AI receptionist for law firms: what it should handle (and what it must not)

An AI receptionist for law firms answers every call, day or night, works out whether the caller is a new prospect or an existing client, collects the intake and conflict-check details, books qualified consultations and transfers urgent calls to an attorney with a briefing. What it must never do is give legal advice, predict outcomes, put a value on a case or suggest that an attorney-client relationship exists. Get that boundary right and it becomes the most reliable member of your intake team.

Below: what to hand to an AI answering service for lawyers, what to keep with people, a sample script, a setup checklist and questions to ask vendors.

What should an AI receptionist for law firms handle?

Think of the receptionist as the first five minutes of every call: nobody reaches voicemail, nobody repeats themselves, and attorney time goes to matters the firm takes.

Answer every call, 24/7

Accident victims, people who have just been arrested and family members in a crisis call at night and on weekends. A voice agent picks up on the first ring at 11 p.m. on a Sunday the same way it does at 10 a.m. on a Tuesday, and it can handle several calls at once, so a busy Monday morning does not produce a queue.

Tell new callers from existing clients

The first fork in the call is simple: “Are you a current client of the firm, or is this about a new matter?” Existing clients get routed, a message taken or a callback booked with their attorney or paralegal. New callers go into intake.

Capture conflict-check information

Before anyone discusses the substance of a matter, the firm needs names: the caller, the other parties involved, and any businesses or insurers on the other side. The receptionist should collect these as structured fields and pass them to whoever runs conflicts. It should not run the conflict decision itself or tell the caller whether the firm can help until that check is cleared under your process.

Qualify against your criteria

Every practice has gatekeeping questions: jurisdiction, practice area, deadlines or dates of the incident, whether the person already has a lawyer, and the basic facts. The AI asks them conversationally, records the answers and applies your rules, so out-of-area or already-represented callers are handled politely without reaching an attorney.

Book consultations

Once a caller fits, the receptionist checks live attorney availability and books the consult on the call, then sends a confirmation by text or email. Automated reminders follow.

Route, take messages and warm-transfer

For an existing client with a question, the right answer is usually a detailed message to the right person, logged against the matter. For an urgent new matter, the right answer is a warm transfer: the agent briefs the attorney first (who is calling, what happened, what has been confirmed) and then connects the call, so the attorney starts with context instead of “How can I help you?”

Speak the caller’s language

If a meaningful share of your callers prefer Spanish or another language, a bilingual voice agent can run the same intake in that language and hand a translated summary to your team. Test it with native speakers before launch, not just with the vendor’s demo.

What an AI receptionist must not do

A law firm’s receptionist, human or AI, works under attorney supervision. Most of the risk with an AI answering service sits in a handful of sentences it should never say. Write these as hard rules in the script and test them on purpose.

• No legal advice. It does not tell a caller whether they have a case, what to do next legally, or how a law applies to their facts. A safe line: “That’s exactly the kind of question the attorney will go through with you at the consultation.”
• No promises about outcomes. No “you’ll definitely win,” “we always get results,” or anything close.
• No case values. Callers ask “What’s my case worth?” constantly. The answer is always a polite deflection to the attorney.
• No implied representation. The agent should not say or imply “we’ll take your case” or “you’re our client now.” Engagement is an attorney decision, made after conflicts and review.
• No collecting more than intake needs. Get the facts needed to qualify and check conflicts. Detailed narratives of the matter belong in the consultation.

The rules on advertising, solicitation, confidentiality and supervising non-lawyer assistance vary by state. Check your state bar’s rules and any ethics opinions on AI before you go live. The ABA’s Formal Opinion 512 on generative AI tools (July 2024) is a useful starting point for the questions to ask, but it is not a substitute for your own jurisdiction’s guidance.

AI receptionist vs. human answering service vs. in-house receptionist

None of the three options is right for every firm. Many firms end up combining them: an in-house person during business hours, with AI covering overflow and after-hours calls. Here is how they compare on the things that matter for intake.

	AI receptionist	Human answering service	In-house receptionist
Availability	24/7, no hold queue, handles several calls at once	Often 24/7, but capacity is shared with other clients	Business hours; one call at a time
Consistency	Asks the same questions the same way every time	Varies by operator and shift	High once trained, drops when covering or out
Knowledge of your firm	Only what you script and connect it to	Usually a generic script and limited firm context	Deepest; knows clients, staff and history
Judgment on unusual calls	Limited; should escalate anything off-script	Moderate; follows escalation instructions	Strongest; can read the situation
Booking and systems	Can book live into your calendar and write to your CRM or case management	Often takes messages; booking depends on the service	Full access to your systems
Empathy with distressed callers	Can be calm and patient if designed well; some callers still want a human	Human warmth, but rushed when busy	Best, especially with existing clients
Main limit	Needs careful scripting, testing and monitoring	Thin context; details lost in handoff	Coverage gaps, turnover, training time

Pricing changes often and is usually per minute or per call, so get written quotes based on your real call volume.

A sample call script for a law firm AI receptionist

Below is a simplified script for a personal injury practice. Adapt the questions to your practice area and your criteria; the structure carries over to family law, criminal defense, immigration or estate planning.

