Engagements
Start with the smallest engagement that can resolve the material uncertainty. Scale only when the institution has evidence that the approach works.
A closed working session for the leaders who own one proposed or operating AI use.
A decision is imminent, accountability is unclear, or the institution needs an independent view of the principal control and evidence gaps.
Install and operate the minimum governance layer around one material AI use.
The organisation needs to move beyond framework design and prove that decision rights, controls, review and evidence can function in a live or near-live workflow.
Design and implement how the organisation authorises, controls, changes and evidences AI use.
Policies or committees exist but lifecycle decisions, risk ownership, operating procedures and production evidence remain fragmented or immature.
Build or migrate the platform foundations with governance signals and operational ownership designed in.
AI delivery is constrained by platform architecture, data controls, observability, resilience or the inability to bind activity and cost to a governed use case.
How work is governed
Scope, assumptions, dependencies, decision owners, acceptance criteria and exit conditions are made explicit before delivery begins.
Resolve the decision, use case, system boundary, material stakeholders and evidence required.
Translate obligations and operating constraints into decisions, controls, architecture and acceptance criteria.
Work through real artefacts, technical changes, workflows, tests and accountable decisions.
Document, rehearse, train and step back only when the internal team can operate the capability.
Start with one real problem
We will test the fit before proposing a larger programme of work.