Operating Model & Delivery ยท OPM-01
Target operating model for AI
The organisation has a defined, approved operating model that states who builds, buys, runs and supports AI, and what is done centrally and what locally.
Why it matters
Without an agreed model, central teams and business units duplicate effort, compete for the same scarce skills, or each assume the other is responsible. Whether the model is central, devolved or hub-and-spoke matters less than everyone knowing which it is.
Levels
| 0 Absent | Confirmed absent: the evidence shows nothing is in place. Not known is recorded as insufficient evidence, not 0. |
|---|---|
| 1 Initial | No defined model; AI work is done wherever capability and interest happen to exist. |
| 2 Developing | A model is described or emerging, but responsibilities between central and local teams are unclear or not followed. |
| 3 Defined | An approved operating model sets out who builds, buys, runs and supports AI, centrally and locally, through named and resourced teams, and is followed across the organisation. |
| 4 Managed | The model's performance is measured, for example delivery times, capacity and demand, and work done outside it is identified and addressed. |
| 5 Optimising | The model is reviewed against outcomes and adjusted as AI use grows or its nature changes. |
Level 3 criteria
Level 3 is met when every applicable criterion is met. Criteria marked with an overlay apply only to organisations with that characteristic (overlays).
- OPM-01.1 An approved operating model for AI is documented.
- OPM-01.2 The model states which parts of building, buying, running and supporting AI are done centrally and which locally.
- OPM-01.3 Each of these activities has a named responsible team with the resources to carry it out.
- OPM-01.4 The interfaces between central and local teams are defined.
- OPM-01.5 AI work across the organisation follows the model.
Assessment questions
- Which teams build, buy, run and support AI, and where is that written down?
- What is done centrally and what is left to business units?
- Where is AI work happening outside the agreed model?
- The approved operating model document, with owner and review date
- Team remits or service catalogue entries that reflect the model
How to reach Level 3
- Map who currently builds, buys, runs and supports AI, including work happening in business units without central involvement.
- Choose a model (central, devolved or hub-and-spoke) and write down which of those four activities sit centrally and which locally.
- Name the team responsible for each activity, the people and budget behind it, and how central and local teams hand work to each other.
- Get the model approved by the executive sponsor (GOV-02), give it an owner and review date, and update team remits to match.
- Tell every business unit how to route AI work through the model and check new work against it.
Common pitfalls
- Writing a model on paper while the people and money stay where they were, so nobody follows it.
- Covering building and buying but forgetting who runs and supports AI once it is live.
Starter artefacts
- Operating model document with owner and review date
- Responsibility matrix for build, buy, run and support
- Updated team remits
Ownership and sequence
- Accountable
- executive sponsor
- Contributors
- delivery lead, ai governance lead, people lead
- Depends on
- GOV-02
- Roadmap step
- 2
- Lifecycle
- strategy, operate
Cross-references
| ISO/IEC 42001 | 7.1 partial; A.4.2 partial |
|---|---|
| JSP 936 Part 1 | para 20 partial; para 103 partial |
| MITRE AI Maturity Model | P3.2 direct |
| UK Data and AI Ethics Framework | SAT 3.10 partial |
| UNESCO AI Maturity Framework | 2.2 direct |
direct: this control at Level 3 meets the item. partial: it covers part of it.