capture gap**"
},
{
"t": "p",
"v": "**Capture** Transcript, chat, files, commitments “Who owns the follow-up?” **Session evidence pack with owner and actions**"
},
{
"t": "p",
"v": "**Classify** Taxonomy: capability, use case, sensitivity, owner, review date “Which of these applies to me?” **Shared taxonomy across communities**"
},
{
"t": "p",
"v": "**Connect** Federated index links to governed sources “Is this the current version?” **Index links rather than copies**"
},
{
"t": "p",
"v": "**Reuse** Another organisation adopts the pattern “Has anyone tried this already?” **Discovery by role, sector, maturity and supplier**"
},
{
"t": "p",
"v": "**Learn** Outcome and corrections returned “Did it actually work?” **Value reported as reuse, not attendance**"
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Service blueprint — an agent in an operational service"
},
{
"t": "p",
"v": "Five lanes across the same five moments: what the user does, what they see, what happens behind it, what supports it, and what is recorded."
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "User actions"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Requests help"
},
{
"t": "small",
"v": "Reviews agent output"
},
{
"t": "small",
"v": "Accepts or challenges"
},
{
"t": "small",
"v": "Escalates an error"
},
{
"t": "small",
"v": "Feeds back on value"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Frontstage"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Agent responds with citation"
},
{
"t": "small",
"v": "Confidence and limits shown"
},
{
"t": "small",
"v": "Human review point"
},
{
"t": "small",
"v": "Escalation route offered"
},
{
"t": "small",
"v": "Feedback captured in service"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Backstage"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Retrieval over authoritative sources"
},
{
"t": "small",
"v": "Guardrails and permission checks"
},
{
"t": "small",
"v": "Logging and audit trail"
},
{
"t": "small",
"v": "Incident triage and rollback"
},
{
"t": "small",
"v": "Evaluation against baseline"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Supporting processes"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Register and ATRS record maintained"
},
{
"t": "small",
"v": "Assurance gate evidence held"
},
{
"t": "small",
"v": "Cost allocated to the use case"
},
{
"t": "small",
"v": "Supplier and model change managed"
},
{
"t": "small",
"v": "Review or retirement decision"
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Evidence and records"
},
{
"t": "row",
"items": [
{
"t": "small",
"v": "Source provenance"
},
{
"t": "small",
"v": "Approval and sign-off"
},
{
"t": "small",
"v": "Monitoring and drift reports"
},
{
"t": "small",
"v": "Incident records"
},
{
"t": "small",
"v": "Benefit evidence"
}
]
}
]
}
]
},
{
"id": "docresources",
"part": "product-documentation",
"title": "Gov.uk AI resources",
"icon": "public",
"eyebrow": "Delivery documentation · external resources",
"heading": "The AI Engineering Lab and other gov.uk resources",
"lede": "This guide interprets and sequences; these resources supply engineering practice, worked examples and published evaluation. Link to them rather than copying them — they move faster than any document.",
"body": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "AI Engineering Lab"
},
{
"t": "title",
"v": "GDS Office of the CTO engineering-excellence project with DSIT",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "Programme site, an open GitHub resource repository under OGL v3.0, a Knowledge Hub community, role-specific material and cross-government hackathons. Enquiries to the GDS engineering-excellence team."
},
{
"t": "title",
"v": "Use it for: Engineering patterns, evaluation practice, contribution route",
"big": false
},
{
"t": "title",
"v": "Open resource",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "AI Knowledge Hub"
},
{
"t": "title",
"v": "i.AI, GDS within DSIT",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "A living resource on building and adopting AI in the public sector, with tool pages carrying what was built, what it cost and what the evaluation found."
},
{
"t": "title",
"v": "Use it for: Worked examples with evidence, before you build your own",
"big": false
},
{
"t": "title",
"v": "Open resource",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "i.AI product family"
},
{
"t": "title",
"v": "Incubator for Artificial Intelligence",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "Caddy, Consult, Extract, Redbox, Minute, Parlex, Scout, Connect, Lex and rAPId — several open-sourced, with published evaluations."
},
{
"t": "title",
"v": "Use it for: Reusable patterns and honest post-mortems, including a sunsetted product",
"big": false
},
{
"t": "title",
"v": "Open resource",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "AI Growth Lab"
},
{
"t": "title",
"v": "DSIT",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "Proposed cross-economy regulatory sandbox with time-limited exemptions under supervision; call for evidence closed January 2026."
},
{
"t": "title",
"v": "Use it for: Where regulation, not capability, is the blocker",
"big": false
},
{
"t": "title",
"v": "Open resource",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Fundamental AI Research Lab"
},
{
"t": "title",
"v": "DSIT with UKRI",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "Up to £40m for long-term research into limitations such as hallucination, unreliable memory and unpredictable reasoning."
},
{
"t": "title",
"v": "Use it for: Why some current risks are structural, not configuration errors",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "AI Playbook for the UK Government"
},
{
"t": "title",
"v": "GDS, February 2025",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "Ten principles and the life cycle this guide sequences; supersedes the Generative AI Framework."
},
{
"t": "title",
"v": "Use it for: The guidance in force",
"big": false
},
{
"t": "title",
"v": "Open resource",
"big": false
}
]
},
{
"t": "h",
"l": 2,
"v": "Which resource, which chapter"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Need an engineering or evaluation pattern",
"big": false
},
{
"t": "title",
"v": "AI Engineering Lab repository and community",
"big": false
},
{
"t": "small",
"v": "3.2 Architecture & quality"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Need a worked example with published evidence",
"big": false
},
{
"t": "title",
"v": "AI Knowledge Hub tool pages",
"big": false
},
{
"t": "small",
"v": "2.2 Use cases & evidence"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Need to see how a product was evaluated",
"big": false
},
{
"t": "title",
"v": "i.AI evaluations and blogs",
"big": false
},
{
"t": "small",
"v": "2.1 Evidence before assertion"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Need to know if regulation is the blocker",
"big": false
},
{
"t": "title",
"v": "AI Growth Lab call for evidence",
"big": false
},
{
"t": "small",
"v": "Resource library (home page)"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Need the authoritative principles",
"big": false
},
{
"t": "title",
"v": "AI Playbook for the UK Government",
"big": false
},
{
"t": "small",
"v": "1.3 The playbook"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Need transparency-record structure",
"big": false
},
{
"t": "title",
"v": "ATRS and the internal registry chapter",
"big": false
},
{
"t": "small",
"v": "4.5 Internal ATRS registry"
}
]
}
]
},
{
"id": "thesaurus",
"part": "knowledge-and-navigation",
"title": "Thesaurus",
"icon": "find_replace",
"eyebrow": "Knowledge and navigation · 10 entries",
"heading": "Thesaurus",
"lede": "Preferred term, the variants you will hear, and why the distinction matters. Cross-government acronyms and loose synonyms are the main cause of transcripts and registers that cannot be searched.",
"body": [
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Agent"
},
{
"t": "small",
"v": "digital worker"
},
{
"t": "small",
"v": "assistant"
},
{
"t": "small",
"v": "copilot"
},
{
"t": "small",
"v": "bot"
}
],
"box": false
},
{
"t": "p",
"v": "Use agent for the governed capability; digital worker only when the employment framing is deliberate."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Assurance"
},
{
"t": "small",
"v": "governance"
},
{
"t": "small",
"v": "compliance"
},
{
"t": "small",
"v": "sign-off"
}
],
"box": false
},
{
"t": "p",
"v": "Assurance tests claims against evidence; governance sets the rules. Do not use them interchangeably."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Criticality"
},
{
"t": "small",
"v": "risk level"
},
{
"t": "small",
"v": "importance"
},
{
"t": "small",
"v": "priority"
}
],
"box": false
},
{
"t": "p",
"v": "Criticality is dependency and consequence. Risk level is the output of scoring it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Register"
},
{
"t": "small",
"v": "inventory"
},
{
"t": "small",
"v": "catalogue"
},
{
"t": "small",
"v": "asset register"
}
],
"box": false
},
{
"t": "p",
"v": "Register for agents; catalogue for data; asset register for information assets. Three registers, integrated, not merged."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Evaluation"
},
{
"t": "small",
"v": "testing"
},
{
"t": "small",
"v": "benchmarking"
},
{
"t": "small",
"v": "evals"
}
],
"box": false
},
{
"t": "p",
"v": "Evaluation covers accuracy, safety and fitness against a baseline; testing alone implies pass or fail."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Guardrail"
},
{
"t": "small",
"v": "control"
},
{
"t": "small",
"v": "policy"
},
{
"t": "small",
"v": "limit"
}
],
"box": false
},
{
"t": "p",
"v": "Guardrails are technical constraints; controls include process and people; policy is the rule they enforce."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Retirement"
},
{
"t": "small",
"v": "decommissioning"
},
{
"t": "small",
"v": "offboarding"
},
{
"t": "small",
"v": "sunsetting"
}
],
"box": false
},
{
"t": "p",
"v": "Retirement is the lifecycle stage; decommissioning is the technical work inside it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Transparency record"
},
{
"t": "small",
"v": "ATRS record"
},
{
"t": "small",
"v": "algorithmic transparency"
},
{
"t": "small",
"v": "publication"
}
],
"box": false
},
{
"t": "p",
"v": "ATRS record is the published artefact; the internal registry extension is proposed, not mandated."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Value"
},
{
"t": "small",
"v": "benefit"
},
{
"t": "small",
"v": "ROI"
},
{
"t": "small",
"v": "savings"
}
],
"box": false
},
{
"t": "p",
"v": "State value as an outcome with a method. Reserve savings for cash-releasing benefits that finance has agreed."
