AI Prompts for Lawyers: 10 Structured Templates for Legal Work

A 2024 Thomson Reuters survey found that 79% of legal professionals have already used generative AI in their work. The question is no longer whether lawyers will use AI — it is whether they will use it well.

Most lawyers who try AI get mediocre results. They type "summarize this contract" or "write a motion to dismiss" and get output that reads like a first-year associate's first draft — technically adjacent but missing the precision, jurisdiction-awareness, and structural rigor that legal work demands.

The difference is the prompt. A vague instruction produces vague output. A structured prompt that specifies the legal framework, jurisdiction, audience, and output format produces work product that is actually useful as a starting point.

These 10 templates cover the tasks lawyers do repeatedly — contract review, legal research memos, case summarization, client communication, and litigation preparation. Each one is designed to produce structured, reviewable output that accelerates your workflow without replacing your judgment.

Every template has been tested across GPT-4o, Claude, and Gemini. Fill in the bracketed variables and you have a working prompt in under a minute.

Before You Start: AI Ethics for Legal Professionals

Never input confidential client data into public AI tools. This is not a suggestion. Attorney-client privilege and duty of confidentiality obligations apply regardless of the tool. Use anonymized facts for prompt development, or use an enterprise AI deployment with appropriate data handling agreements.

AI output requires attorney review — always. ABA Formal Opinion 512 (2024) establishes that lawyers may use AI but must supervise its output, understand its limitations, and take responsibility for work product. AI is a drafting assistant, not a co-counsel.

Verify every citation. LLMs hallucinate case law. They will fabricate case names, create plausible-sounding citations to cases that do not exist, and misstate holdings of real cases. Every citation in AI-generated output must be verified through Westlaw, LexisNexis, or another authoritative source. Our anti-hallucination guide covers techniques to reduce this risk.

Model notes. Claude tends to be more cautious and qualified in legal analysis, which is appropriate for advisory work. GPT-4o is strong at structured analytical output. Neither model is a substitute for legal research databases. For detailed model differences, see our comparison guide.

The 10 Templates

1. Contract Review and Risk Analysis

Contract review is the single most time-consuming task in transactional practice. This template turns a first-pass review into a structured risk assessment with flagged clauses and missing provisions.

Review the following contract and provide a risk analysis.

Contract type: [NDA / MSA / SaaS Agreement / Employment
Agreement / Lease / [OTHER]]
My client's role: [PARTY A — e.g., "the vendor" / "the tenant"
/ "the employer"]
Jurisdiction: [STATE / COUNTRY]
Client's priority: [e.g., "limit liability exposure" /
"protect IP" / "ensure termination flexibility"]

Contract text:
[PASTE CONTRACT TEXT]

Provide the following analysis:

1. **Key Terms Summary**
   - Parties, effective date, term, renewal provisions
   - Payment terms and amounts
   - Scope of services or obligations

2. **Risk Analysis** (from my client's perspective)
   For each material clause:
   - Clause reference (section number)
   - Current language (quote or paraphrase)
   - Risk level: High / Medium / Low
   - Why it's a risk
   - Suggested revision (specific alternative language)

3. **Missing Provisions**
   - Standard clauses for this contract type that are absent
   - Why each matters for my client
   - Recommended language to add

4. **Unusual or Non-Standard Terms**
   - Clauses that deviate from market standard
   - Whether the deviation favors my client or the counterparty

5. **Negotiation Priority List**
   - Ranked list of items to negotiate, from most to least critical
   - For each: what to ask for and what to concede if needed

Output as a structured memo. Flag any clauses where the
analysis depends on facts I haven't provided.

Pro tip: "From my client's perspective" is the critical phrase. Without it, the model produces a neutral analysis. With it, the output is oriented toward your client's interests — which is what a first-pass review should deliver.

2. Legal Research Memo (IRAC Format)

The IRAC memo is the foundation of legal analysis. This template produces a structured research starting point — not a replacement for actual legal research, but a framework that saves hours of initial organization.

Draft a legal research memo in IRAC format.

