Prompt Pilotby Write.law
All prompts

Analyze/Featured

Analyze a contract for real risk

Reads a commercial agreement the way it will read when the relationship breaks, and asks you the business questions first, so the review is about your deal.

About 20 minintermediateTransactional, In-house

Your prompt4,446 characters

Still to fill in: Agreement text, Governing law / forum, Who you represent

RoleYou are a commercial lawyer who has reviewed several thousand agreements and been on the receiving end of the ones that went wrong. You read for what happens when the relationship breaks, not for what the parties hope will happen. You flag the clause that will matter in a dispute, not the clause that is merely unusual, and you never pad a review with observations that carry no consequence.What I needReview the agreement below on behalf of Who you represent under Governing law / forum, and tell me what actually carries risk. Calibrate how hard you push to aggressive.InputsAgreement: Agreement text Governing law / forum: Governing law / forum Who I represent: Who you represent Counterparty: Counterparty Risk posture: aggressiveHow to work this1. Read for the breakup first: termination, limitation of liability, indemnity, IP ownership, data rights and security, dispute resolution, change of control. 2. Rank every flag by consequence to Who you represent, not by how unusual the drafting is. A standard-looking cap sitting below realistic exposure outranks an odd but harmless clause. 3. Sort every flagged term into exactly one bucket: Must change, Should push, or Accept. 4. Quote the operative language and name the section for every flag. No paraphrase-only findings. 5. For each Must change and Should push, supply the actual redline language you would send, not a description of what to ask for. 6. Name what is absent: missing limitation of liability, missing assignment restriction, missing survival clause, missing security or breach-notification obligation. 7. Where the agreement's governing law materially changes the analysis and you are unsure of that jurisdiction's rule, say so rather than assuming.Ask me firstBefore you produce anything, ask me these questions, then stop and wait. I have already given you the documents above. These are the things the paper cannot tell you: 1. What is this deal worth to my client, and what is the realistic worst-case exposure if it goes wrong? A liability cap is only assessable against a number. 2. Does my client have leverage here, or do we need this deal more than Counterparty does? 3. Which terms are already negotiated or genuinely non-negotiable, so I do not spend redlines on settled ground? 4. Has this counterparty performed for us before, and if so, what went wrong? Do not begin the review until I answer. If I tell you to proceed anyway, state each assumption you are making at the top of your output and mark it [ASSUMPTION - verify].Output formatOpen with a three-sentence bottom line: can my client sign this, and what is the single biggest problem. Then the three buckets, each flag with section number, quoted language, why it matters to this deal, and proposed redline. Close with "What's missing" and the assumptions you had to make. End with one line naming the two of my answers that most changed this review, and what you would have concluded without them. If an answer changed nothing, say so. It means I should not have been asked.Never do this- If your review would apply equally to any client in any jurisdiction, it is too generic. Start over from this deal's numbers and this client's position. - No hedging filler. Cut "arguably," "it should be noted," "it depends," and "this clause is fairly standard" as a substitute for analysis. Do not tell me to consult an attorney. I am the attorney. - Every statutory or case reference must come from my inputs or be marked [UNVERIFIED - check before relying]. Never invent a citation or quote a statute you are not certain of. - Where you do not know how Governing law / forum treats a clause, say you do not know. Do not smooth over the gap with fluent prose. - Do not pad. A clean agreement with two real problems gets a short review. Length is not value.Before you answer- Did I quote actual contract language for every flag, with a section number? - Is every Must change and Should push accompanied by redline text I could paste into the document? - Have I assessed the liability cap against the deal's actual numbers rather than in the abstract? - Would this review be useless to someone reviewing a different contract? It should be. - Have I listed what is missing from the agreement, not just what is in it?

Adds driver's-seat tunes: options instead of answers, questions before work, every citation flagged. Your values come with it.

2

Pressure-test it

Makes the AI switch hats and attack its own answer.

Hand this review across the table. The lawyer who drafted the paper for Counterparty has seen every one of these objections before and keeps a rebuttal ready for each; answer as the drafter, not as me. For every Must change item, write the pushback you would give and rate whether my position survives it. Then tell me which one flag I should actually spend my leverage on, and which ones I am likely to lose without damaging the relationship.
3

Go deeper

Pushes the work further once the basics are right.

Whoever opens this on Counterparty's side will read the email before the redline. Turn the Must change and Should push items into that cover email: a short, non-adversarial note that leads with agreement, groups the asks into the two or three that matter, and gives a business reason for each rather than a legal one. Keep it under 250 words.

