24/7 answering with sub-second pickup
Every inbound call gets picked up in under a second, day or night. You can run a single greeting or split it into a separate after-hours flow, whichever fits your firm.
Industry · Legal · ABA Rule 5.3 framework · Live in 1 to 3 weeks
AI receptionist for law firms: conflict-aware intake, booked into Clio, 24/7
Your phone rings at 6:47 PM on a Tuesday, on a Saturday morning, on a Wednesday night before a hearing. Most of those callers are new clients deciding between you and the firm that picked up. Scale me AI builds and operates an AI receptionist that answers in under a second, runs a conflict check against your PMS before opening any file, books the consult into Clio Grow, Cal.com, or Calendly, and escalates emergencies to your on-call attorney. TCPA-safe outbound by default. Built on our AI voice agent service.
- TCPA-safe outbound
- ABA Rule 5.3 + Formal Op 512
- Conflict check wired into your PMS
- Bilingual EN + ES
AI Legal Receptionist
Idle
Demo · Scale me AI's own AI receptionist
Why law firms ship with Scale me AI
- ABA Formal Opinion 512 framework documented per engagement
- TCPA-safe outbound by default (SMS-first, AI-voice with consent only)
- PMS write-back via Clio, MyCase, PracticePanther, Filevine, Smokeball
- 1 to 3 weeks to live, custom-scoped
The intake call problem
The intake call problem most US law firms haven't solved
Clio's 2025 Legal Trends Report finds the average lawyer bills 2.6 hours of an 8-hour day. The other 5.4 hours include the intake calls your firm has to answer to keep growing. Most US small and mid-size firms run that intake on a paralegal who is already double-booked, or a voicemail box that quietly costs the firm a new case every week.
Today, without an AI receptionist
- After-hours new-client calls go to voicemail and never call back
- Spam screening eats your paralegal's day
- Conflict checks happen after the intake form, not before the call
- Outbound recall to old leads risks TCPA exposure
- Bilingual Spanish-speaking callers hit an English-only flow
What we build
We build and operate an AI receptionist that runs the new-client intake flow your firm would design if you had time to design it. We map your call volume, write the script per practice area, wire the conflict check into your PMS, and then operate the agent as a managed service after launch.
You stop running intake. We run it for you.
After launch (1 to 3 weeks in)
- Every call answered in under a second, 24/7
- Conflict check runs as a PMS lookup before the AI opens any record
- Consult booked into Clio Grow, Cal.com, or Calendly automatically
- Urgent matters escalated to your on-call attorney by name
- Outbound recall stays inside TCPA-safe patterns
Source: Clio Legal Trends Report 2025, “lawyers bill 2.6 hours of an 8-hour day.” clio.com/blog/lawyer-statistics
What it does
What the AI receptionist actually does on every call
Six capabilities your firm uses on day one. Each one is configurable per practice area at build, and tuneable per attorney after launch.
Conflict check before any record opens
Before the AI collects any new-client information, it runs a PMS lookup on the caller's name and matter. If a conflict surfaces, the AI declines politely and flags it for your review.
New-client intake by practice area
Custom script per practice area: PI, criminal defense, family, immigration, estate planning, employment, mass tort. The AI collects the same structured fields your paper intake form already collects, in your firm's voice.
Direct booking into Clio Grow, Cal.com, Calendly
The booking writes straight into your scheduling tool of record. Confirmation SMS goes out automatically, and the appointment shows up to your team inside Clio Grow within seconds.
Live transfer for emergencies
Statute-of-limitations cues, criminal arrest, restraining order, immediate threats: the AI routes the caller to your on-call attorney by name, on the line. The escalation tree is configured in Day 0 to 3. Not to a general voicemail.
Bilingual English + Spanish
The AI picks up the caller's language preference in the first sentence, so a Spanish-speaking caller never gets forced through an English-only flow. Both languages run on the same intake script.
Pricing
How much does an AI receptionist for a law firm cost?
