How to Review Any Contract with AI with Eigent

How it works

1. Set up your playbook first for MSA and DPA reviews — the review runs against your stated non-negotiables. For NDAs, you can proceed without one and the skill falls back to CUAD risk categories.
2. Choose the right command for your document type: /review for NDAs, /commercial-legal:review for vendor MSAs, /privacy-legal:dpa-review for DPAs, and /ai-governance-legal:vendor-ai-review for AI vendor terms.
3. Paste the full contract text and state your role explicitly — customer vs. vendor, controller vs. processor, receiving vs. disclosing party — before any clause analysis.
4. Read the output starting with critical and high-severity issues — these are the ones to address before anything else.
5. Take critical and high findings into redlines, or feed the full output into scoring for a sign-off recommendation.

More prompts to try

• /commercial-legal:review — Review this SaaS agreement [paste]. We're the customer. Check liability caps, data ownership, unilateral change clauses, and auto-renewal terms. How does it compare to market norms?
• /privacy-legal:dpa-review — Review this DPA [paste]. We are the controller. Produce a complete GDPR Article 28 compliance checklist — mark each required provision as present, absent, or inadequate.
• /ai-governance-legal:vendor-ai-review — Review these AI vendor terms [paste]. Flag any clause that grants training rights on our data, and assess EU AI Act exposure for our use case.
• Focus the review only on data rights and liability — skip everything else and go deep on those two areas.

How to use

Use the matching command for the document type — using the wrong command applies the wrong framework and produces inaccurate results. Paste the full contract, not a summary or excerpt; the clause mapping depends on reading the whole agreement, and cross-references between clauses matter. State governing law if known — enforceability of liability caps, non-competes, and data rights varies significantly by jurisdiction. For AI vendor agreements, always check training data rights explicitly — this is the highest-frequency issue in AI vendor contracts and the most commonly missed in manual reviews. Run your playbook setup before any MSA or DPA review to get playbook deviation flags alongside market benchmarks.

Expected output

A scored, severity-ranked issue list with clause citations, one-sentence risk explanations, market benchmarks for key terms, and — for MSAs — a playbook deviation table. Critical and high issues appear first, medium and low issues follow.

Limitations

• One document at a time — the review tool is optimized for single-document analysis; batch input produces inaccurate clause mapping.
• Market benchmarks reflect general commercial norms, not live deal data — "below market" and "unprecedented" call for different responses.
• For very long contracts that exceed the context window, paste section by section and ask the skill to consolidate findings at the end.