AI automation for law firms

Fee earners spend a surprising share of the week on work that is not legal work: chasing documents, updating files, telling clients where things stand. This is about removing that layer, without letting software anywhere near the legal judgement.

By Rashid, founder of Layerform System Ltd, based in London.

Where the time actually goes

In most firms I have looked at, the bottleneck is not the legal work. It is everything wrapped around it:

• Chasing clients for documents, then chasing again, and keeping track of who has sent what.
• Filing what arrives. Documents come in by email, post and portal, and someone has to read, rename and put each one in the right place.
• Checking a file is complete before it can move to the next stage, by opening it and looking.
• Telling clients where things are. The same update, written again, because the client cannot see it themselves.
• Keeping two systems in step, because the case management system and the accounts package do not talk to each other.
• Watching deadlines across a caseload, held in someone's head and a diary.

None of that needs a qualified lawyer. All of it currently uses one.

What can genuinely be automated

This is the honest version, with the AI kept to where it earns its place.

• Reading and filing incoming documents. Working out what a document is, which matter it belongs to, and filing it named correctly. This is the strongest use of AI in a firm, because documents arrive in endlessly varying formats.
• Spotting what is missing. Comparing a file against what that matter type requires, and listing the gaps, so nobody opens a file to find out.
• Chasing automatically. The client gets a polite reminder about the specific document outstanding, without a fee earner remembering to send it.
• Drafting the routine update. Drafted, not sent. The fee earner reads it and sends in seconds.
• Deadline tracking across the caseload. Rules, not AI, because this needs to be exactly right every time.
• One view of a matter. Everything about a case in one place instead of across a system, an inbox and a folder.

Where a human stays in the loop

Some of this is professional obligation and some is simple good sense, but the line is the same either way.

• Anything that leaves the firm is approved by a person.
• Legal judgement, advice and strategy stay entirely with the fee earner.
• Anything with a limitation date or court deadline is built on fixed rules and double-checked, never left to a model's interpretation.
• Every automated action is logged, so there is a record of what happened and when.

Compliance remains your firm's responsibility and I am not a compliance adviser. What I will do is design the system so your COLP can actually review it, and explain in plain terms where client data goes and who can see it.

Do you need a legal automation platform, or something built around your firm?

A lot of firms start this search looking for a legal automation platform: a product with a monthly fee, a login for everyone and a list of features. I should say plainly that I do not sell one. What I build is a system for one firm, shaped around how that firm already works. So here is how I would think about which of the two you actually need.

Buy a platform if your firm works the way most firms do. If your firm does one type of work in much the same way as every other firm doing it, and a product already handles the parts that eat your time, buy the product. It is quicker to get running, the cost is predictable, and someone else pays for the maintenance. Most practice management and case management products sit in this group, and they are good at what they do.

Have something built if the way you work is the point, or if a product would force you to change how you work to fit it. The usual sign is a firm that has already bought a platform and still runs most of its process on spreadsheets and email around it, because the product does not do the one thing they needed. Or the awkward part of the work is specific to the practice area, the client base or the way the team is set up, and no feature list covers it.

Most firms land between the two. Keep the case management system you have. Build the missing pieces around it: the document reading and filing, the chasing, the checks for what is missing from a file, the reporting nobody has time to do. Joined up to the platform, so nothing is typed twice. That costs a fraction of replacing the system, and it removes most of the manual work the platform left behind.

If you are weighing this up, the build-versus-buy guide goes through the decision in more depth, including a test you can run in a week without a developer.

Why I understand this particular problem

I am currently building a case management system for a UK immigration law firm, replacing paper files and shared inboxes with one place to work. It is in build rather than finished, and I would rather say that than dress it up.

What those have in common with a law firm is the shape of the problem: a business whose real process lives in spreadsheets, inboxes and the heads of a few people, where the cost of a missed detail is high.

A sensible first project

Not a full system. Pick the single task eating the most fee-earner time, usually document chasing or filing, and automate that one thing. It is small enough to prove quickly, and it builds the trust needed for anything larger. If it does not save the time I said it would, you have found that out cheaply.