vendor-ai-review

Share the vendor's AI addendum, main agreement AI provisions, or terms of service. The review extracts each key term — training on data, liability, model changes, output IP — and compares it against your governance playbook with severity ratings and proposed redlines.

The training-on-data clause is typically most critical, followed by confidentiality of inputs, liability for outputs, model change notification, and incident notification. For stacked vendors, flow-down of commitments from upstream model providers must also be verified.

Yes. A provisional mode runs the review against generic defaults — US jurisdiction, middle risk appetite, lawyer role — and tags every output as provisional. Configuring the practice profile via the cold-start interview produces tailored results.

No. The review covers AI-specific terms only and explicitly excludes DPA provisions, which require a separate DPA review. It does flag when a DPA exists without an AI addendum, since that leaves training, liability, and model-change risks unaddressed.

Terms outside the playbook's fallback positions are rated critical and routed per the escalation table rather than redlined. The review also documents gaps for renewal leverage and suggests fallback positions for each significant or critical finding.