These five AI prompts are built for the actual work lawyers do every day — client intake, case research, contract review, demand letters, and court prep. Each prompt includes specific instructions that produce usable output on the first try, not generic text that needs heavy editing. Copy them directly into ChatGPT, Claude, or your firm’s AI tool and adjust the bracketed fields for your case.
Key Takeaways
- Specific, structured prompts save lawyers 30–60 minutes per task compared to writing from scratch
- The best legal AI prompts include jurisdiction, practice area, and desired format to avoid generic output
- AI-generated legal text requires attorney review — use it as a first draft accelerator, not a final product
- Firms using AI for document drafting report 40% faster turnaround on routine legal work (Thomson Reuters 2024 survey)
Why Prompt Quality Matters for Legal Work
Generic prompts produce generic output. Asking an AI to “write a demand letter” gives you a template that reads like it was pulled from a law school textbook. Asking it to “draft a demand letter for a breach of commercial lease in Colorado, tenant is a restaurant that stopped paying rent after HVAC failure, landlord failed to repair within 14 days per lease Section 8.2, damages are $47,000 in lost revenue plus $12,000 in relocation costs” gives you something your paralegal can actually work with.
The difference is specificity. Legal work depends on jurisdiction, fact patterns, and precise language. Every prompt below is designed to capture those details so the AI output is relevant and useful, not a starting-from-zero exercise. Lawyers at firms using structured prompts report spending 65% less time editing AI-generated drafts compared to those using generic requests.
Prompt 1: Client Intake Summary
Use this after an initial consultation to generate a structured intake memo.
The Prompt:
You are a legal assistant at a [practice area] firm in [state]. Based on the following client intake notes, create a structured intake summary with these sections: Client Information, Case Summary (3-4 sentences), Key Facts (bullet points), Potential Legal Issues, Relevant Statutes or Regulations, Recommended Next Steps, and Conflicts Check Notes.
Client intake notes: [paste your notes from the consultation]
Format the output as a professional internal memo. Flag any information gaps that need follow-up.
Why this works: It gives the AI a clear role (legal assistant), specifies the output structure, includes jurisdiction context, and asks it to identify gaps — which catches things you might miss in a busy intake session. A family law attorney using this prompt reported cutting intake documentation time from 45 minutes to 12 minutes per client.
Prompt 2: Case Law Research Summary
Use this to get a structured overview of relevant case law for a specific legal issue.
The Prompt:
I need a research memo on [specific legal issue] in [jurisdiction]. Identify the leading cases, organize them chronologically, and for each case provide: case name and citation, court and year, key facts (2-3 sentences), holding, and relevance to my issue.
My specific fact pattern: [describe your case facts]
Focus on cases from the last 10 years. Distinguish between binding authority and persuasive authority. Note any circuit splits or unsettled areas of law. End with a brief analysis of how the case law applies to my fact pattern.
Why this works: It asks for structured output that matches how lawyers actually use case research. The chronological organization shows how the law has developed, and the binding vs. persuasive distinction prevents you from relying on authority that does not apply in your jurisdiction.
Important caveat: AI models can generate plausible-sounding but nonexistent case citations. Always verify every citation in Westlaw, LexisNexis, or your jurisdiction’s court records before relying on it. Use AI research as a starting direction, not as your final source. The Mata v. Avianca case in 2023 — where a lawyer submitted AI-fabricated citations — is a cautionary tale that every firm should remember.
Prompt 3: Contract Clause Review
Use this when reviewing a contract and you need to flag risky clauses or missing protections.
The Prompt:
Review the following contract from the perspective of [your client’s role: buyer/seller/tenant/licensee/etc.]. For each section, identify: clauses that are unfavorable to my client, missing protections my client should negotiate for, ambiguous language that could be interpreted against my client, and non-standard terms that deviate from typical [type of contract] agreements.
Jurisdiction: [state]. Governing law: [if specified in contract].
Contract text: [paste contract]
Present your findings as a redline summary table with columns for Section, Issue, Risk Level (Low/Medium/High), and Recommended Change. End with a prioritized list of the top 5 negotiation points.
Why this works: It specifies the perspective (your client’s side), asks for actionable output (recommended changes, not just observations), and includes a risk-level framework so you can prioritize what matters most. The table format makes it easy to share with your client or use as talking points in negotiation. Real estate attorneys using this approach report reviewing standard commercial leases in 20 minutes instead of 90 minutes.
Prompt 4: Demand Letter Draft
Use this to generate a first draft of a demand letter with the right tone and legal specificity.
