Design a technical due-diligence checklist for an AI pilot

Technical due diligence is evidence gathering under decision pressure. A checklist should expose which claims are verified, conditional or missing, rather than create a large count of yes boxes.

Build the decision artifact

Review intended use and prohibited use, data ownership and deletion, model versions and evaluation, integration identity and contracts, reliability and support, baseline and acceptance, vendor dependencies and exit. Sample artifacts back to source. Rate findings by consequence and remediation, and separate technical review from legal or regulatory advice.

The commercial leadership lab makes the artifact inspectable with authored inputs:

from leadership_cases import due_diligence_case

result = due_diligence_case()
assert result["evidence_items"] == 19
assert result["blocking_gaps"] == ["data deletion evidence"]
assert result["approve"] is False
assert result["legal_opinion"] is False

The fixture asks for nineteen evidence items across five domains. Missing deletion evidence is blocking, so approval remains false. The lab also states that the exercise is not a legal opinion.

Protect the decision from weak evidence

A polished demo can anchor the reviewer. Begin from material failure paths and request raw evidence, including failed tests and incidents. Check whether the evaluated build matches the proposed deployment and whether synthetic success transfers to permitted production data.

Keep these artifacts for review:

• system scope and version manifest
• domain checklist linked to artifacts
• finding severity, owner and remediation
• approve, condition or stop decision record

This practice aligns with the discovery, productisation, client enablement, technical leadership and capstone sequence in the FDE for Professionals course. The course link describes the pathway; the local scenario is fictional and does not claim a client engagement, investment result, hiring decision or certificate.

Practice task

Review a fictional pilot with one missing deletion path. Explain why other passing domains cannot compensate for it and what closes the gap.

Continue learning

This article is part of the Commercial judgement and delivery leadership sequence. Use the neighbouring tasks when you need the prerequisite or the next application.

• Review the prerequisite or neighbouring task in Negotiate a scope change using impact evidence.
• Continue with Present an AI investment recommendation with uncertainty.

Reference: NIST AI Risk Management Framework.