Blog

Artificial Intelligence (AI) is transforming legal practice, and mastering prompt engineering is the key to unlocking its full potential. Whether you’re using AI for legal research, contract drafting, or case analysis, how you ask your question determines the quality of the answer. A well-crafted prompt can save time and improve accuracy, while a vague or poorly structured one can lead to misinformation. 

In this guide, we’ll walk through the best practices of legal prompt engineering and show you how to craft effective prompts with real-world examples. 

1. Legal Research: Get Precise Answers with Targeted Prompts 

Bad Prompt: 

“What is the law on employee privacy?” 
Too broad, lacks jurisdiction, no context. 

Good Prompt: 

“Under U.S. federal law, what privacy rights do employees have regarding their work emails? Cite relevant statutes or case law.” 

Includes jurisdiction, specific issue, and asks for sources. 

Best Practice: When researching legal issues, always include: 

  • Jurisdiction (U.S. federal, California, UK, etc.) 
  • Timeframe (e.g., “as of 2024”) 
  • Specific issue (e.g., “work emails” rather than general “privacy”) 
  • Citation request (“cite statutes or case law”) 

2. Contract Drafting: Provide AI with a Blueprint 

Bad Prompt: 

“Draft a commercial lease agreement.” 

Too generic; AI may produce an incomplete or unenforceable contract. 

Good Prompt: 

“Draft a commercial lease agreement between ABC Properties LLC (landlord) and TechStartup Inc. (tenant) for a 5-year term. Include clauses on: 

  1. Rent ($5,000/month) 
  1. Security deposit ($10,000) 
  1. Permitted uses (general office work) 
  1. Maintenance responsibilities (tenant responsible for interior upkeep, landlord for structural repairs) 
  1. Governing law: New York.” 

Specifies parties, key terms, and governing law. 

Best Practice: 

  • Clearly state who the parties are and the type of agreement. 
  • List key clauses you want included (rent, liability, governing law, etc.). 
  • If applicable, specify the tone (formal, simple language, etc.). 

3. Drafting a Letter: Guide AI with Purpose and Details 

Bad Prompt: 

“Write a letter to opposing counsel suggesting favorable edits to the contract.” 
Too vague; AI won’t know the context or issues. 

Good Prompt: 

“Draft a formal letter to opposing counsel, Jane Doe, regarding our client’s objections to the indemnity clause in the contract. 

  1. Politely but firmly explain that our client requires mutual indemnification. 
  1. Reference Section 5 of the draft agreement. 
  1. Propose a revision to include reciprocal obligations. 
  1. Request a response by next Friday.” 

Provides recipient, issue, relevant section, and deadline. 

Best Practice: 

  • Identify who the letter is to and what the dispute is about. 
  • Specify the tone (persuasive, neutral, firm, etc.). 
  • Outline what needs to be included (e.g., reference contract section, set a deadline). 

4. Case Analysis: Teach AI to Think Like a Lawyer 

Bad Prompt: 

“Apply Smith v. Jones to my client’s case.” 

AI may generate a response that simply confirms the case applies without critically analyzing key differences or legal nuances. 

Good Prompt: 

“My client is a commercial landlord who is suing a former tenant for early termination of a lease without proper notice. The lease required six months’ notice before termination, but the tenant vacated in two months, citing economic hardship. The case Smith v. Jones (2022) held that commercial tenants must strictly adhere to notice provisions unless an unforeseeable event makes performance impossible. 

  • Analyze whether Smith v. Jones is relevant to my client’s case. 
  • Identify similarities and differences between Smith and my client’s case. 
  • Highlight any factors that might distinguish my client’s case from Smith. 
  • If Smith does not apply, suggest other relevant precedents.” 

Ensures a deeper analysis by prompting the AI to evaluate both applicability and distinctions. 

Best Practice: 

  • Describe your client’s facts clearly. 
  • Avoid asking AI to assume a case applies; instead, request an analysis of similarities and differences. 
  • Encourage the AI to suggest alternative cases if necessary. 

5. Litigation Preparation: Testing Arguments Before Trial 

Bad Prompt: 

“Does my argument sufficiently establish negligence under California law?” 
 
 

This assumes the AI understands your full argument and the specific legal standard. The AI might respond with a generic or overly confident answer, missing key nuances. 

Good Prompt: 

“Evaluate the strength of this trial argument for a negligence case: ‘The defendant’s failure to install proper safety signage directly led to the plaintiff’s injury, violating their duty of care under premises liability law in California. The plaintiff was a lawful visitor, and the absence of warning signs created an unreasonable risk.’ 

  • Identify weaknesses in this argument. 
  • Suggest stronger phrasing or additional supporting case law.” 

Provides legal argument, jurisdiction, specific issue, and asks for critique and improvement. 

Best Practice: 

  • Clearly state the legal argument you want reviewed. 
  • Include jurisdiction and relevant legal standards. 
  • Ask for specific feedback (e.g., weaknesses, counterarguments, case law support). 

6. Iteration: Building on Prompts for Maximum Effectiveness 

AI works best when prompts are refined and layered over time. Instead of expecting a perfect response in one query, build on the AI’s output by refining the request. 

Example Iteration: 

First Prompt: “Summarize key arguments for and against enforcing non-compete agreements in Texas.” 

AI Response: Provides a general summary of Texas law on non-competes. 

Follow-up Prompt: “Expand on how Texas courts determine whether a non-compete is reasonable in scope and duration. Provide recent case examples.” 

By iterating, you guide the AI toward deeper and more precise insights. Each new prompt builds on the previous response to refine the accuracy and relevance of the AI’s output. 

Key Takeaways for Lawyers Using AI 

  • Be specific – vague prompts lead to vague answers. The more details, the better. 
  • Set context – AI is not a lawyer; provide it with relevant background. 
  • Request a format – want bullet points? A case comparison? Say so. 
  • Iterate and refine – if the first response isn’t perfect, tweak your prompt. 

Final Thought: AI is a powerful tool, but only as smart as the instructions it receives. By applying these prompt engineering best practices, lawyers can work faster and more efficiently—without compromising accuracy. Next time you use AI for legal work, craft your prompt like you would instruct a junior associate: with precision, clarity, and purpose. 

Stay ahead of the curve with Law Tech AI insights. Subscribe for more updates on AI in legal practice.