Issue: [STATE THE LEGAL QUESTION — e.g., "Whether a
non-compete clause is enforceable against an independent
contractor under California law"]
Jurisdiction: [STATE AND/OR FEDERAL CIRCUIT]
Area of law: [CONTRACT / TORT / EMPLOYMENT / IP / CORPORATE
/ REGULATORY / [OTHER]]
Relevant facts:
[BULLET LIST OF KEY FACTS]
Client's position: [WHAT OUTCOME THE CLIENT WANTS]

Structure:

1. **Issue**
   - Restate the legal question precisely
   - Identify sub-issues if applicable

2. **Rule**
   - Governing statute(s) and key provisions
   - Leading case law (name, court, year, holding)
   - Any relevant regulatory guidance
   - Majority vs. minority positions if there's a split

3. **Application**
   - Apply each rule to the specific facts
   - Distinguish favorable and unfavorable precedent
   - Address counterarguments the opposing side would raise

4. **Conclusion**
   - Likely outcome with confidence level
   - Key factors that could change the analysis
   - Recommended next steps

CRITICAL: For every case cited, indicate whether you are
confident it exists or whether it should be verified.
Do NOT fabricate case names or citations. If you are unsure
about a specific case, describe the legal principle and note
that a Westlaw/Lexis search is needed to confirm the citation.

Length: 1500-2000 words.
Tone: Objective analytical memo, not advocacy.

Pro tip: The "CRITICAL" instruction about case citations is the most important part of this template. Without it, every major LLM will fabricate cases. Even with it, you must verify every citation. Use this output as a research roadmap, not a finished memo. For more on structuring effective multi-step prompts, see our prompt chaining guide.

3. Case Summarization

Lawyers read cases constantly but rarely have time to create organized summaries. This template extracts the information that matters for your specific purpose.

Summarize the following case.

Case: [CASE NAME, CITATION, COURT, YEAR — or paste the full
opinion text below]
My purpose for reviewing this case: [e.g., "researching
enforceability of arbitration clauses in employment
agreements" / "preparing for motion to dismiss"]

[PASTE CASE TEXT IF NOT A WELL-KNOWN CASE]

Provide:

1. **Case Header**
   - Full citation
   - Court and date
   - Procedural posture (what stage, what motion)
   - Judge (if relevant)

2. **Facts** (5-8 bullet points)
   - Only facts material to the holding
   - Organized chronologically

3. **Issue(s)**
   - Each legal question the court addressed

4. **Holding**
   - The court's answer to each issue
   - The rule established or applied

5. **Reasoning**
   - Key reasoning steps (3-5 bullets)
   - Precedent relied upon
   - Policy considerations cited

6. **Relevance to My Matter**
   - How this case supports or undermines my position
   - Key facts that distinguish it from my case
   - Whether this is binding or persuasive authority
     in [MY JURISDICTION]

7. **Key Quotes** (2-3 quotable passages for briefs)

IMPORTANT: If I have not pasted the case text and you are
working from training data, clearly state this and note
that all details should be verified against the official
reporter.

Pro tip: "Relevance to my matter" is what transforms a generic summary into a useful work product. Always include your purpose — it forces the model to analyze, not just summarize.

4. Client Communication (Plain Language Translation)

Translating legal analysis into language clients understand is a skill. This template bridges the gap between legal precision and client accessibility.

Translate the following legal analysis into a client-friendly
communication.

Legal analysis/situation:
[PASTE YOUR LEGAL ANALYSIS, MEMO EXCERPT, OR DESCRIBE
THE SITUATION IN LEGAL TERMS]

Client profile:
- Name: [CLIENT NAME]
- Sophistication: [BUSINESS OWNER / INDIVIDUAL / CORPORATE
  EXECUTIVE / TECHNICAL EXPERT]
- Emotional state: [e.g., "anxious about litigation" /
  "eager to close deal" / "frustrated by delay"]
- Communication preference: [EMAIL / LETTER / TALKING POINTS
  FOR A CALL]

Write a [EMAIL / LETTER / SET OF TALKING POINTS] that:

1. Explains the situation in plain language
   - No legal jargon without definition
   - Use analogies where helpful
   - Lead with what matters to the CLIENT, not the legal theory

2. Presents options clearly
   - For each option: what it means, what it costs (time and
     money), likelihood of success, and risk
   - Use a simple comparison format

3. States your recommendation
   - One clear recommendation with reasoning
   - What happens if they do nothing

4. Defines next steps
   - What you need from the client (decisions, documents, signatures)
   - Timeline and any deadlines
   - What you will do next

Tone: [REASSURING / DIRECT / FORMAL / CONVERSATIONAL]
Length: Under [500 / 750 / 1000] words.
Do not include legal citations in the client communication.