Before you run it

What to gather first

  • The full agreement text, including exhibits and any incorporated terms
  • Governing law and forum from the agreement itself
  • What the deal is worth and the realistic downside if it fails
  • Whether your side has leverage in this negotiation
  • Which terms are already settled and off the table

Watch for

  • The model reads only what you paste. Exhibits, order forms, and incorporated policies usually carry the real obligations. Omit them and the review is wrong.
  • Do not paste client-confidential agreements unless your firm's AI policy and the client's engagement terms permit it.
  • Enforceability of caps, indemnities, and non-competes varies sharply by governing law. Confirm the rule in the actual jurisdiction before relying on any conclusion here.
  • The model is weak at spotting what is absent from a contract. Check independently for missing survival, assignment, and insurance provisions.
  • A clause that is "market" is not automatically acceptable for your client. Market terms still lose deals when exposure is asymmetric.

What comes back

Opens with a three-sentence bottom line: can my client sign, and what is the single biggest problem. Then three buckets (Must change, Should push, Accept), each flag carrying a section number, quoted language, why it matters to this deal, and paste-ready redline text. Closes with "What's missing" and the assumptions the review rests on.

See an example of what you’ll get
*(After you answer the four questions, say, $1.2M annual contract, worst case is a payments outage costing roughly $4M, we have moderate leverage, no prior relationship.)* Bottom line. Do not sign as drafted. Liability is capped at 12 months of fees (roughly $1.2M) against realistic exposure near $4M, and the cap covers the indemnity too. Everything else is negotiable noise by comparison. MUST CHANGE § 11.2: Limitation of liability. *"In no event shall Vendor's aggregate liability exceed the fees paid in the twelve (12) months preceding the claim."* Your realistic outage exposure is roughly 3.3× this cap, and § 11.3 makes the cap apply to the § 9 indemnity as well, so even a third-party data-breach claim is capped at $1.2M. > Redline: "...exceed the greater of (a) three times (3×) the fees paid in the twelve (12) months preceding the claim, or (b) $5,000,000. The foregoing cap shall not apply to Vendor's obligations under Section 9 (Indemnification), breach of Section 7 (Data Security), or Vendor's gross negligence or willful misconduct." SHOULD PUSH § 14.1: Assignment. Vendor may assign freely on change of control; you may not assign at all. Given they are Series C, an acquisition is plausible inside the term. > Redline: "Neither party may assign this Agreement without the other party's prior written consent, except that either party may assign to a successor in interest to all or substantially all of its assets, provided such successor is not a competitor of the other party." ACCEPT § 18.4: Delaware forum. Not ideal from California, but you have no leverage worth spending here and Delaware is a competent forum. WHAT'S MISSING - No security-incident notification deadline. § 7 requires "prompt" notice: meaningless. Ask for 48 hours. - No survival clause. As drafted, confidentiality and indemnity obligations end at termination: § 16 lists no surviving sections. - No insurance requirement, which for a payments vendor at this exposure level is unusual. Assumptions I had to make. Exhibit B was not included in what you pasted. If it contains an SLA with service credits, § 11.2 may interact with it in ways I cannot see [verify]. I assumed Delaware law governs enforceability of the cap [safe - § 18.1]. What your answers changed. The $4M exposure figure is what moved the liability cap from a footnote to the headline. At a $200k worst case this agreement is signable as drafted. Your "no prior relationship" answer is why § 14.1 assignment made the list at all. Your leverage answer changed nothing here; I would have flagged the same three items either way.
Why this prompt is built the way it is
## Framework 1. **Ask before reading.** Business context determines what counts as risk. A cap on liability that is fine for a $50k deal is negligent for a $5M one. 2. **Read for the breakup.** Termination, liability, indemnity, IP ownership, data, dispute resolution, and change-of-control are where deals actually hurt. 3. **Rank by consequence, not by oddity.** An unusual clause with no exposure is a footnote. A standard-looking cap that sits below realistic exposure is the headline. 4. **Three buckets.** Must change (walk away or fix), should push (worth a redline), accept (note and move on). Every flagged term lands in exactly one. 5. **Quote the language.** Every flag quotes the operative words and names the section. 6. **Give the ask, not just the problem.** For each "must change" and "should push," supply the redline language you would actually send. 7. **Name what is missing.** Absent clauses are risk too: no limitation of liability, no assignment restriction, no survival clause.