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "h",
"l": 2,
"v": "Adoption"
},
{
"t": "small",
"v": "rollout"
},
{
"t": "small",
"v": "deployment"
},
{
"t": "small",
"v": "enablement"
}
],
"box": false
},
{
"t": "p",
"v": "Adoption is people changing how they work; deployment is the technology going live. They are not the same date."
}
]
}
]
},
{
"id": "taxonomy",
"part": "knowledge-and-navigation",
"title": "Taxonomies & ontology",
"icon": "schema",
"eyebrow": "Knowledge and navigation · classification",
"heading": "Taxonomies and ontology",
"lede": "Eight facets to classify anything in the estate, and the relations that connect them. Use the same vocabulary in the register, the ATRS record and the assurance case, or none of the three can be joined up.",
"body": [
{
"t": "h",
"l": 2,
"v": "Facets"
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Capability class"
},
{
"t": "small",
"v": "Sets which controls carry the weight."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Generative",
"big": false
},
{
"t": "title",
"v": "Agentic",
"big": false
},
{
"t": "title",
"v": "Declarative",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Assurance tier"
},
{
"t": "small",
"v": "Derived from criticality, not from popularity."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Personal",
"big": false
},
{
"t": "title",
"v": "Team",
"big": false
},
{
"t": "title",
"v": "Departmental",
"big": false
},
{
"t": "title",
"v": "Operational service",
"big": false
},
{
"t": "title",
"v": "Citizen impact",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Lifecycle stage"
},
{
"t": "small",
"v": "The status field in the register."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Proposed",
"big": false
},
{
"t": "title",
"v": "Pilot",
"big": false
},
{
"t": "title",
"v": "Live",
"big": false
},
{
"t": "title",
"v": "Restricted",
"big": false
},
{
"t": "title",
"v": "Suspended",
"big": false
},
{
"t": "title",
"v": "Retired",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Autonomy level"
},
{
"t": "small",
"v": "Guardrails apply at every level."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Suggest",
"big": false
},
{
"t": "title",
"v": "Recommend",
"big": false
},
{
"t": "title",
"v": "Prepare",
"big": false
},
{
"t": "title",
"v": "Act with approval",
"big": false
},
{
"t": "title",
"v": "Act and report",
"big": false
},
{
"t": "title",
"v": "Act autonomously",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Maturity level"
},
{
"t": "small",
"v": "Evidence-based, not self-scored."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Uncontrolled",
"big": false
},
{
"t": "title",
"v": "Aware",
"big": false
},
{
"t": "title",
"v": "Experimenting",
"big": false
},
{
"t": "title",
"v": "Governed",
"big": false
},
{
"t": "title",
"v": "Integrated",
"big": false
},
{
"t": "title",
"v": "Federated",
"big": false
},
{
"t": "title",
"v": "Adaptive",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Cost layer"
},
{
"t": "small",
"v": "Six layers, six owners."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Enterprise foundation",
"big": false
},
{
"t": "title",
"v": "Shared platform",
"big": false
},
{
"t": "title",
"v": "Product or service",
"big": false
},
{
"t": "title",
"v": "Team or workload",
"big": false
},
{
"t": "title",
"v": "Use case",
"big": false
},
{
"t": "title",
"v": "End user",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Evidence strength"
},
{
"t": "small",
"v": "Independently validated down to anecdotal."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Strong",
"big": false
},
{
"t": "title",
"v": "Moderate",
"big": false
},
{
"t": "title",
"v": "Emerging",
"big": false
},
{
"t": "title",
"v": "Weak",
"big": false
},
{
"t": "title",
"v": "Not assessed",
"big": false
}
]
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 3,
"v": "Role family"
},
{
"t": "small",
"v": "Determines the learning pathway."
}
],
"box": false
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Leadership",
"big": false
},
{
"t": "title",
"v": "Product and service",
"big": false
},
{
"t": "title",
"v": "Technical and data",
"big": false
},
{
"t": "title",
"v": "Governance and assurance",
"big": false
},
{
"t": "title",
"v": "Commercial and finance",
"big": false
},
{
"t": "title",
"v": "Operations and frontline",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "Ontology — subject, relation, object"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "has"
},
{
"t": "small",
"v": "Purpose, owner, capability class and authority"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "is recorded in"
},
{
"t": "small",
"v": "Register entry and ATRS record"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "is classified by"
},
{
"t": "small",
"v": "Criticality score, which sets its Assurance tier"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Assurance tier",
"big": false
},
{
"t": "small",
"v": "requires"
},
{
"t": "small",
"v": "Evidence set and review frequency"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "consumes"
},
{
"t": "small",
"v": "Data sources with provenance and permissions"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "acts within"
},
{
"t": "small",
"v": "Autonomy level and guardrails"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "generates"
},
{
"t": "small",
"v": "Cost, measured per interaction and per month"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Cost",
"big": false
},
{
"t": "small",
"v": "is allocated to"
},
{
"t": "small",
"v": "One of six cost layers with a named owner"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Agent",
"big": false
},
{
"t": "small",
"v": "affects"
},
{
"t": "small",
"v": "Service outcome, then public experience, then trust"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Incident",
"big": false
},
{
"t": "small",
"v": "triggers"
},
{
"t": "small",
"v": "Failure and recovery loop, then a control change"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner",
"big": false
},
{
"t": "small",
"v": "is accountable for"
},
{
"t": "small",
"v": "Agent through its whole lifecycle"
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Evidence",
"big": false
},
{
"t": "small",
"v": "supports or refutes"
},
{
"t": "small",
"v": "Claim, with method, publisher and date"
}
]
}
]
},
{
"id": "stack",
"part": "knowledge-and-navigation",
"title": "Platforms & repositories",
"icon": "integration_instructions",
"eyebrow": "Knowledge and navigation · platforms and repositories",
"heading": "Platforms and repositories",
"lede": "Links point to the vendors' own documentation. Nothing here is an endorsement or a procurement decision: platform choice sits with commercial colleagues, and the dependency and exit chapter applies to every entry on this page.",
"body": [
{
"t": "h",
"l": 2,
"v": "Government repositories and communities"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI Engineering Lab"
},
{
"t": "p",
"v": "Open resource repository under OGL v3.0, plus the Knowledge Hub community"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "i.AI open source"
},
{
"t": "p",
"v": "Consult, Redbox, Caddy and the rest of the incubator's public repositories"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI Knowledge Hub"
},
{
"t": "small",
"v": "Tool pages with cost, evaluation and lessons"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "GOV.UK Design System"
},
{
"t": "small",
"v": "Components and patterns for government services"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Microsoft"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Copilot Studio and agent builder"
},
{
"t": "small",
"v": "Declarative agents, connectors, publishing controls"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Microsoft 365 admin and Purview"
},
{
"t": "small",
"v": "Permissions, sensitivity labels, audit and retention"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Azure AI Foundry"
},
{
"t": "small",
"v": "Model hosting, evaluation and content safety"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Power Platform governance"
},
{
"t": "small",
"v": "Environment strategy, DLP policies, managed environments"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Google"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Gemini for Workspace admin"
},
{
"t": "small",
"v": "Data handling, availability and admin controls"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Vertex AI"
},
{
"t": "small",
"v": "Model garden, evaluation and grounding"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Google Cloud IAM"
},
{
"t": "small",
"v": "Least-privilege permissions and service accounts"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Atlassian"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Atlassian Intelligence and Rovo"
},
{
"t": "small",
"v": "Agents over Jira and Confluence content"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Jira for the pilot backlog"
},
{
"t": "small",
"v": "Decision records, findings and remediation tracking"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Confluence for evidence packs"
},
{
"t": "small",
"v": "Session packs, taxonomies and curated patterns"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Slack"
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Slack AI and agents"
},
{
"t": "small",
"v": "Assistants inside channels, with workspace data boundaries"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Slack Connect governance"
},
{
"t": "small",
"v": "Cross-organisation channels: membership and data controls"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Workflow Builder"
},
{
"t": "small",
"v": "Declarative automation without code"
},
{
"t": "title",
"v": "Documentation",
"big": false
}
]
}
],
"box": false
}
]
},
{
"id": "federated",
"part": "knowledge-and-navigation",
"title": "Federated guidance service",
"icon": "travel_explore",
"eyebrow": "Knowledge and navigation · proposed model",
"heading": "From fragmented guidance to a federated service",
"lede": "One trusted discovery layer over many authoritative sources, with a clear path from need to assured deployment. A federated index that links to governed sources beats another copy repository — this is a proposed model synthesised from community discussion and existing guidance, not an authoritative standard.",
"body": [
{
"t": "h",
"l": 2,
