AI supplier review — questions and required evidence

A discussion pack for procurement, technology, security and the operational owner. The actual contract and use require appropriate professional review.

Product and workflow

Which exact task does the product support? What does it refuse to do? Which actions need approval? Ask for examples of known failures and the conditions under which the supplier recommends human handover.

Data and operating model

Where are inputs, outputs and logs processed and stored? Who can access them? Are they used for model improvement? Ask for contractual terms, a processing description and configuration evidence, not only a sales statement.

Quality and change

How are results evaluated on your use case? Who maintains the evaluation set? Ask what happens when a model, dependency or knowledge source changes, and how a previous version can be recovered.

Security and incidents

Ask how access boundaries are enforced, how incidents are reported and who investigates disputed outputs. Confirm logging, monitoring, escalation and the ability to disable actions without losing the underlying service.

Exit and portability

Request a sample export and an explanation of what is portable. Identify dependencies that would need rebuilding. Include the cost and effort of switching in the buying decision, and record unanswered questions as open requirements.

AIStrategy.gr working resource · 6 November 2026. Adapt this template to the actual workflow and responsible teams.