The Workflow Edit | Practical AI Tool Tutorial

How to Use Harvey Without Generic AI Instructions

Stop reading every page of every contract by hand. Use legal AI to flag non-standard clauses and risky language with citations you can verify.

Dear Suzannah

Dear Suzannah, I need to review a stack of contracts for non-standard clauses before the end of the week, and manually reading every page is eating my entire afternoon. I keep missing things because my eyes glaze over by page 15. Is there a way to flag the risky clauses faster without sacrificing accuracy?

Upload the contracts and ask for flagged risks.

Harvey reads your contracts and flags non-standard clauses, risky language, and unusual terms with citations to the exact section. You review the flagged items instead of scanning every page, then verify the findings before acting.

The real use case

You have a batch of contracts — vendor agreements, NDAs, or lease documents — and you need to identify non-standard clauses and potential risks before the deadline. Instead of reading every page of every document manually, you upload them to Harvey and ask it to flag clauses that deviate from standard language for that contract type. The tool identifies the risks, cites the specific section, and summarizes the key terms. You verify the flagged clauses against the source text, prioritize the highest-risk items, and prepare your review notes.

The tool-specific prompt to use

After uploading your contracts to Harvey, paste this prompt to guide the review. Replace the bracketed items with your actual contract details before running.

Review the uploaded [contract type, e.g., vendor agreements, NDAs, lease documents] and flag every non-standard clause or term that deviates from typical language for this contract type. For each flagged item, provide: the document name, the section or clause number, the exact quoted language, a one-sentence explanation of why it is non-standard or risky, and the severity level (high, medium, low). Focus specifically on these areas: [list areas, e.g., termination rights, liability caps, indemnification, auto-renewal, payment terms, IP ownership]. For each document, also produce a summary of the five most important terms: parties, term length, payment structure, termination trigger, and liability or indemnification language. Compare the flagged clauses across all uploaded documents and note which contracts have the most risk. Do not paraphrase loosely — quote the key phrase for each flag. If a contract appears to be missing a standard clause that this type should include, note the omission. Rank all flagged items from highest to lowest risk at the end. Flag any clause that creates open-ended or uncapped obligations. Tell me if any clause references an external document or appendix I need to review separately.

Prompt length: 187 words.

Make the result less generic

  • Name the exact contract type so the tool knows what counts as standard versus non-standard.
  • List the specific clause areas you care about (liability, termination, indemnification) instead of asking for a general review.
  • Require direct quotes for every flagged clause so you can verify against the source text.
  • Ask for a severity ranking so you know which risks to address first.
  • Request a note for any missing standard clause, not just for clauses that are present and unusual.

Quick human check

  • Does every flagged item include the exact section number and quoted language you can verify?
  • Are the severity rankings consistent across contracts, or do some seem over- or under-rated?
  • Did the tool catch the clauses you already knew were risky, or did it miss any?
  • Are any flagged items false positives that you can dismiss after checking the source text?
  • Did you review every external document or appendix the tool flagged separately?

Sources and further reading

Whatever Time Finds
In-Person Business Workshop

Your competitor is already using this. Are you?

Learn how to use AI at Work to save hours, clean up repeat tasks, and build one AI-supported workflow your business can actually use.

Date
July 24, 2026
Time
9 AM to 1 PM
Location
Google Meet
Early Bird Seat
$297

Standard seat $497.
Total workshop value $3,750.

Bring your laptop, your AI tool logins, and one business process you are tired of doing the hard way.

Verified by MonsterInsights