"v": "The landscape today"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Government policy & principles"
},
{
"t": "small",
"v": "AI Playbook for the UK Government"
},
{
"t": "small",
"v": "Generative AI Framework"
},
{
"t": "small",
"v": "Civil service guidance"
},
{
"t": "small",
"v": "Departmental policies"
},
{
"t": "small",
"v": "Ethical & legal frameworks"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Technical documentation"
},
{
"t": "small",
"v": "Microsoft Learn (Copilot, Agents, Studio, Purview)"
},
{
"t": "small",
"v": "Technical readiness & configuration"
},
{
"t": "small",
"v": "Security & identity blueprints"
},
{
"t": "small",
"v": "Administration guides"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Product & licensing information"
},
{
"t": "small",
"v": "Microsoft 365 Copilot licensing"
},
{
"t": "small",
"v": "Agent Builder & Studio documentation"
},
{
"t": "small",
"v": "Capacity, credits & consumption"
},
{
"t": "small",
"v": "Connectors & capabilities"
},
{
"t": "small",
"v": "Roadmaps & updates"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Local playbooks & templates"
},
{
"t": "small",
"v": "Departmental playbooks"
},
{
"t": "small",
"v": "Use case templates"
},
{
"t": "small",
"v": "Business cases"
},
{
"t": "small",
"v": "Process maps"
},
{
"t": "small",
"v": "Agent governance examples"
},
{
"t": "small",
"v": "Training materials"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Community evidence & insight"
},
{
"t": "small",
"v": "Weekly xGov Copilot calls"
},
{
"t": "small",
"v": "AI Engineering Lab"
},
{
"t": "small",
"v": "Community Q&A"
},
{
"t": "small",
"v": "Worked examples"
},
{
"t": "small",
"v": "Lessons learned"
},
{
"t": "small",
"v": "Emerging patterns & issues"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Sector & investment guidance"
},
{
"t": "small",
"v": "FinOps guidance"
},
{
"t": "small",
"v": "Commercial & procurement guidance"
},
{
"t": "small",
"v": "Data management & information governance"
},
{
"t": "small",
"v": "Security, privacy & assurance"
},
{
"t": "small",
"v": "Accessibility & inclusion"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "External & market intelligence"
},
{
"t": "small",
"v": "Supplier blogs & updates"
},
{
"t": "small",
"v": "Industry reports"
},
{
"t": "small",
"v": "Standards bodies"
},
{
"t": "small",
"v": "Academic research"
},
{
"t": "small",
"v": "International benchmarks"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "icon",
"v": "warning"
},
{
"t": "title",
"v": "Challenges of fragmentation",
"big": false
},
{
"t": "title",
"v": "Hard to find",
"big": false
},
{
"t": "title",
"v": "Out of date",
"big": false
},
{
"t": "title",
"v": "Inconsistent",
"big": false
},
{
"t": "title",
"v": "Duplication",
"big": false
},
{
"t": "title",
"v": "Gaps",
"big": false
},
{
"t": "title",
"v": "Heavy manual effort",
"big": false
},
{
"t": "title",
"v": "Risk of non-compliance",
"big": false
}
]
}
]
},
{
"t": "h",
"l": 2,
"v": "The target state"
},
{
"t": "p",
"v": "A federated guidance hub: navigate, connect, understand, apply, assure, share."
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "search"
},
{
"t": "p",
"v": "**Single discovery experience** One place to search and navigate."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "verified"
},
{
"t": "p",
"v": "**Authority & evidence clarity** Clear source, status and assurance level."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "hub"
},
{
"t": "p",
"v": "**Connected knowledge** Linked guidance, tools, templates, use cases and decisions."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "notifications_active"
},
{
"t": "p",
"v": "**Current & relevant** Version control, review dates and change alerts."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "badge"
},
{
"t": "p",
"v": "**Role & scenario based pathways** Tailored journeys by role, task and maturity level."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "checklist"
},
{
"t": "p",
"v": "**Implementation tools** Checklists, decision trees, templates, readiness assessments."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "forum"
},
{
"t": "p",
"v": "**Community feedback loop** Capture questions, insights and lessons into guidance."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "insights"
},
{
"t": "p",
"v": "**Measure & improve** Usage insights, gap analysis, continuous improvement."
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "Outcomes and benefits"
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "verified_user"
},
{
"t": "title",
"v": "Consistent, proportionate and compliant AI deployment",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "schedule"
},
{
"t": "title",
"v": "Faster access to the right guidance and tools",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "group_work"
},
{
"t": "title",
"v": "Reduced duplication and wasted effort",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "task_alt"
},
{
"t": "title",
"v": "Stronger assurance and risk management",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "trending_up"
},
{
"t": "title",
"v": "Higher quality outcomes and public value",
"big": false
}
]
}
],
"box": false
}
]
},
{
"id": "atlas",
"part": "deploy-for-real",
"title": "Service atlas",
"icon": "map",
"eyebrow": "Deploy for real · where the work actually is",
"heading": "Fifty-four public services, placed",
"lede": "The highest-volume services across UK central government, local government and the NHS, placed on two axes that decide almost everything else: how much of it there is, and what happens when it is wrong. Select a quadrant or a factor to filter. Open any service to load it into the generator.",
"body": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "54 services",
"big": false
},
{
"t": "title",
"v": "4 quadrants",
"big": false
},
{
"t": "title",
"v": "10 factors",
"big": false
},
{
"t": "title",
"v": "27 land in Tier 4",
"big": false
},
{
"t": "title",
"v": "34 sit in the dangerous quadrant",
"big": false
}
]
},
{
"t": "eyebrow",
"v": "One agent, end to end, in each quadrant"
},
{
"t": "eyebrow",
"v": "Filter by factor"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Services"
},
{
"t": "small",
"v": "54 of 54 shown"
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "How to read this, and how not to"
},
{
"t": "p",
"v": "Volume bands come from published transaction data; criticality and the factor tags are an assessment of consequence, not a classification you can cite. The value is not the placement — it is that placing a service forces a conversation about consequence that a business case usually skips. Move anything you disagree with, and record why you moved it."
}
]
}
]
},
{
"id": "casegen",
"part": "deploy-for-real",
"title": "Case generator",
"icon": "tune",
"eyebrow": "Deploy for real · working tool",
"heading": "Derive the governance set for any agent",
"lede": "Two bands and ten factors in. Tier, autonomy cap, oversight regime, currency window, evaluation set, threat priorities, DPIA route, contract clauses, workforce actions, register fields and kill criteria out — with the rules that produced them, so the answer can be argued with rather than accepted.",
"body": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Volume"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Criticality"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "h",
"l": 2,
"v": "Extended factors"
},
{
"t": "small",
"v": "No factors selected"
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "eyebrow",
"v": "Q1"
},
{
"t": "h",
"l": 2,
"v": "Automate with sampling"
},
{
"t": "title",
"v": "Tier 1 · Light",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "The strongest agent case in government. Volume pays for the governance; a mistake is recoverable. The risk here is not harm, it is quietly losing the ability to do the work."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Effective criticality"
},
{
"t": "title",
"v": "1 / 14",
"big": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Autonomy cap"
},
{
"t": "title",
"v": "L4",
"big": false
},
{
"t": "small",
"v": "Act with post-hoc sampling — exceptions escalate"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Challenge-rate floor"
},
{
"t": "title",
"v": "0.5%",
"big": false
},
{
"t": "small",
"v": "Below this, suspect oversight, not excellence."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Sign-off"
},
{
"t": "title",
"v": "Service owner",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Oversight regime"
},
{
"t": "small",
"v": "2% random sample, weekly"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Data protection route"
},
{
"t": "title",
"v": "Screening assessment, recorded and dated",
"big": false
},
{
"t": "p",
"v": "A screening that concludes no full DPIA is needed is still evidence. Keep it, and re-run it when scope changes."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Evaluations that must exist"
},
{
"t": "title",
"v": "Golden-set regression",
"big": false
},
{
"t": "p",
"v": "A frozen set of real historical cases with agreed correct outputs, re-run on every model or prompt change."
},
{
"t": "title",
"v": "Refusal and out-of-scope set",
"big": false
},
{
"t": "small",
"v": "Cases the agent must decline, escalate or flag rather than attempt."
},
{
"t": "title",
"v": "Cost and latency envelope",
"big": false
},
{
"t": "p",
"v": "Per-transaction cost and response time, with the ceiling that triggers review."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Threat priorities"
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Prompt injection via ingested content",
"big": false
},
{
"t": "title",
"v": "High",
"big": false
}
]
},
{
"t": "p",
"v": "Any document, email or record the agent reads is untrusted input. Test with adversarial content in every ingestion path."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Tool over-permission",
"big": false
},
{
"t": "title",
"v": "High",
"big": false
}
]
},
{
"t": "p",
"v": "The agent holds broader credentials than its task needs. Scope per action, not per agent."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Contract clauses that become mandatory"
},
{
"t": "title",
"v": "Model change notification",
"big": false
},
{
"t": "p",
"v": "Written notice before any model, weight or default change that affects behaviour, with a defined notice period."