Typical all-in cost for US small-firm AI-receptionist services runs $400 to $1,500 per month plus a one-time build fee of $1,500 to $3,500. That is the competitor-band range across the market, not a Scale me AI quote. Scale me AI is custom-scoped: book a discovery call and you get a fixed-scope proposal in three business days.
Market pricing across 6 named legal-vertical answering services (US, 2026)
| Service | Starting monthly | What's included | AI / human / hybrid | Source |
|---|---|---|---|---|
| Scale me AI | Custom-scoped (book a discovery call) | Build once, operated as a managed service; voice-stack math published below | AI + human escalation | scalemeai.com |
| Smith.ai | $95 to $1,950 / mo | AI tier ($95 to $825) or human tier ($285 to $1,950); 30 to 300 calls per tier | Hybrid (AI tier + human tier) | MyCase, Jan 2026 |
| LEX Reception | $425 to $775 / mo | 150 to 500 min per tier; $75 setup on Entry tier; legal-only | Human-first | MyCase + LegalSoft, 2026 |
| Ruby Receptionists | $235 to $1,695 / mo | 50 to 500 min per tier; premium human positioning, no AI | Human-only | MyCase + LegalSoft + Embroker |
| Answering Legal | $330 to $737 / mo | 100 to 250 min per tier; $3.91 per min on starter; legal-vertical | Human-first | MyCase + LegalSoft |
| Posh Virtual Receptionists | $65 to $1,900 / mo | $65 base with no minutes; 50 to 1,000 min plans on top | Human-first | Embroker + LegalSoft |
Voice-stack transparency
What the underlying voice stack actually costs
Most legal answering services will not show you the wholesale cost of the AI itself. We will. Here is the published per-minute and per-month math we use to scope every legal engagement.
- VapiVoice platform (Build tier, per minute): $0.05 per minute on callsvapi.ai/pricing
- RetellVoice platform (PAYG, per minute): $0.07 to $0.31 per minute total stackretellai.com/pricing
- ElevenLabsTTS subscription: $6 to $990 per month (Starter to Business)elevenlabs.io/pricing
- TwilioUS telephony: $0.0140 per minute outbound, $0.0085 per minute inbound localtwilio.com/en-us/voice/pricing/us
- Cal.com OrganizationsScheduling (Org / SOC 2 + HIPAA tier): $28 per user per month (yearly)cal.com/pricing
That is the wholesale infrastructure the call rides on. The Scale me AI monthly retainer wraps build, ops, transcript review, prompt tuning, and integration maintenance on top of that stack.
Get a custom proposal in three business days.
Practice areas
Which practice areas get the most lift from an AI receptionist
These six practice areas cover almost all the AI-receptionist work we do for US law firms. The intake call profile is different in each one. So are the urgency cues, and so is the build. We script per area, not generically.
Personal injury
The highest deal-value vertical we deploy. Heavy after-hours urgent inbound, heavy Spanish-speaking caller volume, intake fields built around incident type, date of accident, injuries, treatment status, and insurance status. PI firms also carry the heaviest outbound recall volume of any vertical we work in, which is exactly where the TCPA-safe configuration earns its weight on day one.
Criminal defense
Urgent inbound at any hour. The AI confirms jurisdiction, charge type, arrest status, and custody status, then triggers an immediate human transfer if the caller is at the station or in custody. Statute-of-limitations cues and bond hearings route to your on-call attorney by name. Not to a general voicemail.
Family law
Emotional triage. The AI is configured warm, slow-paced, and asks about safety first. It never gives advice. Restraining-order matters and any indication of immediate danger trigger an immediate transfer to the on-call attorney. Intake captures custody, separation date, and the structured fields you would collect on a paper form.
Immigration
Bilingual Spanish is mandatory here, not optional. Intake asks about visa type, USCIS case status, deadlines, and prior counsel. Long booking windows are normal in this practice area, so the AI books 2 to 4 weeks out by default and sends an SMS reminder 48 hours and 24 hours before the consult.
Estate planning
Lower urgency, higher booking-rate. The AI qualifies on net-worth bracket, document scope (will, trust, healthcare directive), and any pending health issue. It books a consult 1 to 2 weeks out and sends a pre-call document checklist by SMS so the consult starts with the right context in the room.
