An AI voice agent for legal intake is a voice AI that picks up potential-client calls with empathy-first scripting, runs conflict checks against Clio, MyCase, or PracticePanther in real-time, triages statute-of-limitations urgency, captures detailed intake notes, and books qualified consults. Mid-market law firms recover an average $95,000 per year per attorney by capturing the 70%+ of intake calls that currently leak.
TL;DR
- Law firms capture under 30% of inbound intake calls. Most callers reach voicemail or staff who cannot triage urgency.
- Empathy is scripted, not improvised. Active acknowledgement plus pacing plus explicit no-legal-advice posture.
- Conflict checks run in real-time. Clio, PracticePanther, MyCase integration during the call.
- Statute-of-limitations triage is the highest-stakes loop. 30-day SOL window auto-transfers to attorney.
- $95k/year per attorney recovered. 5-day deploy.
Where the leak shows up · Five legal intake loops · Empathy-first script deep-dive · Conflict check deep-dive · Statute-of-limitations triage · PMS integrations · ABA Model Rules + bar compliance · 5-day deploy · ROI math · Failure patterns · FAQ
1. Where law firms leak intake
Run the math on your last week. Across the 5 legal intake audits luup ran in 2026, median intake-call capture rate was 28%. The other 72% reached voicemail, were placed on hold and abandoned, or were screened by staff who could not triage urgency or run conflict checks. Of the 28% captured, only 60-70% converted to consultation bookings due to scheduling friction.
The leak compounds across three structural failures. First, intake calls arrive at unpredictable times - personal injury after accidents, family law during emotional moments, criminal defence at arrest. The standard 9-5 intake desk misses 40-55% of these. Second, conflict checking is manual and slow; firms either delay the consult by 24-48 hours to run conflicts (losing the caller to a competitor) or skip the check (creating malpractice exposure). Third, SOL triage requires legal judgment that intake staff lack; urgent cases get treated like routine ones and miss the filing window.
Legal ranked best in our cross-vertical 47-deployment voice audit at 60% failure rate (lowest among 8 verticals). The reason: intake scripts are narrower and better-defined than other verticals. The same audit found legal deployments that succeeded did three things consistently - empathy-first scripting, real-time conflict checks, and SOL triage with attorney escalation.
2. Five loops a legal intake voice agent closes
Five loops cover the operational surface for a typical mid-market law firm (1-15 attorneys, 2-5 practice areas, $1-15M annual revenue). Each has defined triggers, integration paths, success metrics.
2.1 Loop 1 - Inbound intake (24/7)
Trigger: any call to the main line that goes unanswered after 3 rings, or any call after hours and weekends. Data path: Twilio inbound to Vapi or Retell agent to Clio, PracticePanther, or MyCase for matter creation. Success metric: 95%+ inbound calls answered within 6 seconds.
2.2 Loop 2 - Conflict check during call
Trigger: caller and opposing party identified during intake. Data path: PMS conflict-check API to result in 5-15 seconds. Success metric: 100% of intakes conflict-checked before consult booking.
2.3 Loop 3 - Statute-of-limitations triage
Trigger: incident date plus jurisdiction captured. Data path: SOL matrix lookup, urgency score, routing decision. Success metric: 100% of SOL-urgent cases reach attorney within urgency window.
2.4 Loop 4 - Consult scheduling
Trigger: caller is conflict-clean and qualified. Data path: PMS schedule API to slot held to confirmation SMS to PMS appointment record. Success metric: booking completed within 90 seconds.
2.5 Loop 5 - Post-consult follow-up
Trigger: 24 hours after a scheduled consult that did not result in retention. Data path: outbound call with scheduling link for next steps. Success metric: 18-25% retention recovery rate on initially-undecided callers.
The Loop Map Generator walks an operator through scoping all five for a specific firm in 10-12 minutes.
3. Deep dive: empathy-first intake script
Most legal intake calls arrive at emotionally-charged moments. Personal injury callers are in pain. Family law callers are in conflict. Criminal defence callers are scared. Estate-planning callers are grieving. The intake script must acknowledge the emotional state before transactional questions, or the caller hangs up and calls a more empathetic-sounding firm.
