Indicative duration
One day for a foundation workshop or two days for a use-case-based governance lab.
AI Governance and Evaluation
Define how delegated AI work should be owned, tested, approved, observed, changed, suspended, and improved in practice.
At a glance
Intended audience
Business owners, transformation leads, risk, compliance, IT, security, and solution teams.
Prerequisites
An agreed use case, relevant policies, named decision owners, and available solution information.
Take-home output
A draft control matrix, evaluation plan, approval map, or operating responsibility model.
Three levels of governance
Organisational governance
Risk appetite, acceptable use, accountability, oversight, incident handling, and change control.
Solution governance
Identity, ownership, permissions, approval conditions, evaluations, traceability, and suspension.
User practice
Responsible delegation, data handling, verification, escalation, and reporting.
What the programme covers
- Agent identity, ownership, and tool permissions
- Approval conditions and human judgment points
- Evaluation criteria, test scenarios, and monitoring
- Prompt injection, data leakage, and misuse risks
- Traceability, versioning, and change control
- Suspension conditions and incident response
Practical output
Participants or project sponsors leave with a draft governance artefact tied to the agreed use case. Depending on scope, this may be a control matrix, evaluation plan, approval map, operating responsibility model, or incident scenario.
A workshop develops capability and draft artefacts. It does not claim to implement production controls across an organisation.