Employment law
Time-sensitive because statutes of limitations bite hard here. The AI captures employer name, separation date, claim type (discrimination, wage and hour, wrongful termination), and surfaces the SOL window in the booking note so your intake attorney sees the deadline before the call begins.
We've built the same playbook for dental practices: the call profile is different, the structure is the same.
Comparison
AI receptionist vs traditional legal answering service vs DIY
No single option is universally right. The honest read on five picks:
- Smith.ai is the right pick if you want a hybrid AI-plus-human service with a tight pricing tier.
- Ruby Receptionists is the right pick if you want premium human-first service with no AI involved.
- LEX Reception is the right pick if you want a legal-only human service with mid-market pricing.
- DIY on Vapi or Retell is the right pick if you have a developer on staff and want to own the build.
- Scale me AI is the right pick if you want the AI receptionist built once, operated as a managed service, with TCPA-safe outbound and an ABA Rule 5.3 supervisory framework documented.
| Capability | Scale me AI | Smith.ai | Ruby | LEX | In-house hire | DIY (Vapi / Retell) |
|---|---|---|---|---|---|---|
| Starting cost | Custom-scoped + voice-stack math published | $95 / mo (AI tier) | $235 / mo | $425 / mo | $45k to $65k / yr fully loaded | Wholesale + dev cost |
| AI / human / hybrid | AI + human escalation, managed | Hybrid (AI tier + human tier) | Human-only | Human-first | Human-only | AI-only (you operate) |
| Bilingual EN + ES | Yes, on the same intake script | Yes | Yes | Yes | Hire-dependent | Yes (you configure) |
| Legal PMS integrations | Clio, MyCase, PracticePanther, Filevine, Smokeball via official APIs | Several PMS named, varies by tier | Implied (no specific legal PMS named) | Clio, MyCase, PracticePanther | Manual data entry | You build it |
| TCPA-safe outbound by default | Yes; SMS-first, AI voice with consent only | Outbound is human-led, not AI | No automated outbound | No automated outbound | N/A | You configure |
| ABA Rule 5.3 documented at build | Yes; documented in Day 0 to 3 | Not mentioned | Not mentioned | Not mentioned | Implicit (lawyer supervises) | You document |
| Ongoing managed-service operation | Daily transcripts first 30 days; monthly performance report | Per tier | Continuous (human staff) | Continuous (human staff) | Continuous | You operate |
- If you want pure human service: Ruby Receptionists or the Smith.ai virtual receptionist tier.
- If you have a developer on staff and want to own the build: DIY on Vapi or Retell.
- If you want the AI built once, operated as a managed service, with the legal-compliance layer written down per engagement: that's the Scale me AI position.
Want the matrix with your firm's call volume plugged in?
Process
How Scale me AI deploys a legal AI receptionist (1 to 3 weeks)
Most US solo and small firms go live in 1 to 3 weeks. Mid-size firms (10 to 25 attorneys) take 3 to 6 weeks. Four phases, and we sign the engagement-level confidentiality and DPA in parallel with discovery.
1
Day 0 to 3Discovery and scope
A 30 to 45 minute call. We audit your missed calls, map call volume by hour and by day of week, identify your top intents (new-client inquiry, conflict check, urgent matter, billing, existing-client question), confirm your PMS, and document 3 to 5 of the objections your prospects raise today. The engagement-level confidentiality and DPA get signed in this phase, so the build starts clean.
2
Day 3 to 10Build and integrate
Script per practice area, voice selection (studio voices only, no attorney voice cloning), conflict-check wiring into your PMS read API, escalation rules wired to your on-call attorney by name and number, SMS confirmation flow, and TCPA-safe outbound configuration with documented-consent gating. We also write down the ABA Rule 5.3 supervisory framework for your engagement record. In writing, on file.
3
Day 10 to 14Test and soft launch
We run 50 to 200 synthetic test calls against your actual call flows. Soft launch goes live on one line first, while your team watches transcripts and signs off on urgency-tree behavior. We tune the prompt and the escalation rules before the AI sees your full call volume.
