Jun 27, 2026 · 9 min read · ⚖️ Lawyers Prompt Guides

25 AI Legal Research Prompts That Save Hours (2026)

🔧 Need to generate case summaries fast? Try our Case Summary Generator: paste in a case citation and get a structured brief in seconds.

You’re staring at a research memo due tomorrow morning. The partner wants a comprehensive analysis of whether your client’s non-compete is enforceable under the new state statute, and you need to find every relevant case from the last three years. In 2024, this meant six hours in Westlaw. In 2026, with the right prompts, you can get 80% of the way there in 45 minutes.

But here’s the thing most lawyers get wrong: they type vague questions into ChatGPT and get vague answers back. The quality of your AI legal research depends entirely on the specificity of your prompts. I’ve spent the last year testing and refining prompts across CoCounsel, ChatGPT-4o, and Claude 3.5: and the difference between a mediocre prompt and a great one is the difference between useless output and genuinely useful research.

Here are 25 prompts that actually work, organized by research task.

Case Law Search Prompts

These prompts are designed to surface relevant precedent quickly. CoCounsel excels here because it searches verified legal databases, but ChatGPT and Claude can help you frame your search strategy.

Prompt 1: Broad case law survey

Identify the leading cases in [jurisdiction] addressing [legal issue] from [year range]. For each case, provide: (1) the holding, (2) the key facts that distinguish it, (3) whether it's been cited approvingly or distinguished in subsequent decisions. Organize by how favorable each case is to the position that [your client's position].

Best tool: CoCounsel ($100/user/mo): it pulls from verified Westlaw databases, so you won’t get hallucinated citations.

Prompt 2: Narrow fact-pattern matching

Find cases where courts addressed [specific factual scenario]. I need cases where the facts closely match this situation: [describe your client's facts in 3-4 sentences]. Prioritize cases from [jurisdiction], but include persuasive authority from [related jurisdictions] if on point.

Best tool: CoCounsel for verified results, then Claude to help you analyze the fact patterns more deeply.

Prompt 3: Recent developments tracker

What are the most significant [legal topic] decisions from [jurisdiction] courts in the last 12 months? Focus on cases that changed, clarified, or created splits in the existing doctrine. For each, explain what changed and why it matters for practitioners.

Best tool: CoCounsel for accuracy. ChatGPT’s training data has a cutoff, so it may miss recent decisions.

Prompt 4: Circuit split identifier

Is there currently a circuit split on [legal issue]? Identify which circuits fall on each side, the key cases from each, and whether the Supreme Court has granted cert or signaled interest in resolving the split.

Best tool: CoCounsel for the initial identification, then Claude for synthesizing the analysis into a memo-ready format.

Prompt 5: Standard of review research

What standard of review applies to [type of motion/appeal] in [jurisdiction]? Provide the controlling authority, any recent modifications to the standard, and practical implications for how I should frame my arguments.

Best tool: ChatGPT-4o ($20/mo) handles this well for established standards. Verify citations independently.

Statutory Interpretation Prompts

Statutory research requires precision. These prompts help you parse legislative intent, find interpretive case law, and identify ambiguities.

Prompt 6: Plain meaning analysis

Analyze [statute citation] using plain meaning interpretation. Identify: (1) any undefined terms that could be ambiguous, (2) how courts have interpreted those terms, (3) whether the statutory context resolves any ambiguity. Apply [jurisdiction]'s canons of construction.

Best tool: Claude ($20/mo): its longer context window handles full statutory text better than ChatGPT.

Prompt 7: Legislative history deep dive

Summarize the legislative history of [statute]. Include: committee reports, floor debates, sponsor statements, and any amendments during passage. Identify what problem the legislature was trying to solve and whether the final text matches the stated intent.

Best tool: CoCounsel for accessing actual legislative documents. ChatGPT and Claude can only work with what’s in their training data.

Prompt 8: Statutory interaction analysis

How does [Statute A] interact with [Statute B]? Are there conflicts? Which controls? Has any court addressed the interaction? If not, apply [jurisdiction]'s rules of statutory construction to predict how a court would resolve any tension.

