AI for law firms pays off first at intake, not in the brief: answering missed calls and after-hours web leads in minutes, summarizing intake and preparing the conflict-check list, then summarizing documents and drafting routine letters that a lawyer verifies. ABA Formal Opinion 512 says you may use generative AI if you understand the tool, protect client confidences, supervise its use and bill fairly. Start with intake, keep lawyers on judgment, and measure signed matters by source.
Key takeaways
- The biggest payoff is intake: missed-call text-back, fast replies to after-hours web leads, and clean intake summaries.
- ABA Formal Opinion 512 (July 2024) applies existing duties to AI: competence, confidentiality, communication, supervision, candor and reasonable fees.
- Intake AI collects facts and books consults. It never assesses a case, quotes unapproved fees or pretends to be a person.
- Every AI-drafted citation and fact gets checked against the source before it reaches a court or a client.
- Measure signed matters by source, not messages sent. This guide is not legal advice; your state bar's rules govern.
How are law firms using AI in 2026?
Most small and mid-sized firms use AI in two places: the front door (intake, follow-up, scheduling) and the paperwork (summaries, first drafts, document review), with a lawyer checking anything that touches advice, a filing or a fee. The firms getting real value treat AI as a fast junior assistant, never as counsel.
The vendor pitch usually starts with legal research. For a small practice, the bigger money is earlier in the funnel. A personal injury, family or criminal defense firm lives or dies on whether it answers the 9 PM call and gets the consult on the calendar. AI is very good at that job and carries far less ethical risk than drafting a brief.
AI for law firms: quick facts
- ABA Formal Opinion 512 (July 29, 2024) is the ABA's first ethics opinion on generative AI. It applies existing Model Rules: competence, confidentiality, communication, supervision, candor and reasonable fees.
- Florida Bar Ethics Opinion 24-1 (January 2024) says lawyers may use generative AI if they protect confidentiality, stay accurate and competent, bill properly and follow advertising rules.
- New Jersey: the state Supreme Court issued preliminary AI guidelines for lawyers in January 2024, built on the existing Rules of Professional Conduct.
- Generative AI: tools such as ChatGPT, Claude, Gemini and Microsoft Copilot that write new text from a prompt. They can state false things confidently.
- Not legal advice: this guide explains operations and marketing. Your state bar's rules and opinions govern your practice.
Where does a law firm lose money that AI can fix?
Most firms lose more cases at intake than in court. The leaks are predictable: calls that ring out, web leads that wait until morning, consults that never get booked, and prospects who go quiet after the first conversation. AI for law firms pays off fastest by closing those gaps.
The five intake leaks we see most
- Missed calls. The receptionist is on another line, at lunch or gone for the day. Many callers try the next firm on the list instead of leaving a message.
- After-hours web leads. Accidents, arrests and family crises do not happen 9 to 5. A form submitted at 11 PM that gets a reply at 10 AM is often already lost.
- Slow consult booking. A lead who has to play phone tag to schedule usually books with whoever offers a time first.
- No follow-up. A prospect says "let me think about it" and nobody calls back.
- Unknown sources. The firm spends on ads and directories but cannot say which ones produced signed cases.
Before you add AI to any of these, measure them. How many calls were missed last month? How fast does a web lead get a reply? Which source produced each signed matter? If your website and search presence are the source of most leads, pair this with our law firm SEO guide, because more traffic into a leaky intake process only makes the leak bigger.
The 12 best uses of AI for law firms, ranked by payoff and effort
The best uses sit at intake and in routine paperwork, where the rules are clear and a lawyer can check the output quickly. Legal judgment, advice to a specific person and anything filed with a court stay with the lawyer.
| Use case | What the AI does | What a person still does | Payoff | Effort |
|---|---|---|---|---|
| 1. Missed-call text-back | Texts the caller within seconds with a link to book or a promise of a callback | Calls back, qualifies, decides | Very high | Low |
| 2. After-hours web lead reply | Replies to forms in minutes, collects basic facts, offers consult times | Reviews the facts, accepts or declines the matter | Very high | Low to medium |
| 3. Intake summary | Turns the call or form into a clean summary in the practice system | Reads it before the consult | High | Low |
| 4. Conflict-check prep | Pulls every party name from the intake into a list for checking | Runs the conflict check and clears it | High | Low |
| 5. Consult follow-up | Sends polite reminders and follow-ups that stop on reply | Writes and approves the messages | High | Low |
| 6. Lead source tracing | Tags every call and form with the ad, page or listing it came from | Decides where marketing money goes | High | Medium |
| 7. Document summaries | Summarizes records, depositions or contracts with page cites | Verifies against the source before relying on it | High | Medium |
| 8. First drafts | Drafts routine letters, client updates and simple documents from templates | Edits, verifies every fact and citation, signs | Medium to high | Medium |
| 9. Client status updates | Drafts plain-English updates from matter notes | Approves and sends | Medium | Low |
| 10. Staff knowledge assistant | Answers staff questions from firm procedures and templates | Keeps procedures current | Medium | Medium |
| 11. Marketing content | Researches questions clients ask and drafts articles | A lawyer edits for accuracy and bar advertising rules | Medium | Medium |
| 12. Intake training | Simulates calls so new staff practice before real ones | Coaches and scores | Medium | Medium |
The ranking follows a simple logic: how often the job happens, how much a slow or missed one costs, and how easily a person can check the output. Missed calls happen daily and cost signed cases. A draft brief happens less often, and checking it properly takes nearly as long as writing it.
