Gemini Prompt: Draft an Internal IRAC Legal Memorandum
Draft a partner-ready internal legal memo using IRAC — facts, issue, rule, application, and a confidence-rated conclusion with recommended next steps.
The Prompt
Variables (3)
| Variable | Description | Example |
|---|---|---|
| {{senior_partner_name}} | Senior Partner Name | Jane Whitfield |
| {{facts}} | Relevant Facts | Client signed a commercial supply contract in March 2024 with a liquidated damages clause set at 25% of contract value... |
| {{legal_question}} | Legal Question | Whether the liquidated damages clause is enforceable under Delaware law given that actual damages were quantifiable. |
Expected Output
A neutral IRAC-structured internal memo with restated issue, applicable rule, fact-driven application addressing strengths and weaknesses, and a confidence-rated conclusion with recommended next steps.
Usage Notes
Model-agnostic but particularly clean on Gemini 3.1 Pro for structure adherence. The explicit "do not cite cases you cannot verify" instruction reduces — but does not eliminate — citation hallucinations; verify every authority before circulating.
Originally featured in: Gemini 3.1 Pro for Lawyers: Best Prompts, Use Cases & Honest Review (2026)
Related Prompts
AI Prompt: First-Draft Generator for Any Legal Document
One reusable template for first drafts of any legal document — feed in document type, practice area, brief facts, client type, and jurisdiction.
Demand Letter Template Prompt with Customizable Variables for Any Practice
Reusable demand letter scaffold with placeholder variables — swap in practice area, jurisdiction, client details, dispute facts, and damages to draft any demand.
AI Prompt: Organize Cases Into Research Memo Outline
Turn your verified case list into a memo outline grouped by sub-issue, ordered persuasively — the AI works only with cases you provide.
AI Prompt: Synthesize Cases Into Structured Research Memo
Organize verified cases into a structured analysis with sub-issues, majority/minority positions, and circuit splits — closed-set, no added citations.