},
{
"t": "title",
"v": "Log and evidence export",
"big": false
},
{
"t": "p",
"v": "All prompts, outputs, tool calls and decisions exportable in a documented open format, during and after the contract."
},
{
"t": "title",
"v": "No training on our data",
"big": false
},
{
"t": "small",
"v": "Explicit prohibition, with the right to evidence of compliance."
},
{
"t": "title",
"v": "Exit and portability",
"big": false
},
{
"t": "p",
"v": "Prompts, instruction blocks, eval sets and grounding indexes are ours, exportable, and documented well enough to rebuild elsewhere."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Workforce actions"
},
{
"t": "title",
"v": "Say where the freed capacity went",
"big": false
},
{
"t": "p",
"v": "Volume automation releases hours. Publish where they were redeployed. Unnamed savings become quiet headcount reduction with no decision behind it."
},
{
"t": "title",
"v": "Consultation before design freeze",
"big": false
},
{
"t": "p",
"v": "Staff and unions see the job-redesign proposal while it can still change. After the build is a briefing, not a consultation."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Register fields beyond the baseline"
},
{
"t": "title",
"v": "Accountable officer",
"big": false
},
{
"t": "small",
"v": "A named person, not a team or a role inbox."
},
{
"t": "title",
"v": "Reversal path",
"big": false
},
{
"t": "small",
"v": "How an effect is undone, tested, with the last test date."
},
{
"t": "title",
"v": "Currency holders",
"big": false
},
{
"t": "p",
"v": "Who currently holds valid approval authority, and when each window expires."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Kill criteria"
},
{
"t": "title",
"v": "Challenge rate falls below floor",
"big": false
},
{
"t": "p",
"v": "Reviewers overturning less than 0.5% of outputs. Either the agent is perfect or oversight has stopped. Assume the second."
},
{
"t": "title",
"v": "Golden-set regression unexplained",
"big": false
},
{
"t": "p",
"v": "Any drop on the frozen set that cannot be explained and fixed within the rollback SLA."
},
{
"t": "title",
"v": "An outcome nobody can explain",
"big": false
},
{
"t": "p",
"v": "A single decision the accountable officer cannot account for in plain language."
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Why it produced that answer"
},
{
"t": "small",
"v": "Low criticality sets a base of 1."
},
{
"t": "small",
"v": "Effective criticality 1 of 14 places this in Tier 1 · Light."
},
{
"t": "p",
"v": "Very high volume with low criticality places this in Q1 · Automate with sampling."
},
{
"t": "p",
"v": "The weights are a defensible starting point, not a measured value. Change them with your assurance function and record the change — a tool whose reasoning is hidden cannot be challenged, and a governance answer nobody can challenge is not governance."
}
]
}
]
},
{
"id": "worked",
"part": "deploy-for-real",
"title": "Worked examples",
"icon": "assignment_turned_in",
"eyebrow": "Deploy for real · four agents, one per quadrant",
"heading": "Correspondence triage, mandate to operation",
"lede": "Every artefact filled in, including the ones that are uncomfortable. Very high volume, moderate criticality — the case most departments would call easy. Named honestly it lands in **Tier 3 · Enhanced**, and that is the most useful thing in this chapter.",
"body": [
{
"t": "p",
"v": "These four agents are composites, not case studies. The figures are internally consistent and illustrative, and the decisions are the ones the sources support. Your organisation will ratify and adapt every one of them."
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "h",
"l": 2,
"v": "Mandate"
},
{
"t": "title",
"v": "1 of 10",
"big": false
}
],
"box": false
},
{
"t": "small",
"v": "Signed by the service owner before any build. One page, and it fits."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Problem",
"big": false
},
{
"t": "p",
"v": "Inbound correspondence arrives across six channels into one queue. Median time to first correct routing is four days; 11% is routed wrong at least once. Statutory deadlines are met by escalation, not by process."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "In scope",
"big": false
},
{
"t": "p",
"v": "Classify inbound correspondence to one of 34 work queues, extract sender, reference, subject and any stated deadline, and draft a suggested reply from the approved template set."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Explicitly out of scope",
"big": false
},
{
"t": "p",
"v": "Deciding any case. Sending anything. Replying to anyone. Handling correspondence flagged as a complaint, a legal threat, a safeguarding concern or a data-subject request — those route to a human unopened by the agent."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Success measures",
"big": false
},
{
"t": "p",
"v": "Time to correct routing, misroute rate, deadline breaches, and reviewer challenge rate. Not volume processed, and not staff satisfaction alone."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Decision rights",
"big": false
},
{
"t": "p",
"v": "The agent proposes. A caseworker commits. Nothing leaves the organisation without a named person attached to it."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Funding and duration",
"big": false
},
{
"t": "p",
"v": "Twelve months, one delivery team, with a stop point at month four and a re-authorisation decision at month twelve."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "What would make us stop",
"big": false
},
{
"t": "p",
"v": "Any statutory deadline missed because of the agent, or a challenge rate below 5% sustained for two months."
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "What changes the answer inside this quadrant"
},
{
"t": "p",
"v": "Same agent, same quadrant, one factor different. Each variant below is scored by the generator on this page, not asserted — so what moves and what does not is the tool's own arithmetic."
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "The agent sends the reply itself",
"big": false
},
{
"t": "title",
"v": "Tier moves",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "+ Irreversible",
"big": false
},
{
"t": "title",
"v": "Tier 4 · Critical",
"big": false
},
{
"t": "title",
"v": "Q1 · Automate with sampling",
"big": false
},
{
"t": "title",
"v": "L1 · Draft only — a person edits and commits every output",
"big": false
}
]
},
{
"t": "p",
"v": "One word in the mandate — send rather than draft — and an effect exists that no later correction undoes. Autonomy is forced to draft-only regardless of tier, and shadow mode becomes mandatory. This is the most common scope change requested after go-live, and the cheapest one to refuse."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Correspondence routinely contains health detail",
"big": false
},
{
"t": "title",
"v": "Tier moves",
"big": false
}
]
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "+ Special-category data",
"big": false
},
{
"t": "title",
"v": "Tier 4 · Critical",
"big": false
},
{
"t": "title",
"v": "Q1 · Automate with sampling",
"big": false
},
{
"t": "title",
"v": "L1 · Draft only — a person edits and commits every output",
"big": false
}
]
},
{
"t": "p",
"v": "The screening guardrail stops being a nicety and becomes the control the DPIA rests on. A leakage probe joins the evaluation set, and the grounding index has to be provably free of the material."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Internal-only queue, rule-based routing",
"big": false
},
{
"t": "title",
"v": "Tier moves",
"big": false
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "− Discretionary judgement",
"big": false
},
{
"t": "title",
"v": "− Vulnerable users",
"big": false
},
{
"t": "title",
"v": "Tier 2 · Standard",
"big": false
},
{
"t": "title",
"v": "Q1 · Automate with sampling",
"big": false
},
{
"t": "title",
"v": "L3 · Act within guardrails — every action logged and sampled",
"big": false
}
],
"box": false
},
{
"t": "p",
"v": "Remove the discretion and the vulnerable recipient and the same agent, at the same volume, drops a tier. Most of the governance in the worked example is paid for by those two factors, not by the volume."
}
]
}
]
}
]
},
{
"id": "boundary",
"part": "deploy-for-real",
"title": "What this does not cover",
"icon": "block",
"eyebrow": "Deploy for real · scope boundary",
"heading": "What this guide does not cover",
"lede": "A guide that does not state its boundary strands the team at the point where the interesting problems start. This chapter exists so nobody reaches gate 3 believing the architecture, the security testing or the legal advice was somewhere in the earlier parts.",
"body": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Fully covered here",
"big": false
},
{
"t": "p",
"v": "Principles, criticality and tiering, autonomy boundaries, registers and transparency records, oversight design including currency, cost and value modelling, roles and adoption, the pilot method."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Structured, needs your content",
"big": false
},
{
"t": "p",
"v": "Instruction blocks, evaluation plans, threat profiles, DPIA routes, contract clause sets, job redesign. The shape is here; the substance is yours."
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Named but not solved",
"big": false
},
{
"t": "p",
"v": "Reference architecture for your estate, engineering pipeline, executed security testing, model selection, sector regulation, organisational design, safety cases."
}
]
},
{
"t": "h",
"l": 2,
"v": "What a team must bring"
},
{
"t": "p",
"v": "Eight things this guide names but cannot supply. Each one has a normal home in an organisation. Find the home before the build starts, not after."