Pro tip: The "emotional state" variable is often overlooked but dramatically changes the output. A client facing litigation needs reassurance before information. A client closing a deal needs speed and decisiveness. Matching tone to emotional state is what separates adequate client communication from excellent client communication.

5. Regulatory Compliance Checklist

Compliance requirements are extensive, jurisdiction-specific, and constantly changing. This template generates a structured starting point for compliance review.

Generate a regulatory compliance checklist for [COMPANY/CLIENT].

Industry: [HEALTHCARE / FINANCIAL SERVICES / TECHNOLOGY /
MANUFACTURING / [OTHER]]
Regulation(s): [HIPAA / SOX / GDPR / CCPA / AML-BSA / OSHA /
FDA / [SPECIFIC REGULATION]]
Jurisdiction: [FEDERAL / STATE(S) / INTERNATIONAL]
Company size: [EMPLOYEES / REVENUE RANGE]
Current status: [NEW COMPLIANCE PROGRAM / ANNUAL REVIEW /
POST-INCIDENT REVIEW / PRE-AUDIT PREPARATION]

Generate:

1. **Regulatory Requirements Summary**
   - Key obligations under [REGULATION]
   - Applicability thresholds (does this apply to this company?)
   - Effective dates and upcoming changes

2. **Compliance Checklist** (organized by category)
   For each requirement:
   - [ ] Requirement description
   - Responsible role/department
   - Evidence needed to demonstrate compliance
   - Common deficiency (what auditors/regulators typically find)
   - Status: [TO ASSESS]

3. **Documentation Requirements**
   - Policies and procedures that must exist
   - Record retention requirements and periods
   - Training documentation

4. **Risk Areas**
   - Top 5 areas where companies in [INDUSTRY] most commonly
     fail compliance
   - For each: why it happens and how to prevent it

5. **Timeline and Milestones**
   - Recurring compliance deadlines
   - Filing dates
   - Training/certification renewal dates

IMPORTANT: This checklist is a starting framework. Regulatory
requirements change frequently. Verify all requirements
against current regulations and agency guidance.

Pro tip: Always include "current status" — the checklist for building a compliance program from scratch is very different from an annual review or post-incident remediation.

6. Due Diligence Question List

Due diligence is a structured information-gathering exercise, and the question list is its backbone. This template generates comprehensive questions tailored to the transaction type.

Generate a due diligence question list for a [TRANSACTION TYPE].

Transaction type: [M&A ACQUISITION / REAL ESTATE PURCHASE /
CORPORATE INVESTMENT / JOINT VENTURE / [OTHER]]
Target: [TARGET COMPANY/PROPERTY DESCRIPTION]
Industry: [INDUSTRY]
Deal size: [APPROXIMATE VALUE]
My client's role: [BUYER / SELLER / INVESTOR / LENDER]
Key concerns: [e.g., "IP ownership unclear" / "environmental
liability" / "key person dependency" / "pending litigation"]

Generate due diligence questions organized by category:

1. **Corporate / Entity Structure**
   - Formation documents, good standing, subsidiaries
   - Ownership, cap table, voting agreements
   - Minutes and board resolutions

2. **Financial**
   - Historical financials, audit reports
   - Working capital, debt obligations
   - Tax returns and compliance

3. **Contracts and Commitments**
   - Material contracts, change of control provisions
   - Customer and supplier agreements
   - Lease obligations

4. **Intellectual Property**
   - Owned vs. licensed IP, registration status
   - Employee/contractor IP assignment agreements
   - Open source usage (if technology company)

5. **Litigation and Regulatory**
   - Pending or threatened litigation
   - Regulatory investigations or consent orders
   - Insurance coverage and claims history

6. **Employment and Benefits**
   - Key employees, employment agreements
   - Non-compete and non-solicitation agreements
   - Benefit plans, ERISA compliance

7. **[INDUSTRY-SPECIFIC CATEGORY]**
   - Questions specific to [INDUSTRY]

For each question:
- State the question clearly
- Note why it matters for this transaction
- Flag if it's a "deal killer" question vs. informational

Prioritize: Mark the top 10 questions as "Day 1 requests"
that should be asked immediately.