4
Day 14 onwardOperate
Daily transcript review for the first 30 days. Monthly performance report covering calls handled, calls booked, calls escalated, calls failed, cost per call, and no-show recovery. Ongoing as a managed service: when ABA guidance changes, when the FCC clarifies a rule, when your PMS releases a new API version, we ship the update on your line. You don't have to chase it.
Step 1 starts with a 30 to 45 minute call. We run a missed-call audit on your firm during the session, so you leave with your own recovery math.
Integrations
Integrations that ship on day one
Two tiers. The first is your firm's system of record. The second is everything that routes around it. We use official APIs, not screen-scraping.
Tier 1: Legal PMS
- Clio
- Clio Grow
- MyCase
- PracticePanther
- Filevine
- Smokeball
Tier 2: Scheduling, workflow, voice stack
- Cal.com
- Calendly
- Zapier
- n8n
- Twilio
- ElevenLabs
If your firm runs a PMS without a public API (CasePeer is the example we see most often), we'll say so upfront. We then propose either a partner-tier workaround or a PMS migration before any work starts. We won't dress up screen-scraping as a real integration. The privilege and confidentiality risk is too high to fake it.
Compliance
Ethics and compliance: the layer most AI vendors won't write about
An AI receptionist that handles US law-firm calls touches at least five regulatory domains: ABA Model Rules (1.6, 1.18, 5.3, 5.5, 7.1 to 7.3), ABA Formal Opinion 512 (July 2024), the TCPA (with the FCC's February 2024 AI-voice clarification for outbound), state bar AI ethics opinions (California and Florida have current guidance; New York runs a task force), and state privacy laws (CCPA when CA-resident clients). HIPAA usually does NOT apply.
ABA Model Rule 1.6 (Confidentiality)
The rule: lawyers must protect information relating to the representation of a client, including from vendor mishandling. What we do: vendor data-use rights restricted to the engagement, transcripts encrypted at rest and in transit, signed DPA with Scale me AI before any audio is captured.
ABA Model Rule 1.18 (Prospective Client)
The rule: information shared by a prospective client gets confidentiality protection even when no engagement forms. What we do: every inbound call carries Rule 1.18 protections from the first word, with no marketing reuse of inbound-call content and no cross-firm training on transcripts.
ABA Model Rule 5.3 + ABA Formal Opinion 512 (July 2024)
The rule: a lawyer must reasonably supervise nonlawyer assistance, and under Formal Op 512 the supervisory duty extends to GenAI tools across six duties (competence, confidentiality, communication, candor, supervision, fees). What we do: the supervisory framework gets written down during Day 0 to 3 and stored on your engagement record.
ABA Model Rule 5.5 (Unauthorized practice of law)
The rule: only lawyers admitted in a jurisdiction may give legal advice in that jurisdiction. What we do: a hard rule in the system prompt says the AI must NEVER give legal advice. Every case-specific question routes to a human transfer or a callback request, and the prompt is verified during testing in Day 10 to 14.
ABA Model Rules 7.1 to 7.3 (Advertising and solicitation)
The rule: lawyer communications must not be false or misleading, and live solicitation of prospective clients is restricted. What we do: the AI identifies itself as automated at the start of every call per 47 CFR 64.1200(b), and the outbound recall flows are reviewed against Rule 7.3 solicitation rules during build.
TCPA + FCC February 8, 2024 declaratory ruling
The rule: AI-generated voice is treated as “artificial voice” under the TCPA, which requires prior express written consent for marketing calls to most numbers. What we do: outbound is SMS-first by default, AI-voice outbound runs only against numbers with documented consent collected at intake, and EBR-eligible existing-client outbound is a separate gated flow.
State bar AI ethics opinions
The rule: the California State Bar published GenAI practical guidance in November 2023, and the Florida Bar issued Ethics Opinion 24-1 on lawyer GenAI use on January 19, 2024 (New York currently runs a state-bar task force). What we do: we map the per-state requirements that apply to your firm in Day 0 to 3 and write them into the engagement record.