The script opens with disclosure plus acknowledgement: "This is the intake assistant for [firm]. I am sorry you are dealing with this; I am here to help you connect with one of our attorneys. Can you tell me what is going on?" The acknowledgement is scripted, not improvised. The pacing allows for emotional response - the agent waits 2-3 seconds for the caller to begin rather than rushing into the next question.
The intake then runs 5-7 questions in conversational flow rather than interrogation: nature of the issue, when it happened, parties involved, urgency, prior legal involvement, location and jurisdiction. Throughout, the agent uses active listening cues ("I understand", "that sounds difficult", "I have got that") rather than mechanical confirmation. The script explicitly avoids legal advice; any caller question about case strength or strategy gets "I cannot give legal advice on that, but the attorney will absolutely discuss it during your consultation."
Special handlers cover distressed callers. Grief or trauma signals trigger a longer-pause script with explicit emotional acknowledgement. Mental health crisis signals (suicidal ideation, active panic, severe distress) route to crisis resources (988 in US) plus the on-call attorney. Active emergency (in-progress assault, immediate harm) routes to 911 guidance plus immediate human transfer. These handlers are reviewed quarterly with the firm's intake supervisor.
4. Deep dive: real-time conflict check
Conflict checking is the loop where most legal voice deployments fail or fly. Manual conflict checks delay the consultation booking by 24-48 hours, during which time the caller often books with a competitor. Skipped conflict checks create malpractice exposure and ethics violations. The voice agent runs the check during the call.
The agent collects three data points: caller's full name (with spelling confirmation), opposing party's name (and any other names they go by), and any related entities mentioned (employer, business partner, related parties). Each data point posts to the PMS conflict-check API in real-time. Clio's Conflict Check API, PracticePanther's conflict module, or MyCase's conflict search return results in 5-15 seconds depending on database size.
Three outcomes:
- Clean check. No conflict found. Agent proceeds to scheduling. ~80% of calls land here.
- Possible conflict. Name match but ambiguous (common names, prior matters that may have closed). Agent does not proceed; routes to conflicts attorney for human review with 4-hour SLA. Caller gets clear next-step expectation. ~12% of calls.
- Hard conflict. Confirmed conflict (current client on opposing side, prior representation in related matter). Agent explains professionally that the firm cannot represent in this matter and offers referral resources where allowed. ~8% of calls.
Documentation matters. Every conflict check is logged with timestamps, data captured, result, and any human review. This satisfies bar association documentation requirements and provides audit trail for any future ethics inquiry.
5. Statute-of-limitations triage
SOL triage is the highest-stakes loop in legal intake because missing the filing window can permanently bar the client's claim. The agent captures incident date plus jurisdiction during intake, then matches against a per-practice-area SOL matrix.
| Practice area | Typical SOL band (US states) | Trigger event | Watch out for |
|---|---|---|---|
| Personal injury | 2-3 years | Date of injury | Discovery rule extensions |
| Medical malpractice | 1-3 years | Date of treatment or discovery | State-specific notice requirements |
| Contract disputes | 4-6 years (written), 2-4 years (oral) | Date of breach | Tolling agreements |
| Employment (federal) | 180-300 days | Date of adverse action | EEOC right-to-sue letter window |
| Family law | Varies wildly by issue | Multiple triggers | State-specific custody timelines |
| Criminal defence | State-by-state | Date of charge or arrest | Speedy trial deadlines |
The agent calculates days remaining based on captured date plus jurisdiction. Three urgency tiers:
- Critical (under 30 days remaining). Immediate phone transfer to on-call attorney. No callback wait.
- Urgent (30-90 days remaining). Same-day callback from partner-level attorney with explicit SOL flag in the matter notes.
- Standard (over 90 days remaining). Standard consult booking flow with SOL noted in matter file.
The SOL matrix is reviewed quarterly with the firm's intake supervisor and updated when state laws change. The matrix is the single source of truth; the agent does not improvise SOL judgments.
6. Legal practice management system integrations
Five PMS platforms cover 90%+ of mid-market law firms. Integration depth varies; the picking decision typically follows what the firm already runs.
6.1 Clio
Dominant in US and Canadian mid-market. Clio Manage API covers matter creation, conflict checks, scheduling, billing. Voice integration ships in 2-3 days. Deepest integration partner ecosystem.