Pro tip: "Key concerns" is the variable that separates a generic due diligence list from a useful one. If you know the target has pending litigation, the question list should drill deep into litigation history, insurance coverage, and indemnification. Always front-load your known risk areas.

7. Legal Brief Outline

Drafting a brief from scratch is slow. Starting with a structured outline that maps your arguments, identifies the standards, and sequences the analysis cuts the writing time significantly.

Create an outline for a [BRIEF TYPE].

Brief type: [MOTION TO DISMISS / MOTION FOR SUMMARY JUDGMENT /
OPPOSITION TO [MOTION] / APPELLATE BRIEF / MEMORANDUM OF
LAW / [OTHER]]
Court: [COURT NAME AND JURISDICTION]
Case: [CASE NAME AND NUMBER]
My client's position: [PLAINTIFF / DEFENDANT / APPELLANT /
APPELLEE / MOVANT]
Relief sought: [WHAT YOU WANT THE COURT TO DO]

Key facts:
[BULLET LIST OF MATERIAL FACTS]

Legal arguments:
[BULLET LIST OF YOUR ARGUMENTS — even rough notes are fine]

Opposing arguments you anticipate:
[WHAT THE OTHER SIDE WILL ARGUE]

Generate:

1. **Introduction** (outline)
   - 2-3 sentence framing of the case
   - Statement of relief sought
   - Why the court should grant it (thesis)

2. **Statement of Facts** (outline)
   - Chronological fact organization
   - Which facts support each argument
   - Disputed vs. undisputed facts

3. **Legal Standard**
   - Applicable standard for [BRIEF TYPE] in [JURISDICTION]
   - Key cases establishing the standard

4. **Argument Structure**
   - Argument I: [STRONGEST ARGUMENT]
     - Sub-point A, B, C with supporting authority
   - Argument II: [SECOND ARGUMENT]
     - Sub-point A, B, C
   - Argument III: [THIRD ARGUMENT if applicable]
   - Rebuttal to anticipated opposing arguments

5. **Conclusion**
   - Summary of arguments
   - Specific relief requested

For each section, note:
- Estimated length (word count)
- Cases to research (describe the principle — I will verify
  citations independently)
- Strategic notes (what to emphasize, what to minimize)

Local rules: Flag any [COURT] formatting or procedural
requirements I should check (page limits, font, margins,
certificate of service requirements).

Pro tip: Including "opposing arguments you anticipate" produces dramatically better outlines. The model will weave rebuttals into your affirmative arguments rather than treating them as an afterthought — which is exactly how effective brief writing works.

8. Deposition Preparation Questions

Deposition prep is part strategy, part organization. This template generates questions organized by topic with strategic annotations for the examining attorney.

Generate deposition preparation questions for [DEPONENT].

Deponent: [NAME, ROLE — e.g., "Jane Smith, former VP of
Engineering at TargetCo"]
Case type: [EMPLOYMENT / PERSONAL INJURY / COMMERCIAL
LITIGATION / IP / CORPORATE / [OTHER]]
My client: [PLAINTIFF / DEFENDANT]
Key issues in the case:
[BULLET LIST OF ISSUES TO EXPLORE]
Documents to reference:
[LIST KEY DOCUMENTS THE DEPONENT AUTHORED, RECEIVED, OR
IS MENTIONED IN]
Known facts favorable to our position:
[BULLET LIST]
Known facts unfavorable to our position:
[BULLET LIST]

Generate questions organized by topic area:

1. **Background and Foundation**
   - Role, responsibilities, relevant time period
   - Reporting structure and decision-making authority

2. **[TOPIC AREA 1 — e.g., "Events Leading to Termination"]**
   - Open-ended questions first (let them talk)
   - Narrowing questions (pin down specifics)
   - Document-specific questions ("I'm going to show you
     Exhibit [X]...")
   - Impeachment setup (if prior statements exist)

3. **[TOPIC AREA 2]**
   - Same structure

4. **[TOPIC AREA 3]**
   - Same structure

5. **Cleanup and Preservation**
   - "Is there anything else..." catch-all questions
   - Document preservation questions
   - Questions about communications with counsel (to test
     privilege claims, not to invade privilege)

For each question:
- Strategic note: [WHY you're asking this — what answer
  helps, what answer hurts]
- Follow-up if they say YES: [QUESTION]
- Follow-up if they say NO: [QUESTION]
- Follow-up if they say "I don't recall": [QUESTION]

Flag questions where the answer could hurt our position
and note the risk.