HIPAA (only when it applies)
The rule: HIPAA covers PHI handled by covered entities and their business associates, and it usually does not reach law firms. What we do: when your firm handles PHI on behalf of a healthcare client (typical only for some PI firms running medical-record discovery), we sign a BAA, route the call through a HIPAA-enabled voice provider tier, and add that cost into the engagement scope.
CCPA / CPRA
The rule: if your firm collects personal information from California-resident clients, California privacy law requires disclosure, retention controls, and an opt-out path. What we do: retention and consent flows are configured to your firm's records-retention policy in Day 0 to 3 (default 7 years, the US median), with an opt-out path accessible from every confirmation SMS.
Regulatory references
- 47 CFR 64.1200 (Cornell LII) (TCPA implementing regulation; AI-voice identification)
- en.wikipedia.org/wiki/TCPA (artificial-voice classification, FCC Feb 2024 ruling)
- en.wikipedia.org/wiki/ABA Model Rules
- Thomson Reuters summary of ABA Formal Op 512
- en.wikipedia.org/wiki/HIPAA (scope of HIPAA; business-associate trigger)
Ask us how we handle compliance for your jurisdiction.
Where this breaks
Where AI receptionists fail (and what we do about it at build time)
Every AI receptionist will fail somewhere. Here are the five places we've watched them fail in legal-vertical builds, and what we do about each one at build time. The point of this section is not reassurance. The point is that you should ask any AI vendor the same five questions on a sales call, ours included.
1. UPL slip: AI giving legal advice
A caller asks “can I sue for this?” and the AI tries to answer. Mitigation: hard rule in our system prompt that the AI must NEVER give legal advice. Any case-specific question routes to a human transfer or a callback request, and the prompt is verified against 50 to 200 synthetic test calls during Day 10 to 14.
2. Urgency mis-classification: emergency call not escalated
An immigration caller hits the line with a deportation deadline tomorrow morning. Mitigation: escalation rules wire specific urgency cues to specific phone numbers during Day 0 to 3 discovery, and the urgency tree is tested with synthetic calls before launch. If the cue is missed, the caller is offered a same-day callback as a safety net.
3. PMS write-fail: booking succeeds in the AI but doesn't appear in Clio
The AI confirms the appointment, the caller hangs up, and nothing lands in your PMS. Mitigation: we use official Clio, MyCase, PracticePanther, Filevine, and Smokeball APIs (no screen-scraping), and we verify write-success on every booking with a synchronous read-back before the call ends.
4. TCPA outbound exposure: AI calling without documented consent
AI-voice outbound to a prospect without prior express written consent risks $500 to $1,500 per call under the TCPA, as clarified by the FCC's February 8, 2024 declaratory ruling. Mitigation: outbound defaults to SMS-first. AI-voice outbound is gated on documented consent collected at intake, with EBR existing-client flows handled as a separate audited path.
5. Voice-cloning consent: synthetic voice imitating an attorney
Illinois (BIPA), Texas, and California treat synthetic voice as a tort if used without recorded consent. Mitigation: our default for the legal vertical is simple: we do NOT clone any attorney's voice. Studio voices only. The decision is written into the engagement record so the firm has a clean paper trail.
ROI
The ROI math for a 1 to 25-attorney US law firm
Clio's 2025 Legal Trends Report finds lawyers bill an average of 2.6 hours of an 8-hour day. UC Berkeley HR research finds it takes about 25 minutes to fully refocus after an interruption. MyCase's 2024 Legal Industry Trends Report finds more than one in three firms gain 1 to 2 leads per week after deploying virtual receptionist services. Those three numbers are where the ROI lives.
A worked example (illustrative, not a Scale me AI client metric)
Take a solo PI attorney with an average case value of $4,200 who is missing 8 calls per week today. In practice, that profile is the median small PI firm we get on a discovery call.