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "A target architecture for your estate",
"big": false
},
{
"t": "title",
"v": "Enterprise architecture",
"big": false
}
]
},
{
"t": "p",
"v": "This guide names what an architecture must satisfy. It does not know your identity provider, your network, your data platform or your existing contracts. Someone with authority over those has to draw it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Executed security testing",
"big": false
},
{
"t": "title",
"v": "Security function or an appointed third party",
"big": false
}
]
},
{
"t": "p",
"v": "The threat model here tells you what to test. Running the test — penetration testing, red-teaming, adversarial evaluation — is specialist work with a report at the end of it."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Legal advice on your specific case",
"big": false
},
{
"t": "title",
"v": "Legal and your DPO",
"big": false
}
]
},
{
"t": "p",
"v": "Everything here about lawful basis, automated decision-making and liability is structure for a conversation with a lawyer. It is not the advice."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Model selection and benchmarking",
"big": false
},
{
"t": "title",
"v": "Technical owner",
"big": false
}
]
},
{
"t": "p",
"v": "Which model, at what cost, with what evaluation on your data. This guide is deliberately model-agnostic, which means it cannot answer the question you will be asked first."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Sector regulation",
"big": false
},
{
"t": "title",
"v": "Your regulator and legal",
"big": false
}
]
},
{
"t": "p",
"v": "Health, policing, immigration, education and financial services each carry duties this guide does not enumerate. Where it conflicts with your regulator, your regulator wins."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Organisational design",
"big": false
},
{
"t": "title",
"v": "Accountable leadership",
"big": false
}
]
},
{
"t": "p",
"v": "Who reports to whom, which team holds the capability, and how it is funded past the pilot. The roles here are role names; you have people, budgets and politics."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Accessibility conformance",
"big": false
},
{
"t": "title",
"v": "Accessibility specialist",
"big": false
}
]
},
{
"t": "p",
"v": "The comprehension and inclusion tests here are necessary and not sufficient. A conformance audit against WCAG is a separate, evidenced exercise."
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "title",
"v": "Safety cases",
"big": false
},
{
"t": "title",
"v": "Clinical or safety governance",
"big": false
}
]
},
{
"t": "p",
"v": "Where an agent touches physical safety or clinical outcomes, a structured safety case is required and this guide does not contain one."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Five things to be honest about"
},
{
"t": "p",
"v": "Read these before citing anything in this guide to a board, an auditor or a regulator."
},
{
"t": "title",
"v": "Every artefact here is a proposal",
"big": false
},
{
"t": "p",
"v": "Nothing in this guide has statutory force. Until your organisation ratifies it, an artefact is a well-argued suggestion. Ratify it, or expect it to be ignored at the first disagreement."
},
{
"t": "title",
"v": "Figures marked illustrative are illustrative",
"big": false
},
{
"t": "p",
"v": "Where this guide shows a threshold, a percentage or a band, it is a defensible starting point drawn from the sources, not a measured value from your service. Replace it with yours."
},
{
"t": "title",
"v": "Volume and criticality placements are judgements",
"big": false
},
{
"t": "p",
"v": "The fifty-four services in the atlas are placed on published volume bands and an assessment of consequence. They are a starting position for an argument, not a classification you can cite."
},
{
"t": "title",
"v": "The generator does not approve anything",
"big": false
},
{
"t": "p",
"v": "It derives a consistent governance set from the inputs you give it. Give it flattering inputs and it will produce a flattering answer. That is a property of every assessment tool, including the ones with logos on."
},
{
"t": "title",
"v": "The four worked examples are composites",
"big": false
},
{
"t": "p",
"v": "They are built from the sources, from published service data and from patterns that recur across deployments. They are not case studies of four real agents, and no organisation in them is identifiable. Their numbers are internally consistent and illustrative — treat them as a model answer to compare your own against, not as evidence of what happened somewhere."
},
{
"t": "title",
"v": "Sequence is not schedule",
"big": false
},
{
"t": "p",
"v": "The life cycle here is ordered. It carries no durations, because the honest answer depends on your assurance capacity, and that is the constraint nobody plans for."
}
]
}
]
},
{
"id": "nations",
"part": "reference",
"title": "UK nations",
"icon": "map",
"eyebrow": "Pan-UK",
"heading": "All four nations have strategy. One has published evaluation.",
"lede": "Counting strategies overstates maturity: a benchmark that measures published strategy ranks all four nations comparable. Only one has a published body of controlled evaluation of deployment. What each nation can borrow from the others is the point of comparing.",
"body": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "UK Government"
},
{
"t": "small",
"v": "England and reserved functions"
},
{
"t": "eyebrow",
"v": "Strategy"
},
{
"t": "title",
"v": "AI Playbook (Feb 2025) · AI Opportunities Action Plan (Jan 2025) · departmental AI plans mandated by Cabinet Secretary objectives 2026-27",
"big": false
},
{
"t": "eyebrow",
"v": "Who leads"
},
{
"t": "p",
"v": "DSIT · Government Digital Service (incl. i.AI and the AI Engineering Lab) · AI Security Institute"
},
{
"t": "eyebrow",
"v": "Signals"
},
{
"t": "p",
"v": "Four published Copilot evaluations plus NAO and PAC scrutiny — the UK's distinctive asset is evaluation, not strategy"
},
{
"t": "small",
"v": "ATRS transparency records mandatory for departments and ALBs"
},
{
"t": "p",
"v": "Cabinet Secretary's AI Innovation Award (quarterly, from Jun 2026); AI & Data Challenge at 252 ideas"
},
{
"t": "title",
"v": "AI Playbook on GOV.UK",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Scotland"
},
{
"t": "small",
"v": "Devolved strategy, refreshed 2026"
},
{
"t": "eyebrow",
"v": "Strategy"
},
{
"t": "title",
"v": "Scotland's AI Strategy 2026–2031 (March 2026) — five-year plan, OECD-aligned, with the AI Scotland programme",
"big": false
},
{
"t": "eyebrow",
"v": "Who leads"
},
{
"t": "small",
"v": "Scottish Government · Scottish AI Alliance · The Data Lab"
},
{
"t": "eyebrow",
"v": "Signals"
},
{
"t": "small",
"v": "Scottish AI Register gives public-sector use cases a public face"
},
{
"t": "small",
"v": "Roughly 1 in 3 councils piloting or using AI — leading the UK"
},
{
"t": "small",
"v": "Independent advisory board plus sector and regional AI Champions"
},
{
"t": "title",
"v": "Scotland's AI Strategy",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Wales"
},
{
"t": "small",
"v": "National AI plan, 2025"
},
{
"t": "eyebrow",
"v": "Strategy"
},
{
"t": "title",
"v": "AI Cymru — the AI Plan for Wales (November 2025); Office for AI established June 2025",
"big": false
},
{
"t": "eyebrow",
"v": "Who leads"
},
{
"t": "small",
"v": "Welsh Government · Centre for Digital Public Services (CDPS)"
},
{
"t": "eyebrow",
"v": "Signals"
},
{
"t": "small",
"v": "Workforce, ethics and service guidance brought together early"
},
{
"t": "small",
"v": "Welsh-language parity a design requirement for public AI services"
},
{
"t": "p",
"v": "Council adoption early-stage (AI Readiness of UK Local Authorities, 2025)"
},
{
"t": "title",
"v": "Centre for Digital Public Services",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Northern Ireland"
},
{
"t": "small",
"v": "Draft strategy, consultation closed 7 October 2026"
},
{
"t": "eyebrow",
"v": "Strategy"
},
{
"t": "title",
"v": "Draft public-sector AI strategy (August 2026) — a draft, not adopted policy; Office of AI and Digital announced 2025",
"big": false
},
{
"t": "eyebrow",
"v": "Who leads"
},
{
"t": "p",
"v": "The Executive Office · Department of Finance · Office of AI and Digital"
},
{
"t": "eyebrow",
"v": "Signals"
},
{
"t": "p",
"v": "Proposes human-oversight teams, redress mechanisms and AI risk-management standards"
},
{
"t": "p",
"v": "Procurement expectations on suppliers: oversight, fairness, transparency, accountability"
},
{
"t": "small",
"v": "Stated aim: close the adoption gap with Scotland and Wales"
},
{
"t": "title",
"v": "NI Department of Finance",
"big": false
},
{
"t": "icon",
"v": "open_in_new"
}
]
}
],
"box": false
},
{
"t": "p",
"v": "Local-government comparison: AI Readiness of UK Local Authorities 2025 (Heriot-Watt University & GoLLM), 208 councils across all four nations — Scotland leads, with roughly one council in three piloting or using AI."
