technicalmarketing.ai Free AI audit

Guide

AI for law firms: draft faster, and keep the client file out of the wrong tool.

Law firms can use AI for first drafts, summaries, intake and marketing pages, with a lawyer checking every output. ABA Formal Opinion 512 says lawyers must stay competent with the tool, protect client information, supervise its use and bill honestly. Start with tasks that involve no client information, then review each tool's terms before it sees a file.

8 uses worth your time.

01

First drafts of client letters

You give the assistant the point you need to make, and it drafts the letter in plain language. A lawyer rewrites the advice and removes anything that goes beyond the facts of the matter.

Tool
A general assistant such as ChatGPT, Claude or Gemini, with the facts anonymized
Limit
A letter that states the law or a deadline gets cite-checked by a lawyer before it goes out.
02

Summaries of long records

Deposition transcripts, medical records and discovery productions can be summarized with a page reference for each point. That turns hours of reading into a map of what to read closely.

Tool
A tool whose confidentiality terms you or your IT adviser have read and approved for client files
Limit
The opinion expects an appropriate degree of review of the output; the summary guides your reading and does not replace it.
03

First-pass contract review

Run a contract against your own checklist and ask the tool to flag missing clauses, unusual terms and defined terms used inconsistently. The associate then reads the flagged sections in full.

Tool
A legal-specific review tool, or a general assistant under terms you have approved
Limit
A missed risk does not announce itself, so a lawyer still reads the whole contract.
04

Research starting points

Use AI to frame a research question, list the issues to check and suggest search terms. The lawyer then researches in the database of record and reads the authorities.

Tool
A research platform built for lawyers, or a general assistant used only to organize your thinking
Limit
Assistants can give answers with no basis in fact, including authorities that do not exist, so no citation reaches a court or a client unread.
05

Witness and meeting outlines

Turn a case chronology into question topics for a witness, or a long client meeting into an agenda and a follow-up list. It gets a blank page to a working outline in minutes.

Tool
An approved tool, with identifying details removed where you can
Limit
A chronology holds client confidences, so only a tool you have cleared sees it.
06

Intake when the office is closed

An AI receptionist answers after hours, says it is an AI assistant, collects the caller's name, number and type of matter, and passes it to your intake team. It does not give legal advice or decide whether you take a case. We build these, so this is our own example.

Tool
An AI receptionist (voice agent) set to collect only what you specify
Limit
Callers start describing facts; set it to stop them and hand off, and settle with your ethics counsel how prospective-client information is stored, since the opinion extends confidentiality duties to it.
07

Practice area pages and attorney bios

AI drafts first versions of practice area pages and bios from facts you supply. A lawyer edits them for accuracy and for your state's advertising rules before they publish.

Tool
A general assistant, followed by lawyer review
Limit
Past results, specialist claims and comparisons are where advertising rules bite, and states differ, so your lawyers approve each page.
08

How assistants describe your firm

Clients ask assistants which firm handles a type of case. We check what assistants say about a firm each month and fix the pages and profiles they draw on.

Tool
Your site, listings and attorney profiles; AI search work like ours
Limit
Nobody can guarantee an assistant names your firm, and everything we write goes to your lawyers for approval.

What to keep away from AI.

  • File or send anything containing authorities an AI supplied until a person has read each one. Opinion 512 warns that uncritical reliance on output can mislead courts, clients and third parties.
  • Enter client information into a self-learning tool without reading its terms and, where the opinion calls for it, getting informed consent. It says boilerplate in an engagement letter is not enough.
  • Let a chatbot or voice agent advise callers or accept a matter. Advice and case acceptance are lawyer decisions that follow your conflicts check.
  • Bill an hourly client for time you did not spend, or pass on the cost of a general AI tool as an expense without explaining the basis first. The opinion treats general overhead as part of your fee.

A safe first month.

  1. Week 1: Name an owner and list approved usesAssign one lawyer to own AI policy. List the tasks that may use AI now, such as marketing drafts, internal checklists and anonymized exercises, and the tasks that may not, which is anything with client information.
  2. Week 2: Read the terms with IT or an expertFor each tool someone wants for client work, read the terms of use and privacy policy, or have your IT adviser do it. Record who can see inputs and whether they are used to train the model.
  3. Week 3: Draft the client-facing languageWrite the specific consent language for tools that will see client information, explaining what is entered, the risks and how the tool is used. Opinion 512 says general boilerplate does not do the job.
  4. Week 4: Pilot on one matter typePick one low-risk matter type and run the tool with a lawyer checking every output. Hold a review at month's end on errors found and time spent, and add intake or a site assistant only after that.

