AI receptionist for law firms: which option fits, what it costs and how to stay within the ethics rules

The short answer

An AI receptionist for law firms is a voice agent that answers the calls your staff cannot, asks the intake and conflict questions they would ask, books consultations on the right attorney's calendar and sends urgent or existing-client calls to a person. It pays off when prospective clients call after hours, at lunch or while your one receptionist is on another line, because a caller who reaches voicemail usually calls the next firm on the list.

You have three ways to get one: the intake AI built into your practice management or intake software (Clio launched Grow AI with a voice agent in August 2026), a standalone AI or hybrid answering service such as Smith.ai, or a custom voice agent built on your own phone system and data. If your firm runs on Clio and its intake flow matches yours, try the built-in option first. If you need practice-area-specific screening, several systems or your own reporting, a standalone service or a custom build fits better.

Why law firms miss so many calls

The problem is well measured. For its 2024 Legal Trends Report, Clio had researchers pose as prospective clients and contact 500 US law firms. Only 40% of firms answered the phone, down from 56% in 2019. Just 33% replied to email inquiries, down from 40%. Taking callbacks into account, 48% of firms were effectively unreachable by phone.

The same study found that prospects who actually spoke with someone were far more likely to recommend the firm than those who reached voicemail. When Clio launched Grow AI in August 2026, it cited its own research that 79% of prospective clients expect a reply within 24 hours (LawNext, August 2026).

Most small firms do not have a staffing problem during office hours so much as a coverage problem around them: evenings, weekends, lunch, court days and the minutes when two lines ring at once. Those are the calls an AI receptionist is built for. The broader AI automation for law firms page covers the document, billing and matter workflows that sit behind intake.

What a legal AI receptionist should handle, call by call

Before comparing vendors, list the calls your firm gets and decide what the AI does with each. Demos focus on how natural the voice sounds; what matters is whether it handles each of these correctly.

Call type	What the AI should do	Hand to a person when
New prospective client	Disclose it is an automated assistant, get the caller's name and contact details and the names of other parties, run the conflict pre-screen, then ask the practice-area screening questions and offer consultation slots	A possible conflict is flagged, or the matter is outside your practice areas and you want a human referral
Urgent matter	Recognize signals such as a recent arrest, a hearing in the next few days, a protective order or a deadline about to pass, and escalate immediately	Always, by warm transfer or text alert to the on-call attorney
Existing client	Identify the client and matter, take a message, create a task for the responsible attorney or paralegal	The client is upset or says it is urgent. The AI should never discuss case strategy or status beyond what you approve
Opposing counsel, courts, adjusters	Take a detailed message and route it to the matter team	Your policy says these always go to a person in office hours
Billing and payment questions	Send a payment link from your billing system if you allow it, otherwise take a message	Fee disputes
Questions asking for legal advice	Say it cannot give legal advice and offer a consultation	Never give advice, even general guidance
Vendors, solicitors, spam	Take a message or end the call politely	Never

The rule of thumb: let the AI book consultations and capture details on everything else, and keep urgent matters, existing-client substance and anything resembling advice with lawyers and staff.

Clio Grow AI, standalone services or a custom build: the options compared

Clio's entry into this market changed the buying decision for the many firms on Clio. Here is what each type of option publishes, checked in October 2026.

• Clio Grow AI (built in). Launched on 20 August 2026, Grow AI monitors a firm's phone lines, inboxes and website chat, answers questions, captures contact and case details, checks leads against the firm's criteria, books qualified consultations and follows up by SMS and email. The voice agent is available in the US only. Pricing is $25 per converted lead, charged when a client captured or revived by Grow AI signs the engagement agreement; existing Grow customers get their first three conversions free and can set monthly spending targets (LawNext, August 2026).
• Smith.ai (standalone AI, with human options). Smith.ai's AI Receptionist pricing page lists a free plan with 25 calls a month and $3 per extra call, a Pro plan from $150 a month covering 75 to 300 calls at $2.00 to $1.67 per call, and an Enterprise plan from $500 a month (checked October 2026). It also sells human-first receptionist plans.
• Ruby (human virtual receptionists). Not AI, but the usual comparison. Ruby's pricing page lists $250 a month for 50 minutes, $395 for 100, $720 for 200 and $1,725 for 500 (checked October 2026).
• Custom voice agent. Built on a usage-based voice platform such as Retell AI, connected to your phone system, practice management and CRM, with your own scripts, conflict logic and reporting.