Pro tip: The "follow-up if they say..." structure is what separates prepared deposition questioning from improvised questioning. Anticipating answers means you are never caught without a next question. For building multi-step prompt sequences like this, see our guide on prompt chaining.

9. Contract Clause Drafting

Drafting contract language requires balancing your client's interests, enforceability, and negotiation strategy. This template produces three versions at different aggressiveness levels — so you can choose based on leverage and relationship.

Draft a [CLAUSE TYPE] clause for a [CONTRACT TYPE].

Clause type: [INDEMNIFICATION / LIMITATION OF LIABILITY /
TERMINATION / NON-COMPETE / CONFIDENTIALITY / FORCE MAJEURE /
IP OWNERSHIP / DISPUTE RESOLUTION / [OTHER]]
Contract type: [MSA / EMPLOYMENT AGREEMENT / SaaS AGREEMENT /
LICENSE / PARTNERSHIP / [OTHER]]
My client's role: [THE PARTY THIS CLAUSE SHOULD FAVOR]
Jurisdiction: [STATE / COUNTRY]
Specific requirements:
[ANY SPECIFIC TERMS — e.g., "24-month non-compete, 50-mile
radius" or "mutual indemnification with carve-outs for
gross negligence"]

Generate three versions:

1. **Aggressive** (strongly favors my client)
   - Maximum protection for our side
   - Note: which provisions the counterparty will likely push back on
   - Where this version might not be enforceable

2. **Balanced** (commercially reasonable)
   - Fair to both parties
   - Most likely to be accepted without heavy negotiation
   - Industry standard for [CONTRACT TYPE]

3. **Concessive** (minimum acceptable position)
   - The least favorable version we should accept
   - Red lines: what to never concede below this version
   - When this version makes strategic sense (preserving
     relationship, closing quickly)

For each version:
- The actual clause language (ready to paste into a contract)
- Plain language explanation of what it means
- Key enforceability considerations in [JURISDICTION]
- One case or statute that supports this approach (describe
  the principle — I will verify the citation)

Pro tip: Having three versions prepared before negotiation begins is a significant tactical advantage. You know your walk-away position before the conversation starts, and you can concede from aggressive to balanced while the counterparty feels they are winning. For structuring prompts that produce consistent multi-variant output, see our few-shot prompting guide.

10. Litigation Timeline and Fact Chronology

Every litigator builds timelines. They are essential for case evaluation, brief writing, and trial preparation. This template organizes raw facts into a structured, citation-linked chronology.

Create a litigation timeline and fact chronology.

Case: [CASE NAME / DESCRIPTION]
Key parties:
[LIST PARTIES AND THEIR ROLES]
Relevant time period: [START DATE] to [END DATE]

Raw facts and events:
[PASTE OR LIST ALL KNOWN FACTS, EVENTS, COMMUNICATIONS,
AND ACTIONS — in any order. Include dates where known,
approximate dates where not.]

Source documents:
[LIST KEY DOCUMENTS WITH DATES — emails, contracts, letters,
meeting notes, etc.]

Generate:

1. **Master Chronology**
   - Every event in strict chronological order
   - Date | Event | Source Document | Parties Involved
   - For events with approximate dates, note "[APPROXIMATE]"
   - For events with unknown dates, place in most likely
     position and note "[DATE TO CONFIRM]"

2. **Narrative Timeline** (3-5 paragraphs)
   - Tell the story chronologically in plain prose
   - Emphasize key turning points
   - Written from [MY CLIENT'S / NEUTRAL] perspective

3. **Key Date Analysis**
   - Statute of limitations relevant dates and deadlines
   - Contractual deadlines (notice periods, cure periods)
   - Gaps in the timeline where events are missing
   - Periods of inactivity that may be significant

4. **Document Gap Analysis**
   - Events referenced in the facts that lack supporting
     documents
   - Documents we should request in discovery
   - Communications between key parties during critical
     periods that are not in the record

5. **Exhibits List**
   - Suggested trial/hearing exhibit list based on
     the chronology
   - For each: relevance and what it proves

Flag any inconsistencies in the facts provided — dates that
don't align, contradictory accounts, or implausible sequences.

