Should our firm put an AI chatbot on its website?

You can, and plenty of firms are being sold one. The sensible position is that a website chatbot is an intake and routing tool, not an advice tool: it should collect enough to work out who the enquirer is and what they want, then hand over to a person. The risk is not the technology; it is what a public-facing tool says to a stranger about their legal problem, and what it quietly records about them while doing it.

Be clear what you want it to do

Most firms who ask this question want one of three different things, and the answer differs for each.

• Capture. Take a name, contact details and a one-line description outside office hours, and email it to the right inbox. This is a form with a friendlier manner. Low risk, modest benefit.
• Triage. Ask a few follow-up questions, work out which department the enquiry belongs to, flag the obvious non-starters (wrong jurisdiction, outside your practice areas, already represented) and tell the enquirer what happens next. More useful, more to get wrong.
• Answer. Respond to the substance of the question: whether they have a claim, how long they have, what the fee is likely to be. This is where firms get into difficulty.

Write down which of the three you are buying before you look at any product. A supplier demonstrating the third while you thought you were buying the first is a common and expensive misunderstanding.

The risks worth weighing

It may appear to give advice. A member of the public does not draw a neat line between information and advice. If a tool on your website, carrying your name and your branding, tells someone their deadline has probably passed or that their case sounds strong, they may act on it. Your regulatory obligations to act in a client's best interests and to communicate in a way they can understand do not switch off because the words came from software. Assume anything the chatbot says will be read as coming from the firm.

It may get the jurisdiction wrong. A general purpose tool will happily answer an English law question for someone writing from Glasgow or Belfast, or an employment question with American assumptions. For a public-facing tool you cannot see the misunderstanding happening.

It creates a conflicts and confidentiality problem. Someone may type the other side's name, the detail of a dispute, or information they would regard as privileged into a box on your website at eleven at night. You now hold it. Think through where that sits, who can see it, how it is screened against your conflicts list, and what you do if it turns out you act for the opponent.

It creates a data protection problem. Enquiry data is personal data, often special category data in family, employment, immigration and personal injury work. You need a lawful basis, a retention period, a privacy notice that actually describes the tool, and clarity on where the data goes and who processes it. UK GDPR expectations apply to the chatbot exactly as they apply to your case management system. If the tool is doing anything substantial with people's information, consider whether a data protection impact assessment is warranted.

It creates a complaints trail. If an enquirer later says they were told their claim was worth pursuing, or were turned away in error, you will want a complete log of the conversation. Make sure you have one, and that it is retained as long as your complaints policy requires.

Questions to ask the supplier

• Where is the conversation data stored, who processes it, and is any of it used to improve a product used by other customers?
• Can we see a full transcript of every conversation, and export it?
• What does the tool do when it does not know? Does it say so, or does it produce something plausible?
• Can we constrain it to our own content, and can we set hard limits on topics it must refuse?
• How is it tested before release, and how do we test changes before they go live on our site?
• Who is liable if it tells an enquirer something wrong?

Ask your insurance broker before you deploy anything in the third category. A public-facing tool that comments on the merits of a stranger's problem is a different risk from an internal assistant used by a supervised fee earner.

A workable middle ground

Most firms of your size get the benefit without the exposure by keeping the chatbot firmly on the firm's own territory: opening hours, locations, practice areas, how to make an appointment, what documents to bring, how fees generally work, how to complain. Then a structured handover to a human, with a clear line on screen saying the tool does not give legal advice and no solicitor-client relationship arises from the conversation.

Review the transcripts weekly for the first couple of months. You will learn quickly where it is overreaching, and you will also learn what the public actually asks you, which is useful on its own.

Where Alesis fits

Alesis sits on the other side of that line. It is an AI assistant for UK law firms, made by L25 Limited and used through the web browser, with one conversation for a matter: it is for fee earners working on files, not for members of the public. It prepares drafts for a qualified person to review and sign off, and it does not file, serve or email anything. It assists qualified professionals and does not replace them, and it does not provide legal advice.