Level
1
1.0–1.99
Initial
Ad hoc and individual-driven
Experiments are isolated, sponsorship is inconsistent and outcomes depend on individual initiative.
Typical evidence
- Uncoordinated pilots
- Local data access
- Informal risk decisions
The standard · Publication 02
Five levels of organisational capability.
The levels describe increasingly deliberate, repeatable and adaptive ways of governing AI. They are cumulative: higher maturity depends on foundations established below.
Level
2
2.0–2.99
Developing
Repeatable but tactical
Useful patterns are emerging, although practices remain uneven and enterprise ownership is limited.
Typical evidence
- Named pilot owners
- Early reusable methods
- Emerging governance
Level
3
3.0–3.99
Defined
Documented and integrated
AI becomes an organisational capability with shared standards, defined roles and managed delivery routes.
Typical evidence
- Published standards
- Enterprise architecture
- Defined accountabilities
Level
4
4.0–4.99
Managed
Measured and controlled
Performance, value and risk are measured consistently across the AI portfolio and inform investment decisions.
Typical evidence
- Portfolio metrics
- Operational controls
- Benefits tracking
Level
5
5.0
Optimising
Adaptive and strategic
Continuous improvement is embedded and AI capability is deliberately renewed as conditions change.
Typical evidence
- Adaptive controls
- Continuous learning
- Strategic advantage