Question	Built into Clio Grow	Standalone AI or hybrid service	Custom voice agent
Setup effort	Lowest if you already use Grow	Low to medium; needs an integration to book	Highest; scoped and built for you
Pricing model	Per converted lead	Per call or per minute, monthly plans	Per minute usage plus a one-time build
Conflict checking	Against Clio data, within Clio's settings	Varies; ask exactly what is checked	Your own rules against Clio, MyCase, PracticePanther or a spreadsheet
Practice-area screening	Firm criteria within the product's options	Script-based	Full control, including different flows per practice area or office
Works across systems	Clio-centered	Some integrations	Yes: phone, PMS, CRM, CallRail, ad sources, dashboards
Who owns scripts, data and number	Clio account	The vendor	You, in your own accounts
Best for	Clio firms happy with standard intake	Firms wanting a quick start with human backup	Multi-practice, multi-office or high-volume firms with specific rules

What an AI answering service for a law firm costs

Costs come in three parts: the platform or service fee, usage (calls, minutes and phone numbers) and the setup work to connect booking and conflict checks. Here is a worked example using the published prices above.

Assumptions. A firm sends 200 after-hours and overflow calls a month to the AI, averaging 4 minutes each (800 minutes), and signs 10 new clients a month from those calls.

Option	Estimated monthly cost	How it is calculated
Clio Grow AI	About $250	10 converted leads at $25 each; you pay nothing for calls that do not convert, on top of your Grow subscription
Smith.ai AI Receptionist	Roughly $334 to $400	200 calls at the published $1.67 to $2.00 per-call range; confirm which tier 200 calls falls into
Custom agent on Retell AI	About $56 to $248 in usage	800 minutes at Retell's published $0.07 to $0.31 per minute, plus $2 a month per phone number and a one-time build
Ruby human receptionists	More than $1,725	800 minutes is above Ruby's largest published plan of 500 minutes for $1,725

The staffing comparison. The US Bureau of Labor Statistics puts the median receptionist wage at $18.27 an hour, or $38,010 a year (May 2025). A receptionist covers office hours, not the evenings and weekends when many prospective clients call, so the fair comparison is usually AI versus voicemail or AI versus a live answering service, not AI versus your front desk.

Your own break-even. Count missed and after-hours calls from your phone system or CallRail, estimate how many were real prospects, and multiply by your consult-to-client rate and average fee. In many practice areas, one additional signed matter a month covers any of the AI options above. Our guide to AI automation agency costs explains how custom build fees are usually priced.

Ethics, confidentiality and conflict checks

This is practical guidance for planning a project, not legal advice. Your firm's ethics counsel or state bar guidance should approve your scripts.

• ABA Formal Opinion 512. Issued on 29 July 2024, it applies existing Model Rules to generative AI tools: competence (Rule 1.1), confidentiality (Rule 1.6), communication with clients (Rule 1.4), supervision of nonlawyer assistance (Rules 5.1 and 5.3) and reasonable fees (Rule 1.5). For an AI receptionist, that means knowing how the vendor stores and uses call data, vetting its security, and supervising what the agent says the way you would supervise a new receptionist.
• Prospective clients and Rule 1.18. Information a prospective client shares can be confidential even if you never take the case, and can disqualify you from representing the other side. The comments to Model Rule 1.18 suggest limiting the initial interview to what is reasonably needed to decide whether to take the matter. In practice, the AI should collect names of all parties and run the conflict screen before asking for the story.
• No legal advice. The agent should never tell a caller whether they have a case, what it is worth or what to do next. A fixed answer that offers a consultation works.
• Disclosure and recording. Tell callers they are speaking with an automated assistant and that the call may be recorded. Some states, including California, require every party's consent to record a call.
• Outbound calls are different. On 8 February 2024 the FCC ruled that AI-generated voices count as "artificial" under the TCPA, so outbound AI calls need the same consent as prerecorded calls. Consultation reminders and follow-ups need consent records and opt-out handling.
• Data location and retention. Ask where recordings and transcripts are stored, how long they are kept, whether they are used to train models and how you delete them.