How should AI for law firms handle intake?
AI should make sure every prospect gets a fast, polite, accurate response and that the facts reach a person in usable form. It should not give legal advice, quote fees it has not been told, or promise outcomes.
Missed calls and after-hours leads
The simplest win is a missed-call text-back: when nobody answers, the caller gets a text within seconds saying the firm got the call, with a link to book a consult or a time to expect a callback. For web forms, an AI reply goes out in minutes, asks a few qualifying questions (type of matter, county, dates) and offers consult times. If you want the phone itself answered, read our guides to the AI voice agent and the AI receptionist.
Conflict-check prep
Intake forms and call notes mention people: the opposing party, a spouse, an employer, an insurer. AI can pull every name into a clean list attached to the new matter, so the person running the conflict check starts with a complete list instead of rereading the notes. The check itself, and the decision, stay with the firm.
What the intake AI must never do
- Tell someone whether they have a case, or what it is worth.
- Quote a fee or a timeline the firm has not approved word for word.
- Pretend to be a person. Say plainly that it is an automated assistant.
- Collect more confidential detail than intake needs before a lawyer is involved.
On that last point, the Florida Bar Ethics Opinion 24-1 warns that website chatbots can create a lawyer-client relationship without the lawyer knowing, should tell prospects they are talking to an AI program, and should carry clear disclaimers. Lawyers stay responsible if a chatbot gives misleading information.
Can AI review documents and draft for a law firm?
Yes, as a first pass that a lawyer verifies. After intake, document work is the second big use of AI for law firms. AI is good at summarizing long records, pulling dates into a chronology, comparing two versions of a contract and producing a first draft from a template. It is not reliable enough to cite, file or advise without a lawyer checking every fact and citation against the source.
Where document AI helps most
- Medical and accident records: a chronology of treatment with page references, for a personal injury file.
- Discovery and depositions: summaries of who said what, with page and line cites to check.
- Contracts and leases: a list of key terms, deadlines and unusual clauses compared with your standard.
- Intake documents: police reports, notices and court papers read into the right fields.
Our guide to AI document processing covers the extraction side in depth.
Drafting with verification
Courts have sanctioned lawyers for filing briefs with citations that a chatbot invented. The best-known example is Mata v. Avianca, where a federal judge in New York sanctioned two lawyers and their firm in June 2023 over fake cases generated by ChatGPT. The fix is a firm rule, not better prompts: every citation is pulled and read from a trusted research service before it goes into a filing, and every factual statement is checked against the record. Treat AI drafts like work from a new hire you have not yet learned to trust.
Routine correspondence is lower risk: engagement letters from your template, appointment confirmations, document requests and client status updates. Those are good places to start drafting with AI, with a person reading each one before it goes out.
What does ABA Formal Opinion 512 say about AI?
ABA Formal Opinion 512, issued July 29, 2024, says lawyers may use generative AI but must meet the same duties as always: understand the tool, protect client information, communicate with clients, supervise others, stay candid with courts and charge reasonable fees. It is not binding by itself; your state's rules and opinions govern. This is a summary, not legal advice.
| Duty (Model Rule) | What Opinion 512 says, in short | What it means day to day |
|---|---|---|
| Competence (1.1) | Lawyers need a reasonable understanding of the capabilities and limitations of the specific AI they use, and must check its output | Know what each tool does; verify before relying on it |
| Confidentiality (1.6) | Before putting client information into a self-learning tool, informed client consent may be required; general boilerplate in an engagement letter is not enough | Read the tool's terms and data settings; get real consent when required |
| Communication (1.4) | Tell clients about AI use when it affects the fee or matters to their decisions | Explain it in the engagement conversation or letter |
| Supervision (5.1, 5.3) | Managing lawyers should set clear policies and train lawyers and staff | Write an AI policy; train everyone who uses the tools |
| Candor and claims (3.1, 3.3) | Lawyers must check AI output before relying on it in filings | Pull and read every citation |
| Fees (1.5) | Bill hourly for actual time spent; do not charge clients for learning a general tool | Efficiency should show up in the bill, not hide in it |
States are adding their own guidance. Alongside the Florida opinion above, the New Jersey Supreme Court issued preliminary guidelines on lawyers' use of AI in January 2024, framed around the existing Rules of Professional Conduct. Check your own bar's opinions before choosing tools, and ask your malpractice carrier what it expects.
