Law firm intake automation: what to automate, what not

Law firm intake automation works best on the mechanical steps: answering, capturing details, confirming receipt, routing to the right team, reminding people of deadlines and sending standard documents. It should not make the judgements: whether to act, whether there is a conflict, what the work will cost, or anything a caller could take as legal advice. Those stay with people.

That line sounds simple, and most firms would agree with it in a meeting. The difficulty is that real automations blur it. An auto-reply that says "we can help with that" has made a decision. A chatbot that answers "how long does a divorce take?" is close to advice. This post goes through intake step by step, sorts each part onto one side of the line, and sets out how to check that automations keep doing what you think. For the full sequence, see our guide to the law firm client intake process.

What is law firm intake automation?

Using software to carry out steps in intake that would otherwise need a person: answering or acknowledging an enquiry, asking set questions, recording the answers, sending the enquiry to the right place, chasing when something has not happened, and generating standard documents.

It ranges from very simple (an automatic email reply to a web form) to more involved (an AI system answering the phone, holding a conversation and recording a summary). The principle is the same at every level: automation should make the firm faster and more consistent at the steps where speed and consistency are the whole job, and should hand over cleanly to a person at every step where judgement is.

Which intake steps can be automated?

Here is our sorting of the intake steps. "Automate" means it can run without a person in the loop each time, subject to design and checking. "Assist" means software can prepare it, and a person decides. "Keep human" means a person does it.

• Answering the phone and acknowledging written enquiries: automate. Speed is the whole point here. In the Clio Legal Trends Report 2024 secret-shopper study, only 40 per cent of firms answered a prospective client's call and 33 per cent replied to an email.
• Asking the firm's first-contact questions and recording answers: automate, using the firm's own questions and wording.
• Telling the caller what happens next: automate, within wording the firm has approved.
• Booking a call back: automate, into fee-earners' real availability.
• Routing to the right team: automate, based on rules the firm sets, with a human owner for anything that does not fit.
• Conflict search: assist. Software can run the name search. Deciding whether a match is a conflict is a judgement.
• Deciding whether to take the matter: keep human.
• Identity checks: assist. Tools can collect and verify documents. What level of checks a matter needs is set by the firm's AML policy.
• Cost estimate: keep human. Software can hold published prices; the estimate for a specific matter is the fee-earner's.
• Client care letter and terms: assist. Templates and document assembly save time; the matter-specific content comes from the fee-earner.
• Reminders and chasers: automate, for internal deadlines and for terms not returned.
• Reporting: automate.

The pattern is that everything before the fee-earner's first call, and everything that is a reminder, is a good candidate. Everything in the middle, where the firm decides whether and how to act, is not.

What should never be automated in legal intake?

Four things, and they should be written into any specification or contract:

• Legal advice or anything resembling it. An automated system should not tell a caller their prospects, their options, or what they should do. It can say that a solicitor will discuss those things.
• The decision to act. Including the decision that a conflict exists or does not.
• Price commitments beyond what the firm has published. A system can point to published prices; it should not quote a figure for a specific matter unless the firm has specifically set that up and approved it.
• Pretending to be human. If the firm uses an AI system on the phone or in chat, it should say it is AI. Our post on whether a UK law firm can let AI answer client calls covers the questions to take to your COLP on this.

How do you automate intake without losing the personal touch?

By automating the parts clients do not experience as personal anyway, and putting the saved time into the parts they do.

Nobody values waiting on hold or leaving a voicemail. Most people are glad to get a prompt answer and a confirmed call back time. What clients do experience as personal is the first real conversation with the solicitor: whether that person already knows why they called, and whether they listen. Good automation makes that conversation better, because the fee-earner starts it with a summary rather than a blank page.

The failure mode is automation that tries to be the relationship: a sequence of chaser emails with fake first names, or an assistant that pretends to be a person. That costs trust, and in legal services trust is the product.

What are the risks of intake automation?

• Silent failure. An automation stops working and nobody notices, because nothing visible breaks. A web form's notifications go to a deleted mailbox. A routing rule sends every enquiry to one person on leave.
• Drift into advice. An auto-reply or chatbot answer is written helpfully and says more than it should.
• Wrong routing. Rules written for last year's team structure send work to the wrong person.
• Data protection. Every automated step that collects personal data needs the right privacy information at the point of collection, which the ICO's guidance on the right to be informed describes, and every tool involved is a supplier your data protection lead needs to know about.
• Over-collection. It is easy to add questions to an automated form. Each one costs completion rate and adds data the firm has to protect.

