Law Firm Intake Automation Without Giving Legal Advice

Legal enquiries arrive at inconvenient times and in poor shape. Someone describes a situation at length without mentioning the jurisdiction, the other party, or the deadline that determines whether the matter is viable at all.

A fee earner then spends twenty minutes on a call establishing facts that could have been captured in a structured exchange, and a proportion of those calls end in a matter the firm was never going to take.

Intake automation is not about replacing that judgement. It is about ensuring the judgement happens with the facts already gathered, and about capturing enquiries that would otherwise be lost to whoever answered the phone first.

This distinction carries regulatory weight in every jurisdiction we are aware of, and it is worth stating in concrete terms rather than in principle.

Intake is gathering: what kind of matter, where, who is involved, when did it happen, is there a deadline, how should we reach you. Advice is applying law to those facts — whether they have a case, what they should do, what it is worth, what the likely outcome is.

The failure mode is subtle. A chatbot answering a general question about limitation periods sounds informational, but a prospective client reading it as applying to their situation has received advice in every way that matters if it turns out to be wrong.

The first message needs to establish three things, in plain language rather than in legalese, because a prospective client who does not understand them may disclose things that create problems for the firm.

First, that this is an automated assistant and not a lawyer. Second, that using it does not create a solicitor-client or attorney-client relationship. Third, that they should not share confidential details of their matter until the firm has confirmed it can act.

That third point is not a formality. Unsolicited confidential information from someone whose opponent you already represent is a genuine problem, and it arrives more often than firms expect.

This is the structural difference between legal intake and every other sector, and it determines the sequence of questions.

You need enough information to run a conflicts check — the names of the parties involved — before the enquirer describes their matter in detail. That is the opposite of a natural conversation, where people explain their situation first and the details emerge later.

Some legal enquiries are genuinely time-critical — an arrest, a court date, an injunction, an immigration detention, an expiring limitation period. A queued enquiry is a failed enquiry in those cases.

Configure detection for urgent circumstances and give those a distinct route: immediate escalation, an out-of-hours contact number displayed directly, and clear language that the person should not wait for a callback if the matter is urgent.

Importantly, the bot should not assess how urgent something legally is. It should recognise the language of urgency and escalate, leaving the assessment to a person.

A good intake record makes the first consultation shorter and better. An over-eager one creates a data protection liability and a file full of information the firm did not need and now has to protect.

Firms that do this well report the same two effects. Consultations start further along, because the fee earner already knows the matter type, jurisdiction and deadline. And fewer enquiries are lost overnight, because someone got a considered response at eleven at night rather than a voicemail prompt.

Neither is dramatic. Both compound, and neither requires the chatbot to say a single thing about the law.