The Prompt:
Draft a demand letter on behalf of [client name/description] against [opposing party] for [cause of action] in [state]. Use a firm but professional tone.
Key facts: [list the relevant facts, dates, amounts]
Damages sought: [itemize: compensatory, consequential, punitive, attorneys fees, etc. with amounts]
Relevant legal basis: [statutes, contract provisions, or common law theories]
Include: a statement of facts, legal basis for the claim, specific demand amount and deadline for response (14 days), consequences of non-compliance (litigation), and a professional closing that preserves the relationship if possible.
Format as a letter from [your firm name] on behalf of the client. Do not include a letterhead — I will add that.
Why this works: Demand letters follow a predictable structure, which makes them ideal for AI drafting. By providing specific facts, amounts, and legal theories, you get a draft that needs minor editing rather than a complete rewrite. The instruction about tone (“firm but professional”) and relationship preservation prevents the AI from producing overly aggressive language that could backfire in later negotiations or litigation.
Prompt 5: Deposition Preparation Outline
Use this to generate a structured deposition outline organized by topic and objective.
The Prompt:
Create a deposition outline for [deponent name and role] in a [type of case] case in [jurisdiction].
Case summary: [2-3 sentences about the case]
Key issues to explore: [list the topics you need to cover]
Documents the deponent has access to: [list key documents]
For each topic area, provide: an opening question (broad and open-ended), 5-7 follow-up questions that progressively narrow the focus, pin-down questions that lock the witness into specific facts, and impeachment setups if the witness is likely to be evasive on this topic.
Organize the outline to start with background and rapport-building questions, move through the key facts chronologically, and end with the most critical admissions. Flag questions where I should have a specific document ready to show the witness.
Why this works: Deposition preparation is time-intensive — attorneys typically spend 3–5 hours preparing for a single deposition. This prompt produces a structured outline that covers your key topics with a strategic progression from broad to specific. The impeachment setup section is particularly valuable because it helps you plan for witness evasion before you are in the room.
Tips for Using AI Prompts in Legal Practice
Always verify citations. This cannot be stated enough. AI models generate text that sounds authoritative, but case citations, statute numbers, and regulatory references may be fabricated. Verify every single legal reference before including it in any document that leaves your office.
Never input confidential client information into public AI tools. If you are using ChatGPT, Claude, or any cloud-based AI, check your firm’s data privacy policy first. Many firms use enterprise versions with data isolation guarantees, or they run AI tools locally to avoid any client data exposure. Your ethical obligations under Model Rule 1.6 do not disappear because the tool is convenient.
Use AI as a first-draft accelerator. The goal is not to eliminate attorney judgment — it is to eliminate the blank-page problem. AI gives you a 70% draft in minutes that you refine with your expertise and knowledge of the case. That is where the 40% time savings comes from.
Build a prompt library for your firm. Once you find prompts that work well for your practice area, save them. Standardized prompts across your firm ensure consistent quality and make it easy to train junior associates and paralegals on effective AI use.
Want to set up AI tools across your entire firm with proper security, prompt libraries, and workflow integration? See how we help law firms adopt AI, or book a free consultation to discuss your firm’s specific needs.
FAQ
Are AI-generated legal documents admissible in court?
AI-generated text has no special admissibility status. It is treated the same as any document — the attorney signing it takes full responsibility for its accuracy and legal validity. Courts have sanctioned attorneys who submitted AI-generated filings without verifying the content, so review everything carefully before filing.
Which AI tool is best for lawyers?
For general drafting and research, Claude and ChatGPT (GPT-4) produce the best output when given detailed prompts. For legal-specific needs, tools like CoCounsel (by Thomson Reuters), Harvey AI, and Clio’s AI features are purpose-built for legal workflows and include citation verification. The best tool depends on your practice area and volume of AI-assisted work.
How do I maintain attorney-client privilege when using AI?
Use enterprise-grade AI tools with data isolation agreements, or run AI models locally. Do not paste client information into free AI tools. Check your jurisdiction’s ethics opinions on AI use — several state bars have issued guidance. Document your AI use policies as part of your firm’s technology governance, and train all staff on what information can and cannot be entered into AI systems.
Can paralegals use these prompts too?
Absolutely. These prompts are designed to be used by anyone on the legal team. Paralegals can generate first drafts, and attorneys review and refine them. This workflow lets your firm handle higher volume without proportionally increasing attorney hours. Many firms report that paralegal-generated AI drafts, reviewed by an attorney, are 85–90% as good as attorney-written first drafts at a fraction of the time cost.