Best tool: Claude for the analytical reasoning, but verify all citations with CoCounsel or manual research.

Prompt 9: Regulatory preemption check

Does [federal statute/regulation] preempt [state law claim/statute]? Analyze under express preemption, field preemption, and conflict preemption frameworks. Identify the leading cases on preemption in this specific area.

Best tool: CoCounsel for case identification, Claude for structuring the three-part analysis.

Prompt 10: Effective date and applicability

When did [statute/regulation] take effect? Does it apply retroactively? What's the controlling authority on its temporal scope? Are there any pending challenges to its applicability?

Best tool: ChatGPT-4o for quick answers on well-established statutes. Always verify.

Regulatory Analysis Prompts

Regulatory research is where AI saves the most time: tracking changes across multiple agencies is brutal without automation.

Prompt 11: Regulatory landscape overview

Provide a comprehensive overview of the regulatory framework governing [industry/activity] in [jurisdiction]. Include: primary statutes, implementing regulations, key agency guidance documents, and any pending rulemaking. Identify the most significant compliance obligations.

Best tool: Claude for synthesis, but supplement with Thomson Reuters Regulatory Intelligence for real-time tracking.

Prompt 12: Comment letter analysis

Summarize the key themes from public comments on [proposed rule/docket number]. What were the main objections? What changes did commenters request? Which comments did the agency address in the final rule?

Best tool: ChatGPT-4o with document upload: feed it the comment summary from regulations.gov.

Prompt 13: Multi-jurisdictional comparison

Compare how [legal issue] is regulated across [list of states/jurisdictions]. Create a table showing: (1) the governing statute, (2) key requirements, (3) penalties for non-compliance, (4) any safe harbors or exemptions. Flag jurisdictions where the law is unsettled.

Best tool: Claude for the comparative analysis and table formatting. Verify each jurisdiction independently.

Prompt 14: Agency enforcement trends

What are [agency name]'s enforcement priorities for [year]? Identify recent enforcement actions in [specific area], the theories of liability used, and the penalties imposed. What patterns suggest about future enforcement focus?

Best tool: CoCounsel for finding actual enforcement actions, ChatGPT for trend analysis.

Prompt 15: Compliance gap analysis

Based on [regulation], identify the specific compliance requirements for a company that [describe client's business]. For each requirement, indicate: what's required, the deadline, the penalty for non-compliance, and whether there's any guidance on what constitutes "good faith" compliance.

Best tool: Claude: give it the full regulatory text and your client’s business description.

Opposing Argument Anticipation Prompts

This is where AI becomes your moot court partner. These prompts help you stress-test your arguments before opposing counsel does.

Prompt 16: Steel-man the opposition

I'm arguing [your position] in [type of proceeding]. Construct the strongest possible counter-argument. What cases would opposing counsel cite? What facts would they emphasize? What policy arguments would they make? Don't hold back: I need to see the best version of their argument.

Best tool: Claude: it’s particularly good at constructing nuanced counter-arguments without hedging.

Prompt 17: Weakness identification

Here's my argument: [paste your argument]. Identify every weakness, logical gap, factual assumption, and potential distinguishing factor that opposing counsel could exploit. Rate each vulnerability from 1-5 in terms of how damaging it would be if raised effectively.

Best tool: Claude or ChatGPT-4o: both handle this well. Claude tends to be more thorough.

Prompt 18: Distinguish adverse authority

[Case name] appears to go against my position that [your argument]. Help me distinguish it. What are the factual differences? Are there procedural distinctions? Has the holding been narrowed by subsequent cases? Is there a way to read the holding narrowly?

Best tool: CoCounsel to find subsequent treatment, then Claude to craft the distinguishing argument.

Prompt 19: Reply brief anticipation

I filed [type of motion] arguing [your position]. Predict what opposing counsel will argue in their response. For each likely argument, draft a one-paragraph rebuttal I can use in my reply brief.

Best tool: ChatGPT-4o: fast and good at generating multiple response scenarios.

Prompt 20: Judge-specific research

Research [Judge Name]'s rulings on [legal issue]. What's their general approach? Have they written any notable opinions on this topic? What arguments tend to resonate with this judge? What should I avoid?