6.2 MyCase
Strong on small-firm and mid-market with integrated billing plus client portal. REST API supports matter creation, conflict module, scheduling. Integration takes 3-4 days.
6.3 PracticePanther
Mid-market firms emphasising automation. Native conflict-check module plus matter management plus document generation. Integration takes 3-4 days.
6.4 Smokeball
Strongest for document automation and small-firm workflow. API surface narrower than Clio or MyCase; integration takes 5-7 days. Best fit for transactional firms (estate planning, real estate).
6.5 Filevine
Complex litigation and personal injury firms. Powerful matter management, slower API. Integration takes 6-9 days. Best for firms with high case volume.
7. ABA Model Rules and bar association compliance
Three compliance surfaces matter for AI voice agents in legal intake.
ABA Model Rule 1.6 (confidentiality). The agent must clearly disclose that it is an intake screening tool, not legal counsel. The intake script opens with explicit disclosure. Conversations are encrypted at rest and in transit. Recordings retain attorney-work-product protections through proper firm-owned storage. ABA Model Rules of Professional Conduct provide the framework; state bars overlay specific requirements.
ABA Model Rule 7.3 (solicitation). Restricts certain outbound contact patterns. The recall sequence (post-consult follow-up) must respect existing client-relationship rules per state bar. Most states allow follow-up to existing prospects within 30 days; some impose stricter limits. Document the recall policy and review annually.
State bar approval. Most US state bars have either approved AI-assisted intake or issued guidance permitting it when properly disclosed. Check your specific state bar. Some bars (notably Florida and California) have issued advisory opinions; others (most northeastern states) operate on case-by-case enforcement.
Documentation requirements. Every intake conversation should be retained per state bar rules (typically 3-7 years post-matter), encrypted, accessible for audit. The PMS handles most documentation automatically; verify the integration captures conflict-check results and SOL triage decisions in the matter file.
8. The 5-day deploy
Day 1. PMS integration scoped. Clio: developer access issued same day. MyCase or PracticePanther: enrolment in developer program (1-3 day lag in parallel).
Day 2. Practice-area vocabulary trained. Agent learns firm-specific terms (practice areas, common case types, intake-specific terminology). SOL matrix loaded for firm's jurisdictions.
Day 3. Empathy script work. Voice cloning of the intake supervisor (with consent) or selection from a hospitality-trained voice library. Distressed-caller handlers reviewed by managing partner.
Day 4. Conflict-check integration confirmed. Test with synthetic intakes against PMS sandbox. Disclosure scripts verified by compliance partner.
Day 5. Live with first 10 callers on soft launch. Intake supervisor listens in for 2 hours. Adjustments made in real-time. End of Day 5: full traffic on agent.
9. Cost + ROI math at three firm sizes
| Firm size | Intake calls/week | Annual leak recovered | Monthly cost | Payback |
|---|---|---|---|---|
| Solo or 2-attorney | 40-100 | $60-110k | €1,500-2,200 | 2-3 weeks |
| 3-7 attorney mid-market | 120-350 | $185-380k | €2,200-3,500 | 1-2 weeks |
| 8-15 attorney larger firm | 400-700 | $420-720k | €3,500-5,500 | 1-2 weeks |
Recovery scales super-linearly because larger firms have higher per-case value (more complex matters, higher hourly rates, deeper client relationships) and the capture-rate lift compounds across more attorneys. Run the Revenue Leak Heatmap for your firm-specific number.
10. Five things that break legal voice deployments
- Robotic empathy. Empathy must be scripted but not robotic. Generic friendliness destroys credibility on emotionally-charged intake calls.
- Skipping the conflict check. Convenient but creates malpractice exposure. The check must run in real-time during the call.
- Generic SOL handling. Per-practice-area, per-jurisdiction matrix is non-negotiable. Never let the agent improvise SOL judgments.
- Drifting toward legal advice. Even subtle "in cases like yours typically..." language crosses the line. Train the agent to redirect every advice-adjacent question to the consultation.
- Bar compliance gap on disclosure. State bars require specific disclosure language. Document the disclosure script per jurisdiction and audit quarterly.
11. Companion services for law firms
The voice agent closes the inbound surface. Three companion services close the marketing and operational surface:
- Law firm marketing automation. The professional services automation pillar covers matter lifecycle, client communications, billing automation.
