vendor-ai-review

To review a vendor AI agreement for risks, perform a term-by-term extraction comparing training on data, liability, model change notice, and output ownership clauses against your internal governance positions to identify contractual gaps.

When checking an AI addendum during SaaS procurement, review confidentiality, incident notification, human review rights, use restrictions, audit rights, subprocessors, data residency, and termination clauses to ensure alignment with your governance playbook.

Stacked-vendor gap detection works by identifying upstream layers like SaaS wrappers, gateways, and foundation model providers, then checking flow-down enforceability to ensure upstream commitments hold across the vendor stack.

Yes, you can compare vendor AI terms against your CLAUDE.md playbook by running a consistency check that evaluates extracted provisions and assigns severity levels from aligned to critical per term based on your governance positions.

Flow-down enforceability risk in vendor AI agreements is the danger that upstream model provider or gateway commitments—such as data residency or incident notification—may not be contractually enforceable against the end customer.

You need a term-by-term AI governance review when deploying vendor AI or SaaS solutions during procurement and legal workflows, especially when agreements contain embedded terms of service or AI addenda with training-on-data and model change provisions.