Rules that apply

The rules don't change because the tool is new.

Competence and verification (Model Rule 1.1)

Opinion 512 says lawyers need a reasonable understanding of a tool's capabilities and limits, and must apply an appropriate degree of independent review to its output. It names hallucinations as a risk. How much review depends on the tool and the task.

Confidentiality and informed consent (Rule 1.6)

Self-learning tools raise the risk that one client's information reaches someone else. The opinion calls for the client's informed consent before information about the representation goes into such a tool, says boilerplate in an engagement letter is not sufficient, and tells lawyers to read a tool's terms of use and privacy policy.

Supervision (Rules 5.1 and 5.3)

Managerial lawyers must create effective measures so that lawyers and nonlawyers in the firm follow the rules, which the opinion applies to AI tools, including vendors outside the office that provide them. It says managers must establish clear policies on permissible use and that supervisors must make sure lawyers and nonlawyers are trained.

Fees (Rule 1.5)

An hourly client is billed for actual time. A lawyer may bill the time spent entering information and reviewing a draft, but not time the tool saved. The cost of a tool used across many clients is overhead unless the client agreed otherwise in advance. Your state adopts its own version of each rule, and your ethics counsel decides how it applies.

This guide describes rules in general terms and is not legal or compliance advice. Check your own regulator's current text before you rely on it.

Questions people ask about AI for law firms.

Is it ethical for lawyers to use AI?

Opinion 512 does not bar it. It says lawyers must meet their existing duties when they use generative AI: competence, confidentiality, communication, candor to the tribunal, supervision and reasonable fees. Your state may have its own opinion, so check with your bar.

Who checks what the AI writes?

A lawyer does, every time something leaves the firm or reaches a court. The opinion expects an appropriate degree of independent review, and the right amount grows with the stakes of the task.

Is a legal-specific tool safer than a general assistant?

Not automatically. The label does not settle the question; the terms do. Ask any vendor where inputs are stored, who can see them, whether they train models on them and how long they are kept, and apply the same duties to either kind of tool.

What is the best AI for lawyers?

No single product is best, and we have not tested enough to rank them. Choose by task and by terms: a legal research platform for authorities, a general assistant such as ChatGPT, Claude or Gemini for drafting without client facts, and a vendor whose confidentiality terms you have read for anything that sees a file. Ask each vendor where inputs are stored and whether they train models.

Full answer

What kinds of AI tools do lawyers use?

Four kinds. General assistants handle drafting and brainstorming, legal research platforms add citation checks, document tools do review and summaries, and front-office tools such as AI receptionists and site assistants handle intake. Practice-management software is adding assistants as well. Each kind has different data terms, so evaluate them separately and record what you approved.

Full answer

Is Claude or ChatGPT better for lawyers?

Neither is built or certified for legal work, and your ethics duties are the same whichever you pick. Compare them on terms, not reputation: whether inputs train the model, who can see them, retention, and the admin controls on the plan you would actually buy. Then test both on the same non-confidential exercise and judge accuracy yourself.

Full answer

Can AI handle intake for a law firm?

It can answer calls and chats, say it is an AI assistant, collect a name, a number and the type of matter, and pass it to your team. It should not advise callers, accept matters or take long accounts of the facts. Opinion 512 extends confidentiality duties to prospective clients, so decide with your ethics counsel what it stores and who sees it.

Full answer

How should I train my lawyers and staff on AI?

Hold a short session on your written policy: approved tools, information that never goes in, how to verify citations, and who approves client-facing use. The opinion expects managers to set policies and train lawyers and staff. Use anonymized exercises, include nonlawyer staff, and repeat the session when you add a tool or your state issues guidance.

Full answer

Do I need to tell clients I use generative AI?

Sometimes. The opinion says disclosure turns on the facts. You must disclose if a client asks or requires it, you need informed consent before client information goes into a self-learning tool, and you must consult the client when AI use bears on your fee or a significant decision. Behind-the-scenes use with no client information may need none; your state's opinion controls.

Full answer

Find out what AI says about you.

A free 20-minute call. We ask the assistants your customers' questions and show you where you come up today.

No pitch unless you ask for one.