AI Client Intake for Law Firms: Chatbots, Triage, and the Ethics Line

The most common AI use case that actually makes law firms money is not research or contract review — it is intake. A lead that gets an immediate, substantive response converts dramatically better than one that waits until morning. An intake process that collects the right information the first time compresses the first call from forty minutes to ten. For a solo or small firm, where every hour of the attorney's time is at a premium, this is where AI earns back its cost quickly and visibly.

But intake is also where the legal ethics obligations come into sharpest focus. An intake chatbot is a public-facing communication on behalf of a licensed attorney — which means the advertising rules, the unauthorized-practice line, the confidentiality obligations, and the conflict-checking duty all apply before the first message is sent.

This guide covers both sides: how to build a genuinely useful AI intake workflow, and where the line is.

What AI intake actually does

24/7 lead capture. A website chatbot can respond to prospective clients at any hour, ask the intake questions the firm actually needs answered, and collect contact information and matter detail — without requiring the attorney or staff to be available. The lead is captured; the attorney reviews and follows up during business hours.

Pre-qualification and triage. A well-configured intake flow can route leads by practice area, identify obvious non-fits (geographic limitations, matters outside the firm's scope, legal emergencies requiring immediate referral), and flag urgent matters for same-day response. This is the triage function: separate the matters worth a consultation from the ones that need a different resource.

Information collection before the first call. Gathering the basic matter facts before a consultation — parties, dates, documents in hand, what outcome the client is seeking — compresses the intake call and lets the attorney spend time on analysis rather than background questions. A structured intake form or chatbot conversation accomplishes this whether or not the client calls during business hours.

Routine follow-up drafting. After intake, AI can draft the initial follow-up email, the consultation confirmation, and the engagement checklist — reducing the administrative overhead of converting a lead to a matter.

What it must never do

This is the part that every vendor's demo glosses over. A law firm intake chatbot is not a customer service bot. It is a communication on behalf of a licensed attorney, governed by the advertising rules of every jurisdiction in which it appears, and subject to the unauthorized-practice-of-law prohibition.

Never give legal advice. An intake chatbot can explain what the firm does, collect information, and schedule a consultation. It cannot tell a prospective client whether they have a claim, what the law is, what they should do, or what a likely outcome is. The line between "what information does your firm handle?" and "do you think I have a case?" is the line between intake and advice — and the chatbot must stay on the intake side.

Never create an attorney-client relationship without disclosure. In many jurisdictions, an implied attorney-client relationship can form when a prospective client submits confidential information in the reasonable belief that they are seeking legal advice. Your intake flow should make clear — before information is collected — that it does not constitute a consultation and does not create an attorney-client relationship. This is a disclosure requirement, not boilerplate.

Never skip the conflict check. Collecting intake information before running a conflicts check means the firm may already hold information about a person it later identifies as a conflict. For small firms, the practical answer is: run a name check before the intake bot proceeds past the preliminary question stage, or disclose clearly that conflict checking will occur before the engagement is accepted and nothing submitted creates an obligation.

Never state the law incorrectly. A chatbot configured to answer jurisdiction-specific questions about filing deadlines, damages caps, or procedural requirements and getting them wrong exposes the firm to advertising-rule violations and potentially worse. If the bot must address jurisdictional questions, it should answer in general terms and disclaim specifics: "Statutes of limitations vary by claim type and jurisdiction — your consultation will cover how they apply to your situation."

Building the workflow

Step 1: Define what the intake flow collects

Before configuring any tool, write out the exact questions you need answered before you can decide whether to schedule a consultation. These typically include:

• What type of matter (to route to the right practice area)
• Jurisdiction / location (to check geographic fit)
• A brief factual description (to run an initial conflict check)
• Contact information
• Timeline / urgency (is there an immediate deadline?)
• How they heard about the firm (marketing attribution)

This is the information that informs a yes/no/refer decision. Do not collect more than this at the pre-consultation stage — more information creates more confidentiality exposure before any engagement has been accepted.

Step 2: Set the scope of the bot explicitly

Write the system prompt or configuration instructions to define what the bot is and is not:

• It introduces the firm, explains the intake purpose, and collects the defined information
• It does not answer legal questions, provide legal advice, or opine on the merits of a matter
• When asked a substantive legal question, it defers: "That's exactly the kind of question we'd cover in a consultation — let's get your basic information and set that up"
• When a matter is outside the firm's scope, it refers with a clear explanation: "Based on what you've described, this appears to be a matter we don't handle — [referral guidance]"

Step 3: Choose the tool

For most solo and small firms, the simplest path is a specialized intake platform rather than a general AI chatbot. Platforms like Lawmatics, Clio Grow (formerly Lexicata), and Filevine's intake tools are designed for law firm intake workflows, include the disclosure language and data handling requirements, and integrate with the practice management system where matters eventually live.

A general AI chatbot (embedded widget with GPT-4 or Claude) offers more flexibility but requires more configuration work to get the scope right — and more discipline to prevent the model from answering questions it should not. If you use a general AI widget, the system prompt is the guardrail; invest the time to write it carefully.

Step 4: Test it as the client would

Before going live, run through the intake flow yourself as a prospective client — including the edge cases. What happens if someone describes a criminal matter and the firm is civil-only? What happens if someone asks directly "do I have a case"? What happens if the matter involves a potential conflict with an existing client? What happens if the request is urgent?

Test until every path either collects the right information, defers appropriately, or routes to a resource that can help. The paths that produce a dead end, a wrong answer, or an implied legal opinion are the ones that create exposure.

Step 5: Wire the backend

Intake data that goes into a form and disappears is not a workflow — it is a data-collection failure. Every intake submission should:

• Create a record in the practice management system (Clio, MyCase, etc.)
• Trigger an automated follow-up acknowledging receipt with a timeline for response
• Flag for the attorney or intake coordinator within a defined window (same day for urgent matters, next business day for standard)

The faster the follow-up, the higher the conversion. Studies on lead response time consistently show that response within the first hour — ideally the first minute — produces dramatically higher contact rates than waiting until the next business day.

The confidentiality dimension

Intake information — a prospective client's name, the nature of their matter, any details they share — is confidential under Model Rule 1.6 even if no engagement is ever formed. The intake platform or chatbot tool must handle this information under terms that protect it accordingly: no training on inputs, data processing agreement, secure storage.

For a chatbot embedded on a law firm website, the data handling terms of the underlying AI tool are as important as the configuration. A general-purpose AI widget that trains on the conversations it has is a confidentiality problem before it answers a single question.

The verdict

AI intake delivers one of the clearest ROIs in the legal AI toolkit because the value is direct: more leads captured outside business hours, faster follow-up, better pre-consultation information, less attorney time on administrative triage. The configuration work to do it ethically — scope the bot, write the disclosures, test the edge cases — is several hours of setup. That investment pays back quickly on the first after-hours lead that converts to a matter because someone responded immediately.

For the full framework of AI tools in a small practice, start with the best AI tools for lawyers. For the advertising and UPL rules that govern client-facing communications, see bar advertising rules for attorney websites.

Related reading: The best AI tools for lawyers | Bar advertising rules for attorney websites | AI for law firms