Legal Chatbots: What a Law Firm Chatbot Should Do, and What It Should Never Do

Most people who land on an injury firm's website at night or at the weekend will not fill in a long form, and many will not call. A legal chatbot gives them a third option: a short conversation in the corner of the screen that takes their details and gets them to a person. Done well, it keeps enquiries that would otherwise leave. Done badly, it gives legal advice it has no business giving, frustrates the people it should help and creates ethics problems for the firm.

This guide covers what a law firm chatbot is for, where it must stop, the rules that apply, and how to tell whether it is working. For the wider picture of artificial intelligence in a practice, see AI for law firms.

What a legal chatbot actually is

The term covers two quite different tools:

• Scripted chatbots follow a fixed set of questions and answers. They are predictable and easy to check, but they cannot cope with anything off the script.
• AI chatbots use a language model to understand free text and reply in natural language. They handle real conversations far better, but they can also say things nobody wrote or approved, which is the main risk for a law firm.

Many firms use a mix: a conversational front end that keeps to a defined set of intake questions. Some firms also use live chat staffed by people, sometimes with a chatbot covering the hours nobody is available.

What a chatbot should do

• Capture contact details early. Name and phone number first, so the enquiry is not lost if the person leaves halfway through.
• Ask a few basic intake questions. What happened, roughly when, where, and whether anyone was hurt. Enough to route the enquiry, not to assess the case.
• Answer simple practical questions. Office hours, locations, which practice areas the firm handles and how fees generally work, all from approved wording.
• Hand over to a person fast. Offer a call back, book a time or pass the details straight to intake. The chatbot's job ends when a human takes over.
• Work around the clock. Its real value is the hours when nobody is at the desk.

What a chatbot should never do

• Give legal advice. It should not tell someone whether they have a case, what it is worth, or what to do about a deadline. Those are judgements for a lawyer.
• Pretend to be a person. Make it clear the visitor is talking to an automated assistant.
• Promise an outcome or a fee. A chatbot that says "you will get compensation" is making a statement the firm cannot stand behind.
• Collect more than it needs. Medical detail and the full story belong in a conversation with intake, not in a chat log.
• Leave someone stuck. If a question falls outside what it can answer, it should say so and offer a person.

The ethics rules that apply

A chatbot speaks for the firm, so the usual professional conduct rules reach it. Under the ABA Model Rules, which most states have adopted in some form:

• Rule 7.1 prohibits false or misleading communications about the firm's services. That covers what a chatbot says about results, fees and experience.
• Rule 5.3 makes lawyers responsible for supervising nonlawyer assistance, which is how many regulators view automated tools.
• Rule 1.6 requires reasonable efforts to protect client information, which applies to chat logs and to any outside provider that stores them.

The ABA's Formal Opinion 512, issued in July 2024, addresses lawyers' use of generative AI tools and covers competence, confidentiality, supervision and fees. State bars have issued their own guidance too, and advertising rules vary by state, so check the rules where the firm practises before launch. The guide to lawyer advertising covers the advertising rules in more depth. A clear disclaimer that the chat does not create an attorney-client relationship and is not legal advice is a sensible baseline.

Where it fits in intake

A chatbot is a front door, not an intake process. What happens after the chat matters more than the chat itself. Every conversation should land in the same place as calls and forms, with the details attached, so intake can follow up quickly and nothing sits in a separate inbox. The guide to legal intake software and the guide to law firm CRMs cover the systems behind it, and law firm automation covers the follow-up rules that make sure an enquiry is never left waiting.

How to set one up well

• Write the script first. Decide the questions, the approved answers and the point at which it hands over, before choosing a tool.
• Keep it short. Every extra question loses people. Ask only what intake needs to call back.
• Put it where it helps. Practice area pages and the contact page, without covering content on a phone screen. The guide to law firm website design covers mobile layout.
• Test it as a visitor. Try awkward questions, a request for advice and a bad connection, and see what it says.
• Review transcripts regularly. They show what people actually ask and where the chatbot falls short.

How to judge whether it works

The number of chats is not the measure. Track how many conversations produced contact details, how quickly intake reached each person, and, most importantly, how many became signed cases. Compare that with calls and forms from the same pages. If chats rarely turn into conversations with intake, the problem is usually the handoff, not the chatbot. The guide to law firm KPIs covers which numbers to review.

The short version

A legal chatbot is useful for one job: catching enquiries when nobody is available and getting them to a person fast. It should capture details, ask a few basic questions and hand over. It should never give legal advice, promise outcomes or pretend to be human, and the professional conduct rules apply to everything it says. Judge it by signed cases, not by the number of chats.