1. Greeting and disclosure. “Thank you for calling [Firm Name]. I’m the firm’s virtual assistant. This call may be recorded. Are you a current client, or are you calling about a new matter?”
2. Existing client branch. “Thanks. Can I get your full name and the best number to reach you? Who are you usually working with here?” Then take a message or book a callback, and log it against the matter.
3. New matter: basics. “I’m sorry you’re dealing with this. I’ll ask a few quick questions so the right attorney can help. What’s your full name, and what’s the best number for you?”
4. Conflict-check details. “Who else was involved? Do you know the names of any other drivers, companies or insurance carriers?”
5. Qualification. “Which state did this happen in, and roughly when? Were you hurt? Do you already have an attorney working on this?”
6. Deflection, when needed. If asked about case value or chances: “I can’t speak to that, but the attorney will go through it with you at the consultation.”
7. Outcome A: qualified. “Thank you. I can book a free consultation with an attorney. I have Thursday at 10 a.m. or 2 p.m. Which works better?” Confirm by text.
8. Outcome B: urgent. “I’m going to connect you with the attorney now. Please hold for a moment while I let them know what’s happened.” Brief the attorney, then connect.
9. Outcome C: not a fit. “Thank you for calling. This isn’t a matter our firm handles, but you may want to contact your state bar’s lawyer referral service.” Log the reason.

Note what is missing: no opinions on fault, no value estimate, no “we’ll take your case.”

How to set up an AI receptionist for your law firm: checklist

Most of the effort in a good deployment goes into decisions, not technology. Work through this list before any vendor starts building.

• Call types. List every reason people call: new matters by practice area, existing clients, opposing counsel, courts, vendors, job applicants. Decide what happens to each.
• Intake and qualification questions. Write the exact questions and the rules that make a caller qualified, not a fit, or urgent.
• Conflict fields. Agree which names and parties are captured, in what format, and who reviews them.
• Escalation rules. Define what counts as urgent, who gets the transfer at each hour, and what happens if that person doesn’t answer (a backup number, an immediate text alert, a guaranteed callback time).
• Calendar. Connect attorney calendars, set consultation lengths, buffers and which attorneys take which matter types.
• CRM or case management. Every call should create or update a record with a summary, transcript and outcome, with duplicates merged by phone number.
• Missed and dropped calls. Set an automatic text back to the caller and an alert to staff.
• Recording and AI disclosure. Call recording consent laws differ by state, and some require every party’s consent. Check your state’s consent law and your bar’s rules, and set the greeting accordingly.
• Data handling. Know where recordings and transcripts are stored, who can access them, how long they are kept and whether the vendor uses them to train models.
• Testing. Run test calls with real scenarios: interruptions, accents, angry callers, people in shock, people asking for legal advice. Fix the script before launch, not after.
• Review cadence. Someone at the firm reads a sample of transcripts every week for the first months.

How to evaluate an AI answering service for lawyers

Demos always sound good. These questions show you what happens on a bad call at 2 a.m.

1. Can I hear real calls, not just a demo? Ask for anonymized recordings from a live legal deployment, including a difficult one.
2. How do you stop it from giving legal advice? Ask them to show you the guardrails and then try to break them yourself on a test call.
3. What happens when it doesn’t know the answer? The right answer involves escalation or a message, not improvisation.
4. How does the warm transfer work? Does the attorney hear a briefing before the caller is connected? What happens if nobody picks up?
5. Which systems does it write to? Your calendar, CRM and case management, or only an email summary?
6. Where is our data stored, who can access it, and is it used for training? Get the answer in writing.
7. Who owns the scripts, prompts and call data if we leave? You should be able to take them with you.
8. Who monitors quality after launch? A named person reviewing calls, or a dashboard you have to check yourself?
9. What does a month actually cost at our call volume? Ask for a cost-per-call estimate based on your numbers.
10. How long until it’s live, and what do you need from us? A vendor that skips the scripting and testing questions is skipping the hard part.

Example: how a Florida personal injury firm answers every lead

Darfoor Law Firm, a personal injury practice in Fort Lauderdale, had steady demand from phone and social media but intake that couldn’t keep up. Calls and DMs after hours waited until the next business day, staff re-keyed details by hand and created duplicate records, and out-of-state or already-represented callers still reached an attorney.

We built a Voice AI receptionist alongside chat agents on Instagram, Facebook and TikTok, all running the same eight-step qualification. Five gates (Florida residency, confirmed physical injury, unrepresented status, a valid phone number and a complete name) are checked before anything reaches the team. Qualified callers get a pre-filled booking link, and urgent cases are warm-transferred to Mr. Darfoor after the agent briefs him. A missed call triggers two text messages within seconds: one to the prospect and one to staff.

The results:

• 100% of inbound leads captured and answered, regardless of time of day or staff availability.
• 4 → 1: Instagram (two accounts), Facebook Messenger, TikTok and phone unified into one intake pipeline.
• 0 duplicate leads entering the pipeline, thanks to de-duplication by phone number and session.
• 5 qualification gates applied automatically before anything reaches an attorney.
• 24 hrs of automated reminders before every booked consultation.

The point for other firms is the design, not the channel list: answer instantly, qualify the same way every time and keep attorney time for matters the firm will take. If your firm handles injury cases, our resources for personal injury law firms go deeper on intake.

Where to start

Pull a month of call logs and answer three questions: how many calls went unanswered, when they came in, and what share were new matters. That tells you whether the gap is after hours, overflow or the whole day. Then write your intake questions and the sentences the receptionist must never say. Those two documents are most of the work. Once live, track answer rate by hour, consults booked and kept, transfers connected and every call where the agent misfired.

If after-hours and overflow calls are the gap, start there: AI covers the hours you can’t, and your team keeps the business-hours calls while you build trust in it. For more on how this fits a legal practice, see our page on AI for law firms and how we build AI voice agents and chat agents for social DMs.