The phone rings at 7:12 PM on a Tuesday
A prospective client was rear-ended on I-35. They are in pain. They are scared. They found your firm third on a search results page. They call. The voicemail picks up. They hang up and dial the next number.
That call was worth a signed retainer. The next firm answered. They got the case.
AI intake for law firms captures that call, qualifies the matter, books the consultation, and routes the file to the right attorney before the morning docket. It does not replace your paralegal. It covers the hours your paralegal is not there.
What happens after hours is a math problem
Most firms staff intake from 8 AM to 6 PM. That leaves 14 hours a weekday and 48 hours a weekend uncovered. A mid-size personal injury practice in a metro area might field ten inbound calls during those dark hours each week. If your average signed case brings $15,000 in fees and you close one in four qualified leads, ten missed calls represent $37,500 in monthly revenue that walked to a competitor who answered.
You can check this arithmetic against your own call logs and fee schedule. The numbers do not require a survey. They require a spreadsheet.
Why law firms lose the after-hours caller
The caller does not want a callback. They want counsel. They are in crisis. They will not wait for a return call at 9 AM. They will call the next listing. The firms that capture these callers share three traits: they answer live, they ask the right screening questions, and they lock in the appointment before the caller hangs up.
A human answering service can answer live. It cannot reliably screen for conflict, preserve privilege, or book directly into your calendar without a paralegal relay. A voicemail box does none of the three.
AI intake for law firms does all three inside the same conversation. It verifies identity, runs a conflict check against your matter management system, explains the scope of representation, and writes the consult onto the attorney’s calendar — all before the caller hangs up.
How the system works in practice
The flow is narrow and deliberate. No chatbot personality. No open-ended conversation.
- Call arrives after hours. The AI answers on the first ring with the firm’s name and a confidentiality notice.
- Caller states the issue. The AI extracts practice area, jurisdiction, and adverse party names using a constrained language model tuned to legal taxonomy.
- Conflict check runs against your live matter database. If a conflict exists, the AI discloses it and terminates the intake without storing privileged detail.
- Qualification questions fire: date of incident, statute of limitations window, insurance status, prior representation. Answers are logged in structured fields, not free text.
- Fee structure and retainer terms are read verbatim from your approved script. Caller confirms understanding.
- Calendar integration offers the next three available slots for the relevant attorney. Caller picks one. The event writes to Outlook or Google Calendar with a secure link to the intake packet.
- Call recording and transcript land in the matter folder with access restricted to the assigned attorney and intake supervisor.
No data leaves your tenant. The model runs in a private VPC. The transcript is encrypted at rest. The system does not train on your calls. It does not share patterns across clients. This is not a consumer chatbot. It is a bounded automation layer.
What it cannot do: it cannot give legal advice. It cannot negotiate fees. It cannot make strategic judgments about case viability. It flags the file for human review with a priority score. The attorney still decides.
Measuring the hit rate
Treat after-hours intake like a marketing channel. Track three numbers each month: calls answered, consultations booked, retainers signed. Divide signed retainers by calls answered. That is your hit rate.
If your daytime hit rate is 22 percent and your AI after-hours hit rate is 18 percent, the system is paying for itself. If it falls below 10 percent, adjust the qualification script or the calendar availability. The metric is observable. The fix is mechanical.
One firm in our current program ran this arithmetic for 90 days. Daytime intake: 140 calls, 31 consults, 7 retains. After-hours AI: 68 calls, 14 consults, 3 retains. The after-hours channel added 43 percent more signed cases without adding headcount. The cost was a fixed monthly fee plus per-minute telephony. The margin was clear.
Where to start without disrupting the docket
Do not rip out your daytime process. Run a shadow intake for 30 days. Forward after-hours calls to the AI. Keep your paralegal on the morning review. Compare the AI’s qualification tags against your paralegal’s tags. Measure disagreement rate. If it stays under 5 percent, promote the AI to primary after-hours handler.
The integration surface is small: a SIP trunk for voice, a read-only API key for your matter management system, a service account for calendar write access. No desktop agents. No browser extensions. Your IT admin can provision it in an afternoon.
If you want a guided setup with a fixed timeline, our AI consulting for your operation maps the workflow, configures the conflict logic, and runs the shadow period with your team. You pay for the build. You own the result.
For firms that prefer a self-directed pilot, Zephyr, our AI assistant layer includes a pre-built legal intake skill pack. You configure the screening questions, upload your conflict database, and point the calendar. The skill pack handles the rest.
The revenue leak stops when the phone is answered
After-hours callers do not wait. They do not leave detailed messages. They do not call back. They call the next firm. Every night you rely on voicemail, you underwrite your competitor’s marketing budget.
AI intake for law firms is not a transformation initiative. It is a coverage layer. It costs less than a part-time paralegal. It works the hours no human wants. It hands you qualified, conflict-cleared, calendar-locked consults at 7 AM.
The goal is not replacing staff. The goal is capturing the revenue that currently evaporates at 6:01 PM. The arithmetic is yours to verify. The fix is available now.