
Every AI model makes trade-offs. In most organisations they are made informally and not documented. When a regulator, an auditor, or a claimant asks why the model was designed the way it was, the organisation needs an audit trail. The MI AI Governance Toolkit closes that gap with two registers. The Toolkit includes:
Trade-Off Register — Documents every material design choice in each AI system: what was traded, why, what alternatives were considered, the regulatory and values justification, and who approved it.
Exit Criteria Register — Defines, before deployment, the conditions under which each AI initiative will be Scaled, Restructured, or Stopped.
COSO x GenAI Reference — The COSO five-component internal control framework applied to AI governance: GenAI risks, expected controls, and specific audit tests for each component. For audit committees and internal auditors.
Board Summary — Quarterly summary view for governance board reporting, with status dropdowns and board actions required column.
For: Chief Risk Officers completing pre-deployment governance documentation. AI Governance Sponsors preparing evidence for board papers. Internal auditors needing auditable criteria. Company Secretaries documenting AI governance decisions.