Best tool: CoCounsel for finding the judge’s actual opinions. ChatGPT may have outdated or incomplete information about specific judges.

Brief Research and Writing Support Prompts

These prompts help you structure arguments and find the right authority for each point.

Prompt 21: Argument outline generator

I'm writing a [type of brief] arguing [position]. Generate a detailed outline with: (1) the strongest argument structure, (2) what authority I need for each point, (3) where to address weaknesses preemptively, (4) the most persuasive order of arguments for [court type].

Best tool: Claude: excellent at structural thinking and argument organization.

Prompt 22: Standard language finder

What's the standard formulation courts use when [granting/denying] [type of motion] in [jurisdiction]? Give me the typical language from recent opinions so I can mirror it in my brief.

Best tool: CoCounsel for pulling actual judicial language from recent opinions.

Prompt 23: Fact section framing

Here are the facts of my case: [paste facts]. I'm arguing [position]. Help me organize the fact section to subtly emphasize facts favorable to my position while remaining accurate and complete. What should I lead with? What deserves its own paragraph?

Best tool: Claude: strong at persuasive organization without crossing ethical lines.

Prompt 24: Procedural history check

What's the correct procedural posture for [type of motion] in [jurisdiction]? What must I establish to survive [standard: e.g., 12(b)(6), summary judgment]? What's the burden of proof and who bears it?

Best tool: ChatGPT-4o for well-established procedural standards. Verify with local rules.

Prompt 25: Citation string builder

I need a citation string supporting the proposition that [legal principle]. Find 3-5 cases from [jurisdiction] that directly support this point, ordered from most authoritative to least. Include parenthetical descriptions for each.

Best tool: CoCounsel: this is exactly what it’s built for. Never trust ChatGPT or Claude for citation strings without verification.

Ethical Guardrails for AI Legal Research

A critical reminder: every AI-generated citation must be verified. This isn’t optional: it’s an ethical obligation. Courts have sanctioned lawyers for submitting AI-hallucinated cases (remember Mata v. Avianca?), and bar associations across the country now have guidance on AI use in legal research.

My workflow: use AI to identify potential cases and structure arguments, then verify every citation in Westlaw or Lexis before it goes into any filing. CoCounsel reduces (but doesn’t eliminate) hallucination risk because it searches actual databases, but even its output deserves a spot-check.

Which Tool for Which Task?

Task	Best Tool	Why
Finding real cases	CoCounsel ($100/mo)	Searches verified databases
Analyzing arguments	Claude ($20/mo)	Best reasoning, long context
Quick procedural questions	ChatGPT-4o ($20/mo)	Fast, good for established law
Full research projects	CoCounsel + Claude	Accuracy + analysis
Budget-conscious research	ChatGPT-4o + manual verification	Cheapest option that works

The bottom line: these prompts are starting points. The best legal research prompts are the ones you refine based on your practice area, jurisdiction, and the specific AI tool you’re using. Start with these, iterate, and build your own prompt library.

FAQ

Do I need ChatGPT Plus to use these prompts?

No: most prompts work with the free version of ChatGPT, Claude, or Gemini. Paid versions give you faster responses and longer outputs, but the prompts themselves work on any tier.

How do I customize these prompts for my specific situation?

Replace the bracketed placeholders with your actual details. The more specific context you provide (your industry, audience, goals), the better the output. Start with the template, then iterate based on the first response.

Can I use these prompts with Claude or Gemini instead of ChatGPT?

Yes. These prompts are model-agnostic: they work with any large language model. Claude tends to produce more nuanced writing, while Gemini integrates well with Google Workspace.

How often should I update my prompts?

Revisit your prompt library every 2-3 months. AI models improve regularly, and what required detailed instructions six months ago might now work with simpler prompts. Also update when your business context changes.

Is it ethical to use AI-generated content in my work?

Yes, as long as you review, edit, and take responsibility for the final output. AI is a drafting tool: the expertise, judgment, and quality control still come from you. Disclose AI use where required by your industry or employer.