AI Capability Maturity Framework
AI-CMF Body of Knowledge and Practical Guidebook
Thirty-eight management capabilities, each broken into building blocks you can assess on five observable levels and improve one step at a time.
- 5macro-capabilities
- 38critical capabilities
- ~500building blocks
- 5maturity levels
Five macro-capabilities, one mark per critical capability. Choose one to see what it contains.
The framework in two minutes
Most organisations can list the AI projects they are running. Far fewer can say which capabilities they need to run AI well, how mature each one is, and what to improve first.
That gap shows up in familiar ways. A model passes its pilot and cannot be deployed safely. Governance is written down but nobody applies it. The same problems return in every project because nothing learned in one reaches the next.
AI-CMF answers one question: which capabilities does this organisation need to run AI well, and how mature is each of them today?
AI as management capability, not technology. The framework describes the organisation around AI in five macro-capabilities: AI Strategy and Governance (9 critical capabilities), Data, Technology and Infrastructure (9), AI People, Culture and Organisation (6), AI Delivery and Operations (8) and AI-Enabled Business Transformation (6).
Building blocks you can assess. Each critical capability is broken into capability building blocks, typically eleven to sixteen of them. Each block is assessed on its own, because a capability is rarely mature everywhere: an organisation can have excellent test automation and no independent sign-off at all.
Five observable levels. Initial, Basic, Intermediate, Advanced and Optimising. Every level describes what is done and what results, never how people feel about it. A written policy on its own does not reach Level 3; the practice has to be applied, with records to show it. Level 5 is earned through measured improvement inside the organisation, not through awards or reputation.
Evidence, not assertion. Each level descriptor comes with the evidence that would demonstrate it: documents, records of the practice being carried out, interviews, observation and metrics.
Proportionate. There is one scale for every organisation, with guidance on the target level that fits the context. A small firm using AI internally does not need the same targets as a regulator-facing, high-risk deployment.
From gaps to a plan. Every capability carries a catalogue of practices, outcomes and metrics, and a catalogue of improvement actions tagged to the building block and the level step each one moves. After an assessment, the gaps the organisation chooses to close become a Capability Improvement Plan.
What the framework produces. A maturity profile across the capabilities in scope, down to building-block level; the gap between where you are and where you need to be; and a sequenced plan to close it.
Where to start. Read how the framework is built, see one capability in full, or request a free executive briefing to walk through it with your team.
AI-CMF in three minutes
A short overview for readers who prefer to watch: the five macro-capabilities, how capabilities break into building blocks, what the five levels mean, and how an assessment turns into an improvement plan.
Go deeper
How the framework is built
The architecture, all thirty-eight critical capabilities, what a building block is, how the five levels are written, what every capability chapter contains, and how capabilities depend on each other.
Read how it is built Sample capabilityOne capability in full: AI Testing and Quality Assurance
Definition, scope, sixteen building blocks with all five level descriptors and the evidence behind them, target levels by context, and extracts from the practice and action catalogues.
Read the sample capability MethodCapability maturity assessment
Two stages: a self-assessment of all thirty-eight capabilities, validated in workshops and interviews, then a deep dive into the five to eight that matter most.
Read the assessment method MethodCapability Improvement Plan
How the deep-dive gaps become a sequenced plan, delivered through an improvement programme your own teams lead, with a worked example from AI Testing and Quality Assurance.
Read the improvement methodFour ways to start
-
Executive briefing Free
A walkthrough of the framework with your leadership team: which capabilities matter most for your AI plans, and what an assessment would involve. No commitment.
-
Pilot assessment Free
The full method applied end to end to one critical capability, typically AI Governance and Policy Framework or AI Testing and Quality Assurance.
-
Capability maturity assessment
All thirty-eight capabilities self-assessed and validated in workshops and interviews, then a building-block deep dive into the five to eight that matter most, with an executive read-out.
-
Capability Improvement Plan
The deep-dive gaps turned into a sequenced plan with owners, measures and a re-assessment point, and support for the improvement teams that deliver it.