Organize
Evaluate AI vendors through evidence
Gather evidence and decide whether an AI vendor is ready for a governed proof of value.
Due diligence guide · 9-minute read · Updated
The executive decision
What evidence should a vendor provide before it earns a governed proof of value?
A defensible evaluation connects requirements, vendor answers, supporting evidence, control coverage, and a documented institutional judgment.
Use this guide to leave with
- A shared evidence standard across business, technology, risk, and finance
- A traceable evaluation that separates capability from assertion
- Clear conditions and stop signals before contracting or activation
Define what matters before the demos begin.
Translate priorities and risk appetite into a written evaluation structure before product polish influences the process.
Connect every judgment to something reviewable.
Record what was asked, what was answered, what was proven, and why the institution reached its conclusion.
Make cross-functional disagreement useful.
Use independent review and calibration to turn different assumptions into one documented institutional position.
Know which answers require more than another demo.
Withhold scope, data, traffic, or commitment until material gaps are resolved with evidence.
Questions worth putting on the agenda
- What evidence would we require if the demo were unavailable?
- Which capabilities are live, configured, or dependent on custom development?
- Can the institution operate, monitor, change, and stop the capability without waiting on the vendor?
- Which unresolved gaps change the acceptable scope of a proof of value?
This is a working discussion resource. Tailor it to your institution's policies, risk appetite, and legal or regulatory guidance before using it to support a decision.