8 AI Plays for a Small Law Firm

A small law firm is mostly language work: reading documents, tracking deadlines, and explaining status to clients. That is exactly what AI is good at, as long as privileged client material lives in a private/business AI workspace rather than a public model, and a licensed attorney reviews everything before it goes out or gets filed. AI drafts; the lawyer signs.

Eight concrete plays:

1. Document digester. Paste a stack of discovery, records, or exhibits; AI returns a fact timeline, the key players, and the three things that cut against your client, so you walk into review knowing where to dig.
2. Intake-to-matter summarizer. Turn a new prospect's intake form or call notes into a clean matter summary, a suggested practice-area route, and a flag list of names to run a conflict check on before you take the case.
3. Plain-English client update. Feed the current status; AI drafts a calm, jargon-free email that answers "what's happening with my case?" That kills the anxious check-in calls that eat your day.
4. Billing-narrative cleaner. Turn cryptic time entries ("tc w/ OC re: MSJ") into clear, defensible invoice narratives clients don't dispute. This alone can lift realization on every bill you send.
5. Demand & routine correspondence drafter. First drafts of demand letters, engagement letters, and standard correspondence from a short prompt: you edit instead of starting from a blank page.
6. Deposition prep partner. Give the key facts and the deponent's role; AI generates a themed question outline and the likely evasions, so prep starts at draft two.
7. Contract redline explainer. Paste two versions of a clause or a proposed redline; AI explains in plain English what changed, who it favors, and the risk to your client.
8. Deadline & obligation extractor. Pull every date and duty out of a court order, scheduling order, or contract into a checklist. Then a human confirms each against the actual rules and calendars it. Never trust the extraction alone.

A hard note on data and accuracy: everything here touches privilege. Strip or keep identifiers in a private/business workspace, and verify every legal citation and quote yourself: public models still invent case law that looks real. AI is a drafting and triage tool, not a source of legal authority or a substitute for attorney judgment.

Ready-made tools from this list:

• The Obligation Extractor: Paste Any Contract, Get Every Deadline and Auto-Renewal on One Page — builds "Deadline & obligation extractor"

Build it now

Pick the play above that fits your firm and paste this into Claude or ChatGPT to build it:

Role: You are an operations consultant for small law firms who is also a sharp prompt engineer.
Context: I run a small firm. I want to build this specific play: [paste the play name and one sentence on what it should do]. My practice areas are [e.g. family, PI, small-business/transactional]. My team is [size/roles]. I use [practice-management/DMS tools, e.g. Clio, MyCase].
Task: (1) Design the play as a repeatable tool. (2) Write the exact, reusable prompt I'll paste each time, in Role/Context/Task/Format/Constraints form, with clearly labeled [BRACKETED] fields I swap in. (3) Spell out the human-in-the-loop checkpoint and the confidentiality, privilege, and citation-verification cautions. (4) Show one worked example using realistic but 100% fictional facts and parties.
Format: A short setup section, then the ready-to-paste prompt in a code block, then the example.
Constraints: Assume no technical skills. Never instruct me to route privileged client data through a public model. Remind me to keep sensitive material in a private/business workspace and to verify every citation myself.