}
]
},
{
"id": "world",
"part": "reference",
"title": "International",
"icon": "public",
"eyebrow": "International",
"heading": "Benchmarking the UK's position",
"lede": "Five reference jurisdictions, five different bets: from the EU's binding risk tiers to Singapore's testing-led pragmatism. Genuinely transferable: Canada's public assessment and inventory approach, Australia's named AI ownership and workforce model, Singapore's central adoption flywheel, US inventory discipline, and the EU's joint procurement direction.",
"body": [
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Where"
},
{
"t": "eyebrow",
"v": "Key instrument"
},
{
"t": "eyebrow",
"v": "Model"
},
{
"t": "eyebrow",
"v": "For public bodies"
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "UK",
"big": false
},
{
"t": "p",
"v": "AI Playbook + sector regulators; ATRS; AI Security Institute evaluations"
},
{
"t": "small",
"v": "Principles-led, pro-innovation; non-statutory for now"
},
{
"t": "p",
"v": "Playbook principles and ATRS records; four published evaluations; frontier legislation under consideration"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "EU",
"big": false
},
{
"t": "p",
"v": "AI Act (in force 1 Aug 2024; Digital Omnibus 2026 defers high-risk duties to Dec 2027, public authorities to Aug 2030); EuropAI joint procurement"
},
{
"t": "small",
"v": "Binding, risk-based regulation"
},
{
"t": "p",
"v": "High-risk public deployments: conformity assessment, logging, human oversight, fundamental-rights impact assessment"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "US",
"big": false
},
{
"t": "small",
"v": "NIST AI Risk Management Framework; OMB memoranda; GAO oversight"
},
{
"t": "small",
"v": "Voluntary framework + agency rules; state laws vary"
},
{
"t": "p",
"v": "Federal AI use-case inventories and Chief AI Officers; rapid growth with persistent policy and budget challenges"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Canada",
"big": false
},
{
"t": "p",
"v": "Directive on Automated Decision-Making + Algorithmic Impact Assessment; agentic-AI guidance (May 2026)"
},
{
"t": "small",
"v": "Binding on federal automated decisions"
},
{
"t": "p",
"v": "Mandatory AIA before deployment; names skill atrophy as an oversight risk; bounded autonomy and recoverability"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Singapore",
"big": false
},
{
"t": "small",
"v": "Model AI Governance Framework · AI Verify toolkit"
},
{
"t": "small",
"v": "Voluntary, testing-led"
},
{
"t": "p",
"v": "Central funding, shared platforms, problem intake, applied learning — an adoption flywheel"
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "The EU AI Act's four risk tiers"
},
{
"t": "p",
"v": "In force 1 August 2024. The Digital Omnibus on AI (Regulation (EU) 2026/1744, in force 27 July 2026) moved stand-alone high-risk obligations to 2 December 2027, embedded high-risk systems to 2 August 2028 and high-risk systems for public authorities to 2 August 2030. Article 50 transparency duties have applied since 2 August 2026. Obligations follow the use case, not the technology. The benchmark other regimes are measured against."
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Unacceptable"
},
{
"t": "p",
"v": "Banned outright: social scoring, subliminal manipulation, real-time remote biometric ID in public spaces (narrow exceptions)."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "High risk"
},
{
"t": "p",
"v": "Essential public services, law enforcement, employment, critical infrastructure: conformity assessment, documentation, human oversight, EU database registration."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Limited risk"
},
{
"t": "p",
"v": "Transparency duties: chatbots must say they are AI; synthetic media must be labelled."
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Minimal risk"
},
{
"t": "p",
"v": "Most systems: spam filters, recommenders. No mandatory obligations; voluntary codes encouraged."
}
]
}
],
"box": false
}
]
},
{
"t": "h",
"l": 2,
"v": "The supranational layer"
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "OECD AI Principles + OECD.AI Observatory"
},
{
"t": "p",
"v": "The shared definition of AI used by the UK Playbook and the EU AI Act. The OECD.AI Policy Observatory tracks national AI policies worldwide; it is the model for this app's observatory."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "Council of Europe Framework Convention on AI (2024)"
},
{
"t": "p",
"v": "The first binding international AI treaty: human rights, democracy and the rule of law. The UK is a signatory."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "G7 Hiroshima AI Process (2023)"
},
{
"t": "p",
"v": "International guiding principles and a code of conduct for organisations developing advanced AI systems."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "AI Safety Summit series: Bletchley, Seoul, Paris"
},
{
"t": "p",
"v": "Frontier-safety declarations and the network of national AI safety and security institutes. Started at Bletchley Park, November 2023."
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 3,
"v": "UN General Assembly resolution (2024)"
},
{
"t": "p",
"v": "First UN consensus resolution on safe, secure and trustworthy AI for sustainable development."
}
]
}
],
"box": false
}
]
},
{
"id": "pace",
"part": "reference",
"title": "Pace & parity",
"icon": "bolt",
"eyebrow": "Reference · pace and parity",
"heading": "Named once. Controlled nowhere.",
"lede": "Skill atrophy appears in this guide as a single sentence in the evidence chapter, credited to Canada. Nothing sits behind it: no control, no metric, no register field, no gate criterion, no owner. That is the honest answer to the question, and it is now backlog item **PACE-01**. This chapter closes it, then applies the same test to everything else the world has built that we have not.",
"body": [
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Named"
},
{
"t": "title",
"v": "Yes",
"big": true
},
{
"t": "small",
"v": "One sentence, evidence chapter, attributed to Canada"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Controlled"
},
{
"t": "title",
"v": "No",
"big": true
},
{
"t": "small",
"v": "No control, no owner, no frequency"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Measured"
},
{
"t": "title",
"v": "No",
"big": true
},
{
"t": "small",
"v": "No metric in the barometer or control tower"
}
]
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "Gated"
},
{
"t": "title",
"v": "No",
"big": true
},
{
"t": "small",
"v": "No criterion in the quality gate"
}
]
}
],
"box": false
},
{
"t": "h",
"l": 2,
"v": "What Canada actually did"
},
{
"t": "p",
"v": "Six moves, and the last is the operable one. The instrument is the Guide on the Use of Agentic Artificial Intelligence, published by the Treasury Board of Canada Secretariat on 22 May 2026, sitting on top of a directive that already made assessment mandatory."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Named the risk inside an instrument",
"big": false
},
{
"t": "p",
"v": "Continued offloading of decisions to AI agents can lead to knowledge and skill atrophy and weakened human judgment. Employees become less practiced at core tasks and slower to notice when something is wrong."
},
{
"t": "eyebrow",
"v": "Guide on the Use of Agentic AI, TBS, 22 May 2026"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Tied it to the validity of oversight",
"big": false
},
{
"t": "p",
"v": "Maintenance of relevant skill sets among agentic AI users and deployers is treated as a prerequisite for meaningful human oversight. Not a wellbeing concern — a validity condition on the control itself."
},
{
"t": "eyebrow",
"v": "Same instrument, mitigations"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Added two agent-specific principles",
"big": false
},
{
"t": "p",
"v": "Bounded autonomy and recoverability, introduced alongside the existing generative-AI principles rather than replacing them."
},
{
"t": "eyebrow",
"v": "Same instrument"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Named the consequence chain",
"big": false
},
{
"t": "p",
"v": "Service quality, privacy and security, legal defensibility and public trust. The risk is framed as legal and operational exposure, not a training preference."
},
{
"t": "eyebrow",
"v": "Same instrument"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Made assessment mandatory and public",
"big": false
},
{
"t": "p",
"v": "Algorithmic Impact Assessment before deployment, four impact levels, external peer review at levels III and IV, questionnaire published openly including on GitHub."
},
{
"t": "eyebrow",
"v": "Directive on Automated Decision-Making"
}
]
},
{
"t": "card",
"kids": [
{
"t": "title",
"v": "Pointed at a concrete drill",
"big": false
},
{
"t": "p",
"v": "Periodic human-only execution of agent-handled tasks for comparison; staff trained to interrogate agent outputs rather than accept them."
},
{
"t": "eyebrow",
"v": "Commentary on the guide, May 2026"
}
]
}
],
"box": false
},
{
"t": "card",
"kids": [
{
"t": "eyebrow",
"v": "PACE-01 · proposed instrument"
},
{
"t": "h",
"l": 2,
"v": "Oversight currency"
},
{
"t": "p",
"v": "Canada names skill maintenance as a prerequisite for meaningful human oversight. A prerequisite that nobody measures is a sentiment. Aviation, nuclear operations and clinical practice all solved this the same way: recency requirements, evidenced, with authority attached. If you supervise the work, you stay current in the work. Six components, all of them auditable."
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Component"
},
{
"t": "eyebrow",
"v": "What it does"
},
{
"t": "eyebrow",
"v": "Who holds it"
},
{
"t": "eyebrow",
"v": "Evidence"
}
]
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Manual-mode drill",
"big": false
},
{
"t": "p",
"v": "A reviewer executes an agent-handled task unaided, then compares. Tier 1 monthly, tier 2 quarterly, tier 3 twice yearly."
},
{
"t": "small",
"v": "Reviewer"
},
{
"t": "small",
"v": "Drill record holding both outputs"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Currency window",
"big": false
},
{
"t": "p",
"v": "A reviewer whose last unaided execution predates the tier window loses approval authority on that queue until re-current. Enforced, not advised."
},
{
"t": "small",
"v": "Service owner"
},
{
"t": "small",
"v": "Authority state in the register"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Divergence log",
"big": false
},
{
"t": "p",
"v": "Every drill records agreement or disagreement and why. Sustained perfect agreement is itself a signal — it is what rubber-stamping looks like from outside."
},
{
"t": "small",
"v": "Reviewer"
},
{
"t": "small",
"v": "Divergence entries with reasons"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Challenge rate",
"big": false
},
{
"t": "p",
"v": "Proportion of agent outputs materially amended before acceptance. A falling challenge rate against a flat error rate is drift, not improvement."
},
{
"t": "small",
"v": "Control tower"
},
{
"t": "small",
"v": "Monthly trend per queue"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Rotation",
"big": false
},
{
"t": "p",
"v": "No reviewer supervises the same agent-handled queue continuously. Rotation preserves unaided practice somewhere in the role."