- 8 calls per week × 50 weeks = 400 missed calls per year
- AI receptionist captures 25 percent of those: 100 calls answered and qualified
- 10 percent of qualified intake calls convert to signed cases at this firm: 10 new cases per year
- 10 cases × $4,200 average case value = $42,000 per year of recovered case revenue
Set that against a typical $7,200 per year market cost for an AI-receptionist build and ops in the legal vertical (the competitor-band data from the pricing section). The math holds in either direction. Book a discovery call and we'll plug your firm's actual missed-call rate, average case value, and current intake-conversion rate into the same model. When outbound is in scope, we also wire the recall side of this into our lead-generation automation for law firms.
Sources for this section
- Clio Legal Trends Report 2025: clio.com/blog/lawyer-statistics
- MyCase 2024 Legal Industry Trends Report: mycase.com/reports/2024-legal-industry-report
- UC Berkeley HR (25-minute refocus cost of interruption): hr.berkeley.edu
Book a discovery call and we'll run the math with your firm's numbers.
Frequently asked questions from managing partners
Should I tell my clients they are speaking to an AI?
Yes. 47 CFR 64.1200(b) requires identifying the responsible business at the start of any prerecorded or AI-voice call and offering an opt-out within two seconds of identification. Beyond the rule, transparency reduces caller friction; in our experience callers tolerate an AI receptionist far better when the AI says so at the open. We configure every legal-vertical AI receptionist to identify itself on every call, in plain language, in your firm's brand voice.
Is an AI receptionist HIPAA-compliant for a law firm?
Usually not relevant. HIPAA applies to law firms only when your firm handles PHI on behalf of healthcare clients, typical only for some PI firms running medical-record discovery. When it applies, you need a BAA with Scale me AI AND with the underlying voice provider, and we route the call through a HIPAA-enabled provider tier. For most law firms the load-bearing rules are ABA Model Rules and state bar AI guidance, not HIPAA.
What happens if a caller says it's an emergency?
The AI is configured to detect urgency cues (immediate danger, threat to life, statute-of-limitations deadlines, criminal arrest, restraining order, custody status) and to transfer the call to your on-call attorney's phone or page the office. Escalation rules are wired during Day 0 to 3 discovery, and the urgency tree is tested with synthetic calls before launch. The AI never gives legal advice. Every urgent matter routes to a human.
Can the AI do outbound recall calls to old leads?
Yes, but only inside TCPA-safe patterns per the FCC's February 8, 2024 declaratory ruling. AI-generated voice is "artificial voice" under the TCPA, which requires prior express written consent for marketing calls to most numbers. Existing-client outbound on the established-business-relationship exception is the safer path; cold outbound is restricted. We default outbound to SMS-first; AI-voice outbound is enabled only on numbers with documented consent.
Will the AI work with Clio, MyCase, PracticePanther, Filevine, or Smokeball?
Yes. All five have documented integration paths via official APIs, not screen-scraping. Appointments the AI books are immediately visible to your team in your PMS, and we verify write-success on every booking with a synchronous read-back before the caller hangs up. If your firm runs a PMS without a public API (CasePeer is the common example), we'll say so upfront and propose either a partner-tier workaround or a PMS migration before any work starts.
Still have questions?
Related services
Where this fits in the rest of the Scale me AI stack
Workflow
When the call ends, the back-office work begins
We build the n8n and Make automations that push the booked consult into your case management, draft the engagement letter, send the document checklist, and update your conflict log.
See workflow automation→Lead gen
Outbound and inbound pipelines that respect TCPA
When you want recall and reactivation outbound to old leads on top of the inbound receptionist, we build the lead-gen automation that stays inside the TCPA and the established-business-relationship rules.
See lead-gen automation→Sister vertical
Same playbook, different vertical
We built the conflict-aware, PMS-integrated receptionist pattern first for US dental practices. If you have a friend running a dental office, this is the page to send them.
See the dental playbook→Ready to stop losing intake calls?
Book a 30 to 45 minute discovery call. We'll audit your current intake flow, map your call volume, and tell you honestly whether an AI receptionist is the right pick for your firm right now. Custom-scoped proposal in three business days. No spec work, no boilerplate quote.