Agentic AI Assurance and Audit Committee Judgement Adviser

The mandate

The audit committee's standing question is what assurance it can reasonably take from current agentic AI evaluations and control descriptions. The adviser will challenge the quality and limits of that evidence. The role is not a model safety certification and must not imply that an audit committee opinion authorises a management deployment.

Four days monthly are reserved for assurance evidence, a discussion with audit and risk owners and committee attendance. Attendance is included. A consequential evidence question receives acknowledgement within one working day and a preliminary view within three working days; a fuller opinion may require additional tests commissioned by authorised internal owners.

The retainer begins on 19 October 2026 and runs nine months. The audit chair will review renewal against the remaining AI assurance agenda and the independence of the first term. The adviser holds no line authority over audit or AI teams, and carries no executive responsibility for deployment or risk acceptance. Management and assurance owners retain their obligations.

The sponsor supplies authorised evaluation records, control ownership and existing assurance conclusions. Advice must distinguish a test of model capability from a test of an operating control, and a controlled scenario from the evidence needed for continuing reliance. Where assurance depends on human review, the practical capacity and behaviour of that review must be examined rather than assumed.

Concurrent non-competing advice is permitted. Vendor remuneration, a deployment role in the same programme or assurance work that would be reviewed by this committee creates a conflict requiring disclosure and possible recusal. Model development, operating remediation and formal audit opinions are excluded. The committee wants a GRC practitioner's independent judgement, not an adviser whose involvement is used to turn limited evidence into broad assurance.

What you will own

• Challenge assurance claims by tracing the evaluated behaviour to the operating risk, pressing owners to show whether the test actually addresses the committee's concern.
• Test reliance on human review against capacity and observed execution, advising where a nominal safeguard does not provide dependable control over an agent workflow.
• Shape questions about continuing assurance after model or workflow changes, identifying the evidence trigger that should invalidate an earlier positive conclusion.
• Examine control ownership and management acceptance boundaries, preventing committee discussion from being represented as authorisation of an operational risk exception.
• Press assurance teams to retain failure evidence and coverage limits, so favourable aggregate results do not conceal consequential scenarios that remain untested.
• Review proposed audit work for independence where teams helped build the AI control, advising how self-review risk should be disclosed or addressed.
• Record independent opinions and recusal limits in a committee note, leaving deployment, remediation and formal assurance conclusions with the authorised internal owners.

Candidate qualifications

• Evidence senior audit, GRC or assurance judgement involving AI or another technically complex transformation. Explain a claim you challenged, the test relevance assessed and how the committee's understanding changed.
• Demonstrate risk-control-test lineage competence with actual operating evidence. Candidates should describe a safeguard that existed in policy but lacked reliable execution, and show why the assurance conclusion required qualification.
• Provide experience examining model or workflow change effects on continuing assurance. Explain the trigger that invalidated earlier evidence and the additional work required before confidence could responsibly be restored.
• Show independence from vendors, deployment teams and self-review incentives. Describe a conflict or assurance boundary you disclosed, keeping specialist advice separate from management risk acceptance and formal audit certification.
• Be able to sustain four reserved days monthly and produce technically precise advice directors can use. Professional assurance capability must be evidenced. Candidates should show how a human-review assumption was tested, how contradictory results stayed visible and why the committee could not treat one favourable evaluation as universal assurance over autonomous operation. Explain an assurance opinion that remained conditional because the operating control was changed after evaluation. Show how you identified the invalidated evidence, framed the next useful audit question and prevented a committee summary from implying continued confidence without a fresh test.

Application

Applications for this mandate are received in one way only: through the India Board Terminal's application process. It is automated end to end. Your Executive Passport travels to the mandate holder in its confidential form, your answers to the three questions below are read before anything else in your file, and every stage that follows is recorded on your applications page.

There is no address to write to and no intermediary to call. The mandate holder reads what the Terminal delivers and nothing else, which is what keeps the process the same for every applicant and keeps your name out of it until you release it. Applications close on 13 October 2026. Mandate reference PCT-ADV-2026-IND-16.