},
{
"t": "small",
"v": "Line manager"
},
{
"t": "small",
"v": "Rota showing unaided work"
}
],
"box": true
},
{
"t": "grid",
"cols": 4,
"items": [
{
"t": "title",
"v": "Displaced-skill field",
"big": false
},
{
"t": "p",
"v": "The register records which skills the agent displaces, who must retain them and to what standard. If nobody is named, the deployment is not approvable."
},
{
"t": "small",
"v": "Deployment owner"
},
{
"t": "small",
"v": "Register field, tested at each gate"
}
],
"box": true
}
]
},
{
"t": "p",
"v": "The uncomfortable component is the divergence log. A reviewer who agrees with the agent every time is either supervising a perfect agent or not supervising at all, and the record cannot tell the difference. Convergence has to be read as a warning rather than a result."
}
]
},
{
"t": "h",
"l": 2,
"v": "Eight instruments elsewhere, measured against ours"
},
{
"t": "p",
"v": "Three are outright gaps: something binding or published elsewhere with no UK counterpart. Four are partial. One is context we should track rather than copy."
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "EU"
},
{
"t": "title",
"v": "AI Act Article 4 — the AI literacy duty",
"big": false
},
{
"t": "eyebrow",
"v": "Gap"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "Binding since 2 February 2025 and unaffected by the proposed high-risk delay. Applies to deployers of every AI system, not only high-risk ones, and covers staff and contractors regardless of seniority. The compliance test is a dated, role-mapped programme that plausibly matches actual use."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "Training is encouraged throughout UK guidance but is nowhere a duty, and course completion is not evidence of competence."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Singapore"
},
{
"t": "title",
"v": "Model AI Governance Framework for Agentic AI",
"big": false
},
{
"t": "eyebrow",
"v": "Gap"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "World-first agentic framework, Davos 22 January 2026; version 1.5 on 20 May 2026 after feedback from over sixty organisations, with more than ten deployment case studies. Four pillars, unique agent identities tied to a supervising agent or user, threat modelling for memory poisoning, tool misuse and privilege compromise."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "We hold secure-AI guidance and our own patterns, but no single agentic instrument and no agent identity requirement."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 3,
"items": [
{
"t": "eyebrow",
"v": "Canada"
},
{
"t": "title",
"v": "Directive on Automated Decision-Making, plus the agentic guide",
"big": false
},
{
"t": "eyebrow",
"v": "Gap"
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "Assessment is mandatory before deployment, scored across four impact levels, with external peer review at the top two. The questionnaire is published openly, including on GitHub, and has become the instrument others copy."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "ATRS is transparency after the fact, not assessment before it, and it is not a gate."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Australia"
},
{
"t": "title",
"v": "AI Safety Institute inaugural report, 9 August 2026",
"big": false
},
{
"t": "eyebrow",
"v": "Open to all"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "Found that NIST's risk framework, OWASP's Agentic Top 10, Singapore's agentic framework and ASD's agentic guidance all rest on the same hidden assumption: that one organisation owns every agent in a system. Cross-organisational agent risk is a structural gap none of them covers."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "Our register assumes single-owner agents too. The gap is identified and unimplemented everywhere, including here — which makes it available."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Global"
},
{
"t": "title",
"v": "Singapore Consensus companion report on agentic risk",
"big": false
},
{
"t": "eyebrow",
"v": "Partial"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "Ten foundational principles for managing agentic risk: least privilege, traceable identity, auditability, validated deployment, adversarial resilience, multi-agent stability, runtime assurance, interruptibility, legibility and human oversight."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "We hold most of these as scattered controls. None is consolidated into a named principle set an approver can test a deployment against."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "OECD"
},
{
"t": "title",
"v": "Digital Government Outlook 2026",
"big": false
},
{
"t": "eyebrow",
"v": "Partial"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "Thirty-two of thirty-six countries run AI training programmes; only thirteen of thirty-six train on using AI in public services or in policymaking. Broad capability is improving faster than role-specific readiness. Thirty of thirty-six have an institution governing public-sector AI; twenty-one of thirty-six offer central procurement support."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "Our academy is role-based on paper. Depth of oversight-specific training has never been tested against this benchmark."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "G7"
},
{
"t": "title",
"v": "GAIN network and the GovAI Grand Challenge",
"big": false
},
{
"t": "eyebrow",
"v": "Partial"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "A G7 AI network convened and chaired by Canada, tasked with a roadmap to scale successful public-sector AI projects and a catalogue of open-source, shareable AI solutions for members."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "We participate. We have contributed no reusable instrument to the catalogue."
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "row",
"items": [
{
"t": "eyebrow",
"v": "Asia-Pacific"
},
{
"t": "title",
"v": "Korea AI Basic Act (2026), Taiwan AI Basic Act (2025)",
"big": false
},
{
"t": "eyebrow",
"v": "Tracking"
}
]
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "p",
"v": "Regional convergence on human oversight, data safeguards and prohibitions on harmful or deceptive use, arriving through basic law rather than sectoral guidance."
},
{
"t": "group",
"kids": [
{
"t": "eyebrow",
"v": "Where we stand"
},
{
"t": "p",
"v": "Context rather than gap: the UK route is guidance-led by choice. Worth tracking for interoperability, not imitation."
}
]
}
],
"box": false
}
]
},
{
"t": "h",
"l": 2,
"v": "The backlog, MoSCoW'd"
},
{
"t": "p",
"v": "Thirteen items and three deliberate refusals. Every item has an owner and a first deliverable, because an unowned backlog item is a wish."
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Must",
"big": false
},
{
"t": "p",
"v": "This quarter. Each one closes a gap that leaves an existing control unsound."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-01",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Oversight currency",
"big": false
},
{
"t": "p",
"v": "Manual-mode drills, currency windows, divergence log, rotation and the displaced-skill register field. Turns the Canada finding into something an auditor can test."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Learning lead with service owners",
"big": false
},
{
"t": "title",
"v": "First: Control spec and tier windows",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-02",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Evidenced literacy duty",
"big": false
},
{
"t": "p",
"v": "A dated, role-mapped competence requirement for anyone who operates or approves an agent, contractors in scope. The Article 4 test, applied voluntarily and ahead of any obligation."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Learning lead",
"big": false
},
{
"t": "title",
"v": "First: Role-to-competence map",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-03",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Agent identity",
"big": false
},
{
"t": "p",
"v": "Every agent carries a unique identity bound to a named human owner and, where orchestrated, to a supervising agent. Prerequisite for audit, for handover and for retirement."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Architecture",
"big": false
},
{
"t": "title",
"v": "First: Identity schema in the register",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-04",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Bounded autonomy and recoverability",
"big": false
},
{
"t": "p",
"v": "Adopt both as named approval principles. Every agent has a tested interrupt and a tested rollback before it goes live, with the test evidenced."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Assurance lead",
"big": false
},
{
"t": "title",
"v": "First: Two gate criteria and a test script",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-05",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Drift telemetry",
"big": false
},
{
"t": "p",
"v": "Challenge rate and divergence surfaced in the control tower beside assurance status, per queue, monthly."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Control tower",
"big": false
},
{
"t": "title",
"v": "First: Two new tiles",
"big": false
}
]
}
]
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Should",
"big": false
},
{
"t": "small",
"v": "Next two quarters. These extend reach rather than repair a fault."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-06",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Cross-boundary agents",
"big": false
},
{
"t": "p",
"v": "Register fields for agents we interact with but do not own: supplier, other department, arm's-length body. The first implementation of the gap Australia named."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Architecture with commercial",
"big": false
},
{
"t": "title",
"v": "First: Field set and one worked case",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-07",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Multi-agent systemic risk",
"big": false
},
{
"t": "p",
"v": "An assessment step for orchestrated patterns, covering interaction effects that no single-agent assessment can see."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Assurance lead",
"big": false
},
{
"t": "title",
"v": "First: Assessment addendum",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-08",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Oversight-specific training",
"big": false
},
{
"t": "p",
"v": "Not general AI awareness. Training in interrogating outputs, recognising plausible-but-wrong, and running a manual-mode drill."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Academy",
"big": false
},
{
"t": "title",
"v": "First: Curriculum for three roles",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-09",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Assessment before deployment",
"big": false
},
{
"t": "p",
"v": "A scored pre-deployment assessment distinct from the transparency record, with independent peer review at the top tiers."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Governance office",
"big": false
},
{
"t": "title",
"v": "First: Scoring instrument",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-10",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Extended threat model",
"big": false
},
{
"t": "p",
"v": "Memory poisoning, tool misuse and privilege compromise added to the agent threat model and to the security review."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Security lead",
"big": false
},
{
"t": "title",
"v": "First: Threat model revision",
"big": false
}
]
}
]
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Could",
"big": false
},
{
"t": "p",
"v": "Worth doing, sequenced behind the above. This is where the leadership case is actually made."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-11",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Publish the instruments",
"big": false
},
{
"t": "p",
"v": "Register schema, gate criteria and the currency control released openly under the Open Government Licence, versioned, as Canada did with its assessment."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Governance office",
"big": false
},
{
"t": "title",
"v": "First: Public repository",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-12",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Reciprocal assurance pilot",
"big": false
},
{
"t": "p",
"v": "One partner nation, one deployment class, mutual recognition of assurance evidence. Tests whether our evidence travels before we offer it widely."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Governance office",
"big": false
},
{
"t": "title",
"v": "First: Pilot terms with one partner",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-13",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Atrophy measurement study",
"big": false
},
{
"t": "p",
"v": "Longitudinal measurement of unaided performance in agent-supported roles, with an academic partner. Nobody holds this data, which is why everyone is asserting rather than measuring."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: Evaluation lead",
"big": false
},
{
"t": "title",
"v": "First: Study protocol",
"big": false
}
]
}
]
}
],
"box": true
}
]
},
{
"t": "card",
"kids": [
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "Won't — for now",
"big": false
},
{
"t": "small",
"v": "Named, so the omission reads as a decision rather than an oversight."