Where AI goes wrong in a law firm (and the guardrail for each)
The risks of AI for law firms are rarely dramatic. A confident wrong answer, a leak through a consumer tool, a chatbot that sounds like it is giving advice, or a sequence that texts a grieving family at the wrong moment. Each has a guardrail you can put in place before launch.
| Failure | What it looks like | Guardrail |
|---|---|---|
| Hallucinated citations or facts | A draft cites a case that does not exist | Every citation pulled and read before filing; a named reviewer |
| Confidentiality leak | A paralegal pastes a client file into a free personal chatbot | Approved tools only, business accounts with data controls, a written "never paste" list |
| Accidental advice | The intake bot tells a prospect they have a strong case | Scripted scope: facts in, consult out; no opinions on merits or value |
| Wrong tone | An upbeat auto-text after a death or an arrest | Sensitive matter types skip automation and go straight to a person |
| Over-automation | A follow-up sequence keeps texting after the client hired someone else | Sequences stop on reply, booking or opt-out, with quiet hours |
| Bad data in | Conflict list misses a party because intake notes were incomplete | Required intake fields; a person runs and clears every conflict check |
| Advertising problems | AI-written pages make claims the bar would not allow | A lawyer reviews all marketing against your state's advertising rules |
On marketing claims about AI itself, the FTC said in September 2024 that there is no AI exemption from the laws on the books. If your firm advertises its use of AI, the claims need to be true and provable.
What has Unified built that a law firm can use?
We build the intake and follow-up layer that sits between a firm's website, phone and practice systems. We have not published a law firm case study, so here is what we have built for other businesses, and how the same pieces fit a practice.
Every lead traced to its source
The problem: businesses spending on ads, directories and SEO could not say which source produced paying clients. What we built: UnifiedIQ, our own platform. Calls, taps on the phone number and forms from 8 popular form plugins, its own form builder and GoHighLevel land in one inbox with their source attached. Spam and job seekers are sorted out, and the business gets an email within minutes. Every Monday a test lead checks each form still works. The owner's email shows the contact and their answers; our team gets a separate copy with the full source.
For a firm: you can see which ads and pages produce consults, and when the CRM is connected, booked and won matters can be sent back to Google Ads so bidding chases signed clients, not clicks. A person approves every change to the ad account.
Form to CRM in seconds, with follow-up that knows when to stop
We built a hand-off from a website form into GoHighLevel: the lead lands as a tagged contact with the answers in the right fields within seconds (about 3 seconds in an October 2026 test), so the firm's own CRM automations fire. We also built missed-call text-back and follow-up sequences that stop the moment the lead replies, books or opts out, and that respect quiet hours. They are rolling out client by client.
A live call assistant and a training simulator
For a South Florida wellness practice's front desk, we built a pilot live call assistant. During a call it suggests responses to common objections from a playbook the practice approved, then drafts the call notes for the CRM. A person runs the call; the AI only suggests. We also built a training simulator with 8 realistic caller scenarios, a practice mode that grades each response best, okay or poor, and a scorecard. The same pattern fits a firm's intake team: an approved script, practice before live calls, and notes that write themselves.
AI chat in a regulated industry
For HonestyInsure.com, a health insurance site, we built 144 pages and guides engineered to be cited by AI search, 31 state pages and a 24/7 AI agent chat in the hero. Regulated topics need tight scope, and that discipline carries straight over to legal intake. For a custom pipeline and intake build, see HuntReserve: 9 drag-and-drop pipeline stages, digital waivers and a form builder that records the page each inquiry came from.
What we can build for a firm: intake assistants scoped to facts and scheduling, document extraction into your practice system, a staff assistant that answers from firm procedures, and reports with every number sourced. Our AI software and intelligent workflows teams scope each one with you.
Should a law firm buy legal AI tools or build custom ones?
Buy the legal research and drafting tools built for lawyers, and use the AI features inside your practice management system. Build custom where the job is your intake, your follow-up and your reporting, because those depend on how your firm takes in cases and no single product covers them end to end.
| Need | Usually buy | Usually build or connect |
|---|---|---|
| Legal research and citations | A legal research service with AI features | Rarely |
| Practice and matter management | Your practice system (Clio and similar) and its AI features | Connections into it from intake |
| General writing help | A business account on ChatGPT, Claude, Gemini or Microsoft Copilot | Rarely |
| Intake, follow-up, lead tracing | Partly, through a CRM such as GoHighLevel or HubSpot | Yes: the rules, routing and source tracking |
| Firm-specific knowledge assistant | Sometimes | Often, so it answers only from your procedures |
Whatever you buy, read its data terms against Opinion 512's confidentiality section before anyone uploads a client file. For the wider picture, see our guides on custom AI solutions, AI CRM and AI knowledge bases.