How do you check intake automations are still working?

Do not rely on the automation's own configuration screen. It will show what is set up, not what is happening. Check outcomes instead.

A weekly routine that takes about fifteen minutes:

• Test each channel. Ring the main number out of hours. Submit the web form. Send an email to the enquiries address. Check each one arrived, was answered or acknowledged, and was routed correctly.
• Count the enquiries with no owner and no outcome.
• Read five automated summaries or replies end to end. Would you be happy if a client read them to a regulator?
• Check routing rules against the current team list whenever someone joins, leaves or moves.

A worked example: automating one firm's intake

Here is a hypothetical firm, with invented figures, to show where the time goes.

A six-fee-earner firm receives around 120 enquiries a month. Before automating, reception spends roughly ten minutes on each: answering, taking details, emailing the team, chasing a call back. That is around 20 hours a month. Out-of-hours calls go to voicemail, and the firm does not know how many of those callers never leave a message.

The firm automates four things: the phone out of hours, web form acknowledgements, routing, and a chaser for terms not returned. Reception now spends its time on the enquiries that need a person, such as the distressed caller and the existing client with a new problem, and on making sure every enquiry has an outcome. Fee-earners start their first calls with a written summary. Nothing about the decision to act, the conflict check or the price has changed hands.

The point of the example is not the hours saved. It is that the automation is placed on the steps where the firm was slowest and least consistent, and nowhere else.

Where do you start with intake automation?

With the step your data says you lose the most at. For most firms that is first contact: calls that ring out, and forms that sit unanswered. Map the process first (our client intake process flow chart gives a twelve-step version with six exits), count what falls out at each exit for a month, and automate the biggest leak first.

Then look at tools. Our guide to client intake software for law firms covers what to ask vendors, and if any tool will talk to the public for you, work through our AI client intake governance checklist first. For automation beyond intake, see law firm workflow automation.

Who signs off an intake automation?

Somebody with authority over the process and somebody with responsibility for compliance, before it goes live and after every material change. In most firms that means the intake owner and the COLP, with the data protection lead involved for anything that collects personal data.

The sign-off should cover the exact wording the automation uses with the public, the rules that decide where an enquiry goes, what happens when the automation cannot handle something, and who is told when it fails. Keep a copy of what was approved and when. If the wording changes, the approval should be refreshed; an automated reply that has been edited five times since it was signed off is, in practice, unapproved.

Does automation replace intake staff?

Usually it changes the job rather than removing it. The mechanical half moves to software; the judgement half (the complicated caller, the possible conflict, the enquiry that fits no rule) stays, and gets more attention. Our post on what a legal intake specialist does looks at how that role is changing.

Where WasilX fits

Cam, our AI Engagement Coordinator, automates the first-contact steps on the firm's phone line. It takes the call at any hour, says it is AI, asks the firm's own qualifying questions, books the call back and sends the firm a summary. It does not give legal advice, run conflict checks, quote for a matter or decide whether the firm acts. If you would like to see where the line falls in your own intake, start a conversation.

Frequently asked questions

What parts of law firm intake can be automated? The mechanical steps: answering and acknowledging enquiries, asking set questions, recording answers, booking call backs, routing, reminders and reporting. Conflict searches, identity checks and document assembly can be assisted by software, with a person deciding.

What should never be automated in legal intake? Legal advice or anything resembling it, the decision to act (including on conflicts), price commitments beyond what the firm has published, and any pretence that an automated system is a person.

Is it allowed to use AI to answer a law firm's phone? That is a question for your COLP and your own advisers. Our view is that any AI on the phone should say it is AI, should not give advice, and should hand over to the firm with a clear record.

How do you know if intake automation is working? Check outcomes, not configuration. Test each channel weekly, count enquiries with no owner, read a sample of automated messages, and check routing rules whenever the team changes.

Will intake automation replace reception staff? It usually changes the job. Software takes the repetitive first-contact work; people handle the calls that need judgement and make sure every enquiry has an outcome.