- Law firm website on a 7-day sprint. The law firm website generation pillar ships practice-area sites optimised for local search.
- SEO for law firms. The luup SEO pillar covers local pack rankings, practice-area pages, and review velocity.
Sibling voice-agent verticals: dental, medspa, restaurants, home services, automotive.
12. What to ship this week
Pull last week's intake call log. Count voicemails. Count calls placed on hold over 90 seconds. Multiply by your average new-client value (typically $4,200-12,000 in mid-market US legal). That number is your weekly leak. Run the Revenue Leak Heatmap or book a 30-minute review.
13. Frequently asked questions
How does the agent handle empathy?
Scripted, not improvised. Active acknowledgement plus pacing plus explicit no-legal-advice posture. Outperforms generic friendliness 25-35% on caller satisfaction.
How does conflict checking work?
Real-time API call to Clio, PracticePanther, or MyCase during intake. 5-15 second response. Three outcomes: clean, possible (4-hour human review), hard conflict (professional decline).
How does SOL triage work?
Per-practice-area matrix. Critical under 30 days: immediate phone transfer. Urgent 30-90 days: same-day partner callback. Standard over 90 days: regular consult booking.
Which PMS systems integrate cleanly?
Clio cleanest for mid-market US/Canada. MyCase strong on small-firm. PracticePanther mid-market automation. Smokeball for transactional. Filevine for complex litigation.
What about ABA Model Rules?
Two surfaces: Rule 1.6 (confidentiality, requires disclosure of agent role); Rule 7.3 (solicitation, governs recall sequence). Most state bars have approved properly-disclosed AI intake.
Does the agent give legal advice?
Never. Every advice-adjacent question gets redirected to the consultation. Agent collects facts and runs administrative checks only.
How does the agent handle distressed callers?
Specialised handlers for grief, trauma, mental health crisis, active emergency. Crisis routes to 988 plus on-call attorney. Active emergency routes to 911 plus immediate human transfer.
What is realistic ROI for a 3-attorney firm?
$185-280k annually. Capture rate goes 28-32% to 65-78%. Average new-client value $4,200-12,000. Payback 2-4 weeks at €2,200-3,500/month.
14. Field notes from 5 legal intake engagements
Five patterns surface in legal voice deployments. The smaller sample (5 firms vs 12-18 in other verticals) reflects how few firms had production deployments running at the time of audit; the patterns held consistently across the small sample.
Note 1 - the managing partner is the gatekeeper. 4 of 5 firms had the managing partner personally approve every script element. Skipping the partner review cost time later when compliance objections surfaced. Bring the partner in on Day 3 (script review), not Day 5 (post-launch).
Note 2 - the intake script must avoid the appearance of triage by AI. Callers accept AI screening for administrative tasks; they reject AI assessing the merit of their case. Train the agent to capture facts and run conflict/SOL checks but never assess case strength. Any merit-adjacent question routes to the attorney.
Note 3 - referral handling is its own loop. 3 of 5 firms generated 25-40% of intake from referrals (other attorneys, prior clients, professional networks). Referred callers behave differently - higher trust, faster qualification, often pre-qualified by the referrer. Build a referral-aware intake variant that recognises referral source and adjusts conversational tempo.
Note 4 - personal injury is the highest-volume but lowest-margin practice area for AI. PI firms get the highest call volume but the lowest qualification rate (many calls are non-qualifying). The agent must filter aggressively; otherwise attorney consult time gets consumed by non-cases. Estate planning, business law, family law have lower volume but higher qualification rates.
Note 5 - billable-hour culture resists AI deployment. 2 of 5 firms had partner-level resistance because the agent threatened the billable-hour intake culture. The fix: position the agent as a tool that protects partner time for billable work, not as a threat to intake-staff jobs. Frame the value as "more billable hours per partner" rather than "lower intake cost per call."
The fix in every case: bring the managing partner in early, position the agent as fact-capture only, build referral-aware intake, filter aggressively for PI volume, frame value as billable-time protection. Cross-vertical patterns from the 47-deployment audit generalise. Run on your specific firm at luup voice agents for legal or book a review.
Last updated: 4 May 2026.