An AI intake agent for law firms
that never gives legal advice, only routes to one

Where a firm’s hours disappear

A firm’s front desk, or the partner who answers the firm’s own phone, spends real hours on questions that never needed a lawyer. What practice areas the firm covers. Roughly what a matter costs. What documents to bring to a first meeting. Intake and scheduling eat time that could go to billable work, industry surveys of small and mid-size firms consistently find. A slow first response is one of the more common reasons a prospective client calls a competitor instead.

Inconsistency is the next problem. One associate quotes a rough fee range on the phone. Another refuses to discuss cost until a meeting is booked. A prospective client who gets three different answers from three different people loses confidence before the firm has done any actual work.

There is also real risk in saying too much. A well-meaning staff member answering a general question can drift into something that sounds like advice on a specific situation. That is exactly the exposure a firm wants to avoid. Without a clear line, the easiest fix firms reach for is answering nothing quickly, which loses leads instead.

After-hours volume is a quieter problem. Firms with international or expat clients get questions outside normal office hours, in a different time zone. A prospective client who hears nothing back for a day often assumes the firm does not handle their kind of matter, when the real issue was simply timing.

The agent we build for law firms

An intake and FAQ agent that answers from your firm’s own material: practice area descriptions, typical fee ranges, process and timelines, nothing else. It collects the documents a matter typically needs before a first meeting, using a checklist specific to the matter type the client describes. The first consultation starts with paperwork already gathered, instead of a lawyer asking for it live.

The agent has one hard rule in its playbook: it never gives legal advice on a client’s actual situation. The moment a question moves from “what do you handle” to “what should I do about my case,” it says so plainly. It routes the conversation to a lawyer with the full exchange attached, so nothing needs repeating. Booking a consultation checks your calendar directly and sends a confirmation and reminder.

Typical integrations: your site chat widget, WhatsApp and Telegram for clients who prefer a messenger, and Google Calendar or your booking tool for consultations. For intake records, that is Clio, HubSpot, amoCRM or a Google Sheets log, with role-based access and a stated retention period.

For firms with several practice areas or several partners, the agent can route by matter type to the attorney who actually handles it. A family-law question goes to the family-law partner’s calendar. A corporate question goes to the corporate team. The same qualifying answer the client already gave decides it.

Two to three weeks, step by step

1. Audit of the intake flow. We read a sample of real client questions and your current booking process. We map which questions a non-lawyer can safely answer today.
2. Playbook. FAQ content, document checklists by matter type, and the exact line the agent will not cross, written with a partner and reviewed before any code.
3. Build and test. The agent is built against your real fee structure and practice descriptions, tested on recorded real questions first.
4. Launch with guards. A clear escalation rule for anything case-specific, full conversation logging, confidentiality-aware access.
5. Tuning. Two weeks of reviewing real conversations with your intake team and adjusting the FAQ content.

What it costs

Running cost is separate and usually $20 to $200 a month in model usage depending on volume, with a hard budget cap set before launch.

What usually moves after launch

First-response time is among the strongest predictors of whether a prospective client books a consultation at all, industry benchmarks for professional services show. Firms that move routine intake off a human’s plate typically report real gains. Often 40 to 60 percent of inbound questions get resolved without taking a lawyer’s time.

These are typical ranges reported across the vertical, not a promise for every firm. Practice area, market and how clearly the firm’s own material is written all move the number. Our own numbers sit in the case studies below. A visa consulting centre’s client-facing AI consultant with a strict hand-off rule to a human is detailed in the visa centre AI support case study. A B2B playbook-driven inbox agent with objection handling is in the certification body AI sales agent case study.

Why work with us

We build these agents the way we build them for ourselves. A closed set of tools. A hard rule the agent cannot cross, here, no legal advice, enforced in the playbook rather than hoped for. The price is fixed once we agree the plan. You get a working demo every week during the build, and the conversation logs and source code stay in your firm’s name.

We also tell a firm when a simpler fix would do. Sometimes the real problem is a fee page nobody has updated in two years. No amount of agent polish fixes a question the firm’s own website should already answer. We would rather scope a smaller project that actually closes the gap than sell an agent on top of a site that works against it.

Pair this agent with a site built around booking a consultation: the website for law firms page. See the full package breakdown on the AI agents service page, or get a written plan with a fixed price for your firm.