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-W1",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Statutory literacy penalties",
"big": false
},
{
"t": "p",
"v": "Not ours to set. We build the evidence base a policy decision would need."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: —",
"big": false
},
{
"t": "title",
"v": "First: Revisit when policy opens",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-W2",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "A UK agentic act",
"big": false
},
{
"t": "p",
"v": "The guidance route moves faster, and we do not yet hold the evidence to legislate well."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: —",
"big": false
},
{
"t": "title",
"v": "First: Revisit after PACE-13",
"big": false
}
]
}
]
}
],
"box": true
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "title",
"v": "PACE-W3",
"big": false
},
{
"t": "group",
"kids": [
{
"t": "title",
"v": "Certifying supplier agents",
"big": false
},
{
"t": "p",
"v": "Capacity we do not have. Contract terms and evidence requirements instead."
},
{
"t": "row",
"items": [
{
"t": "title",
"v": "Owner: —",
"big": false
},
{
"t": "title",
"v": "First: Revisit at scale",
"big": false
}
]
}
]
}
],
"box": true
}
]
},
{
"t": "h",
"l": 2,
"v": "Four things nobody has published yet"
},
{
"t": "p",
"v": "Parity is a floor, and reaching it puts us level with instruments already a year old. Each of these is a control we could be first to publish, and three of the four came out of our own case work rather than the international scan."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "bolt"
},
{
"t": "title",
"v": "Oversight currency as a revocable authority",
"big": false
},
{
"t": "p",
"v": "Canada names skill maintenance as a prerequisite for oversight. Nobody has made it measurable and revocable — an approval right you can lose by going stale. Publishing the control with tier windows and drill records would be first."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "link_off"
},
{
"t": "title",
"v": "The orphaned-agent control",
"big": false
},
{
"t": "p",
"v": "Machinery-of-government change moves an owner; active agent work continues unattended and is later inherited by a team that never approved it. No framework surveyed handles ownership transfer on reorganisation."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "hourglass_top"
},
{
"t": "title",
"v": "Attention budgets for review",
"big": false
},
{
"t": "p",
"v": "Approval capacity treated as a finite, costed resource and rationed by risk tier, so review is real where it matters instead of uniform and hollow everywhere."
}
]
},
{
"t": "card",
"kids": [
{
"t": "icon",
"v": "handshake"
},
{
"t": "title",
"v": "Cross-boundary agent handshake",
"big": false
},
{
"t": "p",
"v": "Australia's institute named cross-organisational agent risk as a gap no framework covers. A register field plus a handshake at departmental and supplier boundaries would be the first implementation of it."
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Keeping pace is a rhythm, not a document"
},
{
"t": "p",
"v": "Four of the eight instruments above were published or revised in the last eight months. A guide refreshed annually is a guide that is wrong for eleven of them."
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "eyebrow",
"v": "Weekly"
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "Horizon scan into the guidance ticker: new instrument, new version, new enforcement action."
},
{
"t": "title",
"v": "Scan rota",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "eyebrow",
"v": "Fortnightly"
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "Instrument diff across the eight tracked jurisdictions. What changed, what it obliges, whether it touches a control we hold."
},
{
"t": "title",
"v": "Governance office",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "eyebrow",
"v": "Monthly"
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "Re-MoSCoW the backlog. Anything sitting in Should for two cycles is promoted or killed, never carried."
},
{
"t": "title",
"v": "Governance office",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "eyebrow",
"v": "Quarterly"
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "Publish the delta pack and contribute one reusable instrument to a shared catalogue."
},
{
"t": "title",
"v": "Governance office",
"big": false
}
]
}
],
"box": false
},
{
"t": "grid",
"cols": 2,
"items": [
{
"t": "eyebrow",
"v": "Annually"
},
{
"t": "group",
"kids": [
{
"t": "p",
"v": "Reciprocal review with a partner nation: they read our instruments, we read theirs, both publish the findings."
},
{
"t": "title",
"v": "SRO",
"big": false
}
]
}
],
"box": false
}
]
},
{
"t": "card",
"kids": [
{
"t": "h",
"l": 2,
"v": "Where the work goes once it exists"
},
{
"t": "p",
"v": "Five routes already open to us. None of them takes a position paper; all of them take a working instrument."
},
{
"t": "title",
"v": "G7 GAIN",
"big": false
},
{
"t": "p",
"v": "The network's remit includes a catalogue of open-source, shareable AI solutions for members. A register schema and a currency control are exactly that shape."
},
{
"t": "title",
"v": "OECD.AI and the Digital Government Outlook",
"big": false
},
{
"t": "p",
"v": "The comparative baseline every country is measured against. Contributing method, rather than only returning survey answers, changes what the baseline measures."
},
{
"t": "title",
"v": "Council of Europe Framework Convention",
"big": false
},
{
"t": "p",
"v": "The first binding international AI treaty, on human rights, democracy and the rule of law. The UK is a signatory, and the validity of human oversight sits squarely inside its scope."
},
{
"t": "title",
"v": "The AI safety institute network",
"big": false
},
{
"t": "p",
"v": "The UK allocated roughly A$460m at the 2025 spending review, against about A$50m over five years in Canada and about A$11m a year in Singapore. Capacity to convene is not the constraint."
},
{
"t": "title",
"v": "Open publication",
"big": false
},
{
"t": "p",
"v": "Open Government Licence v3.0, public repository, versioned. Canada published its assessment questionnaire openly and it became the reference others adopted."
}
]
}
],
"box": false
},
{
"t": "p",
"v": "Sources: Guide on the Use of Agentic Artificial Intelligence and Directive on Automated Decision-Making, Treasury Board of Canada Secretariat; EU AI Act Article 4 and the AI Office Q&A on AI literacy; IMDA Model AI Governance Framework for Agentic AI v1.5 (20 May 2026); Australian AI Safety Institute inaugural report (9 August 2026); Singapore Consensus companion report on agentic risk; OECD Digital Government Outlook 2026 and Building an AI-ready public workforce (2026); G7 GovAI Grand Challenge and GAIN, Treasury Board of Canada Secretariat and Impact Canada. Figures are as published by each source and are not converted or combined."
}
]
}
],
"withheld": [
{
"id": "raicatalogue",
"title": "AI tool catalogue",
"part": "Decide and prioritise",
"reason": "Organisation-specific approved-tool statuses and links to internal pages; publishing them would read as endorsement."
},
{
"id": "raianalytics",
"title": "Analytics, estimators & dashboards",
"part": "Decide and prioritise",
"reason": "Its charts count entries in the withheld tool catalogue; the estimators it links to are published in \"Cost, value and investment\"."
},
{
"id": "doccoverage",
"title": "Blueprint coverage",
"part": "Product documentation",
"reason": "Coverage map of internal source documents and blueprints, which are not published."
},
{
"id": "raispec",
"title": "Portal specification trace",
"part": "Product documentation",
"reason": "Traces requirements from an internal specification document."
},
{
"id": "graph",
"title": "Knowledge graph",
"part": "Knowledge and navigation",
"reason": "Interactive graph of the internal edition; replaced by the generated knowledge map in this playbook."
},
{
"id": "sitemap",
"title": "Portal & site map",
"part": "Knowledge and navigation",
"reason": "Map of the internal edition, including internal pages. The public playbook has its own contents and drawer."
},
{
"id": "tours",
"title": "Guided tours",
"part": "Knowledge and navigation",
"reason": "Guided tours route through internal pages; the home page personas and journeys replace them."
},
{
"id": "media",
"title": "Media centre",
"part": "Knowledge and navigation",
"reason": "Films and decks: never kept in a repository (owner rule). They stay in the media library outside GitHub."
},
{
"id": "selfcheck",
"title": "Self-assurance console",
"part": "Knowledge and navigation",
"reason": "Self-test console of the internal edition; the public site runs its own checks in CI (see Transparency)."
},
{
"id": "sources",
"title": "Sources & blueprints",
"part": "Reference",
"reason": "Lists internal source documents. Public sources are in the resource library on the home page."
},
{
"id": "glossary",
"title": "Glossary",
"part": "Knowledge and navigation",
"reason": "Merged into the portal glossary, so the whole portal has one glossary: 39 new terms added, existing definitions kept."
}
]
}