← All engagements

AI Decision Audit

A review of a specific investment, procurement or deployment decision before the team commits budget or authority. The audit examines the rationale, evidence and controls and identifies which unknowns could change the choice.

What to bring

A decision owner, the options under consideration, the decision deadline, the business case and relevant non-confidential evidence. Include owners of the workflow, risk controls and exception handling.

Example structure

  1. Scope the decision

    Agree the options, constraints, decision owner and the evidence needed by the commitment date.

  2. Review the case

    Trace the most important assumptions, stakeholder incentives and control gaps in the proposed workflow.

  3. Challenge the shortlist

    Compare a limited pilot, a staged rollout and other available options. Set thresholds for the findings that would alter the recommendation.

  4. Return the decision memo

    Review the evidence, unresolved questions and suggested tests with the accountable owner.

What you take away

  • A concise decision memo with options and evidence gaps
  • An assumption and risk register with owners
  • A short list of pre-commitment tests with decision criteria

Example scenario

Before licensing an automated claims triage system, review whether the pilot can support the forecast and whether reviewers can still catch the cases where classification has material consequences.

Discuss this engagement