What we'd do first: a 30/60/90-day AI plan for a law firm
Fix intake first, then follow-up and measurement, then the paperwork. Each step has a named owner at the firm and one number that shows whether it worked. This is the order we follow when we scope AI for law firms.
Days 1-30: stop losing callers
- Write a one-page firm AI policy: approved tools, a never-paste list, who verifies what. This is your Opinion 512 supervision step.
- Measure missed calls, web lead reply time and consults booked per week.
- Turn on lead source tracing for every call and form.
- Launch missed-call text-back and an after-hours form reply, scoped to facts and scheduling, with an attorney approving the scripts.
Days 31-60: book and follow up
- Send intake straight into your CRM or practice system with a clean summary and a list of party names for the conflict check.
- Add consult reminders and a short follow-up sequence that stops on reply.
- Route sensitive matter types to a person with no automation.
- Start a weekly report: calls, leads by source, consults booked, matters signed.
Days 61-90: the paperwork
- Pilot document summaries on closed files first, comparing AI output with what your team wrote.
- Template routine letters and client status updates for AI drafting with review.
- Review the 90 days with the managing partner: what to keep, change or switch off.
If you want help sequencing this, our AI readiness assessment covers data, people and process, and the broader AI for small business guide maps the same jobs across other industries.
How do you measure whether AI is working for your firm?
Judge AI for law firms on intake outcomes, not AI activity. The number of messages sent means nothing; the number of consults booked and matters signed from each source means everything.
| Metric | Why it matters | Where it comes from |
|---|---|---|
| Missed-call rate | Each unanswered call is a prospect who may call the next firm | Phone system or call tracking |
| Web lead reply time | Speed decides who books the consult | CRM timestamps |
| Consults booked per lead | Shows whether intake converts | CRM or practice system |
| Consult show rate | Reminders should move it | Calendar |
| Signed matters by source | Tells you where to spend | Lead tracing joined to the CRM |
| Staff hours on retyping | Shows the paperwork payoff | A simple time log before and after |
Most firms can pull the first three this week. The last two take a connected system, which is where our analytics and reporting work comes in. When you are ready to talk it through, contact our team in Florham Park, New Jersey or South Florida.
AI for law firms: questions, answered
Is it ethical for lawyers to use AI?
Yes, within the existing rules. ABA Formal Opinion 512 says lawyers may use generative AI if they understand the tool's capabilities and limits, protect client information, communicate with clients, supervise its use, verify output and charge reasonable fees. State bars, including Florida's Opinion 24-1, say much the same. This is not legal advice; check your own state's guidance.
What is the best AI for a small law firm to start with?
Start at intake: a missed-call text-back and a fast, scripted reply to after-hours web leads that collects basic facts and offers consult times. It happens daily, it costs cases when it fails, and a person can check the output quickly. Drafting and research tools come after.
Can I put client information into ChatGPT?
Not into a free personal account. Opinion 512 says informed client consent may be required before putting client information into self-learning tools, and that boilerplate in an engagement letter is not enough. Use business accounts with data controls, read the terms, and keep a written never-paste list.
Can AI do conflict checks for a law firm?
AI can prepare the conflict check by pulling every party name from intake notes and forms into a clean list. A person at the firm should still run the check in your practice system and clear it. The decision stays with the firm.
Can a law firm bill for time saved by AI?
Opinion 512 says hourly billing should reflect time actually spent, and that firms generally should not charge clients for learning a general AI tool. Efficiency gains should show up in the bill. Flat fees raise their own reasonableness questions, so check your state's guidance.
Do law firm chatbots need to say they are AI?
Florida Bar Ethics Opinion 24-1 says lawyers should tell prospective clients they are talking to an AI program, not a lawyer or employee, and should use clear disclaimers. It also warns that a chatbot can create a lawyer-client relationship without the lawyer knowing. It is good practice everywhere.
Will AI replace paralegals or receptionists?
In our experience it changes the job more than it removes it. AI takes over retyping, first summaries and after-hours replies, so staff spend more time on calls, client care and checking AI output, which Opinion 512 expects someone to supervise.
How do I know if AI intake is working?
Track missed-call rate, web lead reply time, consults booked per lead, consult show rate and signed matters by source. If those move, it is working. If only message counts go up, it is not.
Fix your intake first
We will map where your firm's calls and web leads come from and where they slip away, then scope the intake and follow-up build that fits your rules.
Get a free proposal