Every organisation that adopts AI without governance architecture ends up in the same place: a hundred people using AI a hundred different ways, each interaction producing local value, none of it accumulating into organisational capability.
The governance model follows an engineering change order logic that manufacturing executives already understand. Every system message has an identifier and version number. Every change to a system message goes through a defined process: evidence from practice surfaces through the secretariat, proposed revisions are assessed for impact and consistency, the Council approves changes, and the new version is documented with an effective date. You can trace forward — which processes are using which version of which system message — and backward — what AI guidance was operative when a particular decision was made. This is not bureaucracy. It is traceability. When something goes wrong, you can ask what the system message surfaced or missed and learn from it. When something goes right, you can understand why and replicate it.
Individual experiments never become collective learning. Good practices discovered by one person remain invisible to others. Poor practices — AI reinforcing biases, producing confident nonsense, substituting speed for judgment — persist because no one is responsible for the quality of how AI is used. The organisation gets busier with AI without getting better at it.
The Process Design Council solves this structurally. It is a cross-functional governance body — not an IT committee, not a policy group — that takes explicit responsibility for the design of the management system, including how AI integrates into it. The Council does not govern individual AI use. It governs the architectural layer: the system messages that shape how AI thinks within the organisation, the processes where AI adds value at defined judgment points, and the feedback mechanisms through which learning from AI-assisted work flows back into the system.
The secretariat is the operational heart of the system. It manages intake and routing for process design proposals. It stewards the system messages — drafting new ones, proposing revisions based on accumulated evidence, maintaining the version archive. It observes Pair Design Reviews and captures one-line learning notes about where the system message helped and where it fell short. It synthesises panel notes using AI assistance, compresses them into Executive Decision Capsules for the Council, and maintains the decision register and conditions log. The secretariat does not make decisions. It creates the conditions for the Council to decide well, using the same information flow design principles the OAC Fellowship teaches for the management system as a whole.
The programme teaches you to design and install this governance structure. Phase one — which can serve as the entry point for the broader OAC Fellowship — has participants designing their own secretariat, defining its roles and rhythms, and building their first governed system message. They use AI to do this work, which means they experience the method while building the governance for it. The deliverable from this programme is not a policy document. It is an operating governance structure with defined roles, rhythms, artefacts, and the architectural capability to ensure that AI integration compounds organisational capability rather than fragmenting it.
Course Theme
This course will be released on the Learning Hub 10th April 2026
theme
1
Strategy
Why Governance Architecture, Not AI Policy
theme
2
Planning
Designing the Council Structure
theme
3
Follow-up
The Pair Design Review and Intake Process
theme
4
Results Analysis
System Message Governance in Practice
theme
5
Discovery
Scaling and Adapting the Governance Architecture
theme
6
Capability Development
Installing and Sustaining the Council