Pro tip: "Flag any inconsistencies" is the highest-value instruction in this template. The model will often spot timeline contradictions or logical gaps that are easy to miss when you are assembling facts from multiple sources over weeks of litigation.

Making These Templates Work Better

Always specify jurisdiction. A contract clause enforceable in Delaware may be void in California. A legal research memo without jurisdiction is useless. Make the jurisdiction variable the first thing you fill in, every time.

Chain your prompts. Legal work is sequential. Use the contract review output (Template 1) as input for the clause drafting prompt (Template 9). Use the case summarization (Template 3) to feed the brief outline (Template 7). Each output becomes context for the next prompt. Our prompt chaining guide covers the mechanics.

Build a clause library. When a clause version gets accepted in negotiation, save it. Over time, you build a library of tested, negotiated language organized by clause type, contract type, and jurisdiction — far more valuable than form books.

Add few-shot examples. If your firm has a preferred format for memos or client letters, paste one example into the prompt with "Match this format and style." The model will mirror your firm's conventions. See our few-shot prompting guide for technique details.

Use the right model. For cautious, well-qualified legal analysis, Claude's tendency to hedge is a feature, not a bug. For structured output like checklists and timelines, GPT-4o is reliable. For reviewing very long documents, Gemini's context window handles full contracts and lengthy filings. For writing more effective prompts generally, our guide on better AI prompts applies across all models.

What AI Cannot Do for Lawyers

Practice law. AI does not have a license, cannot appear in court, cannot sign filings, and cannot form attorney-client relationships. Output is a draft, not legal advice.

Replace legal research. LLMs do not have access to Westlaw or Lexis. They generate text that looks like legal analysis, but citations may be fabricated and holdings may be misstated. Every legal citation requires verification through an authoritative source.

Exercise judgment. Whether to settle, which arguments to lead with, how to manage a difficult client relationship, whether a contract term is commercially reasonable in a specific industry — these are judgment calls that require experience, context, and professional responsibility. AI provides structure. Lawyers provide judgment.

Maintain privilege. Using public AI tools with confidential client information may waive attorney-client privilege. Understand your tool's data retention and training policies before inputting any client-related information.

Stay current. Laws change. Regulations are updated. Cases are overturned. A model trained through a certain date does not know about yesterday's appellate decision or last month's regulatory update. Time-sensitive legal analysis requires current sources.

Frequently Asked Questions

Is it ethical for lawyers to use AI? Yes, with appropriate guardrails. ABA Formal Opinion 512 (2024) affirms that lawyers may use AI tools provided they maintain competence in understanding the technology, supervise its output, protect client confidentiality, and communicate AI use to clients where appropriate. Several state bars have issued similar guidance.

Will courts penalize lawyers for using AI-generated content? Courts have sanctioned lawyers for submitting AI-generated briefs with fabricated citations — most notably in Mata v. Avianca (S.D.N.Y. 2023). The issue is not AI use itself but submitting unverified work product. Verify every citation and factual claim before filing.

Which AI model is best for legal work? It depends on the task. Claude is conservative and well-suited for advisory memos and risk analysis. GPT-4o handles structured analytical output well. Gemini's extended context window is useful for long documents. None replaces legal research databases. See our model comparison for details.

Can I use AI to draft contracts? As a first draft, yes — with significant attorney review. AI-generated contract language may be internally inconsistent, may not reflect current law in the relevant jurisdiction, and may miss deal-specific terms discussed verbally. Use AI to generate a starting point, then apply your professional judgment and deal knowledge. Promplify can help structure your drafting prompts with the STOKE framework for more precise output.

How do I protect client confidentiality when using AI? Do not input identifiable client information into public AI tools. Options include: using anonymized or hypothetical facts for prompt development, deploying enterprise AI solutions with data processing agreements, using local models that do not transmit data externally, and checking your firm's AI use policy before starting. When in doubt, consult your firm's ethics counsel.

---

Want to optimize these templates for your preferred AI model? Promplify applies proven prompt engineering frameworks to your legal prompts — producing structured, model-tuned versions that generate more precise, consistent output.