How a conflict pre-screen works. The agent asks for the caller's full name, the other parties' names and any businesses involved, then searches contacts and matters in your practice management system through its API. An exact or close match does not reject the caller: it pauses intake, tells the caller someone will follow up, and sends the details to staff for a proper conflict check. The AI flags; a person decides.

How to set up an AI receptionist for your law firm, step by step

1. Pull your call data. Export 60 to 90 days of calls from your phone system or CallRail: calls by hour, missed calls, voicemails and sources. This shows where to start, usually after hours and overflow.
2. Write your intake scripts by practice area. Personal injury, family, criminal defense and estate planning need different screening questions, disqualifiers and urgency signals. Keep each script short.
3. Define the conflict pre-screen. Decide which names to collect, which systems to search and what happens on a possible match.
4. Decide routing. After hours only, overflow after a set number of rings, or first answer for new-client lines with transfer to staff. Start narrow.
5. Connect booking and records. Give the agent access to the right calendars and permission to create leads in Clio Grow, Lawmatics or your CRM, using a service account rather than one person's login.
6. Set urgent escalation. Warm transfer to the on-call attorney, then a backup, then a text alert with the transcript if nobody answers within a set time.
7. Test on real calls. Replay 30 to 50 transcripts of past calls, including distressed callers, people asking for advice, Spanish speakers if you serve them and callers whose names match existing contacts. Fix every wrong booking and missed escalation.
8. Go live in shadow mode. For the first one to two weeks, have staff review every AI-booked consultation and every flagged conflict each morning.
9. Review monthly. Track answered, booked, escalated and abandoned calls, and consult-to-client conversion for AI-booked consultations.

A built-in option can be switched on quickly, though the testing steps still apply. A custom voice agent with conflict checks, booking and reporting typically takes 4 to 8 weeks in stages.

Questions to ask before you choose a legal answering service

• Can it create a lead or matter in my practice management system, or does it only send an email?
• What exactly does the conflict check search, and what happens on a partial name match?
• Can I set different scripts for each practice area, office and phone number?
• How does it recognize urgent matters, and who does it call or text?
• Where are recordings and transcripts stored, for how long, and are they used to train models?
• How is it billed in a heavy month, and what does an overage cost?
• Can I export every recording, transcript and call outcome?
• Who owns the phone number and the scripts if I cancel?
• Does it handle texts and STOP requests for follow-ups?
• Can it reach a human backup when callers ask for a person?

For the wider questions to ask any automation partner, see how to choose an AI automation agency. If you also run trades or service businesses, our AI receptionist for HVAC guide shows how the same approach adapts to dispatch.

Common mistakes, and when not to buy

• Taking the full story before the conflict screen. It creates Rule 1.18 exposure and wastes the caller's time if you cannot act.
• Letting the AI drift into advice. "You probably have a strong case" is exactly what it must never say. Test for it.
• Booking every caller. Without practice-area and jurisdiction screening, attorneys spend their week on consultations that were never a fit.
• Nobody reads the unbooked calls. Every call that ended without a booking should land in a morning callback list.
• Measuring calls answered instead of clients signed. The number that matters is signed engagements from AI-handled calls.

Skip it or wait if: your staff or a live answering service already answers nearly every call and you are happy with conversion; your firm takes very few new matters a month; or your work is mostly referrals from other lawyers who call attorneys directly. If your real problem is slow follow-up after the first call, a CRM automation project for intake follow-up may pay back faster.

Getting an AI receptionist built for your law firm

If you run on Clio and Grow AI matches your intake, try it. Bring in an agency when you have several practice areas or offices, need custom conflict and urgency rules, use systems beyond Clio, or want intake results in one dashboard next to your marketing spend.

We build AI receptionists and AI voice agents for law firms on Clio, MyCase, PracticePanther and Lawmatics. Every project starts with a free review of your call data, gets a fixed price, ships in stages and runs on accounts you own, including the phone numbers, scripts and code.

To get started, book a free review and tell us which practice management software you use and roughly how many new-client calls you miss each week.

FAQ

Frequently asked questions