AI for law firms: five workflows worth automating first

The five law firm workflows we’d automate first are new-matter intake and conflict-check preparation, chasing client ID and source-of-funds documents, attendance notes, filing emails and documents to the matter, and first drafts of standard letters from your precedents. In each, the automation does the gathering, typing and filing. A solicitor keeps every decision that needs legal or regulatory judgement.

Our founder, Ross Owen, has worked in IT since 2000, specialising in legal and professional services firms. These are the jobs we see filling the gaps around the case management system. For all five, we build with Anthropic’s Claude through its business service, which doesn’t train on customer data and is covered by a data processing agreement. We never use consumer chatbots with client data, and in-house AI on our own hardware in Cardiff is coming soon for work where data shouldn’t leave the building.

1. New-matter intake and conflict-check preparation

Today: an enquiry arrives by email, web form or phone. Someone retypes the details into the case management system, works out the other parties, runs the conflict search and opens the matter. The search is only as good as the names someone thought to enter.

1. The automation reads the enquiry and pulls out the prospective client, the other side and related parties, such as a co-buyer or a company’s directors.
2. It creates a draft intake record and a list of names to search, in your format.
3. Where your system allows, it runs the conflict search and attaches the results for review.
4. Once the matter is approved, it drafts the client care letter from your template.

Stays with a person: deciding whether there’s a conflict and whether you can act. A name match isn’t a conflict, and a clean search isn’t clearance. Under paragraph 6.2 of the SRA Code of Conduct you don’t act where there’s a conflict of interest, with limited exceptions, and paragraph 8.1 requires you to identify who you’re acting for.

Watch: enquiries contain information about people who never become clients, including the other side. Keep only what the conflict check needs.

How it’s delivered: usually a single project at a fixed quote, because it depends on what your case management system can connect to.

2. Client ID and source-of-funds document chasing

Today: someone emails a list of documents, then chases, and chases again. Passports and bank statements arrive in pieces, in different inboxes.

1. When the matter opens, the automation sends your request for ID and, where needed, source-of-funds evidence.
2. It sends reminders on a schedule and stops as soon as each item arrives.
3. It files each document to the matter and ticks it off a checklist the fee earner can see.
4. If you use an electronic ID verification service, it sends the link and records when it’s done.

Stays with a person: the ID verification decision and every judgement about source of funds. The Money Laundering Regulations 2017 require identity to be verified using documents or information from a reliable, independent source, and transactions to be scrutinised, including source of funds where necessary. The SRA’s 2025 thematic review warned against treating source of funds as a tick-box exercise: someone must read and understand what arrives.

Watch: customer due diligence records must normally be kept for five years after the business relationship ends, so keep them in one controlled place, not scattered across mailboxes.

How it’s delivered: a single project, or built from plan hours if you’re on Monthly AI Support.

3. Attendance notes and call or meeting notes

Today: after a meeting or call, the fee earner writes a note from memory, or dictates it for a secretary to type. On a busy day it’s late, or thin.

1. A Teams meeting is transcribed and drafted as an attendance note in your format: attendees, date, advice given, instructions received, next steps and time spent.
2. Where your phone system can record, calls are summarised onto the matter and callbacks become tasks. Where it can’t, a quick dictation does the same job.
3. The fee earner checks, corrects and approves the note, and it’s filed to the matter.

Stays with a person: the content. An attendance note is your record of the advice given, and AI can miss or soften a caveat. The SRA’s August 2026 warning notice on AI says solicitors stay responsible for AI-assisted work, including work by people they supervise.

Watch: tell everyone in the meeting or on the call that it’s recorded and transcribed, and why: that’s part of the right to be informed under UK GDPR. Decide how long recordings are kept once the note is approved.

How it’s delivered: Meeting minutes and call summaries can be built using your Monthly AI Support plan hours, set up to your attendance-note template.

4. Filing emails and documents to the matter

Today: fee earners file emails from Outlook into the case management system when they remember. Attachments land on desktops, and the gaps show up at file review.

1. The automation reads incoming email and attachments and matches each to a matter by reference, client name, property address or the other side’s reference.
2. It files confident matches, with attachments named consistently.
3. Uncertain ones go to a short queue for a person, and anything that looks urgent, such as a deadline or a complaint, is flagged to the fee earner.

Stays with a person: every ambiguous match, and the response to anything urgent.

Watch: a misfiled email puts one client’s affairs on another’s file, a confidentiality problem under paragraph 6.3 of the Code. Sent to the wrong person, it’s likely to be a personal data breach too, which must be recorded and, where it’s likely to pose a risk to people, reported to the ICO within 72 hours. So the automation asks rather than guesses.

How it’s delivered: the inbox assistant covers shared mailboxes. Filing into your case management system, using our document processing, is usually a single project.

5. First drafts of standard letters from precedents

Today: someone finds a similar letter on an old file, copies it and edits the names, and occasionally another client’s detail survives the edit.

1. The fee earner asks for the letter they need, such as a chaser to the other side or a letter confirming instructions.
2. It picks the right precedent from your approved bank, not an old file, and fills in the matter details.
3. The draft lands in Word with the filled-in details highlighted for checking.

Stays with a person: the advice, the tone, anything non-standard, and the decision to send. Your precedent bank stays yours, maintained by a lawyer.

Watch: the Code requires you to give clients information in a way they can understand, and the best possible information about how their matter will be priced. Costs wording in a client care letter needs checking on every matter.

How it’s delivered: plan hours on Monthly AI Support (four a month on Essentials, eight on Growth), or a single project for a large precedent bank.

How to pick the first one

• Count it. Take a team of ten fee earners who each spend 20 minutes a day on attendance notes and filing. That’s over 16 hours a week. Do the same sum for each workflow.
• Start with what’s easy to check. A drafted attendance note or a chasing schedule is quick to review. Matter filing needs more testing first.
• Check what your systems allow. Some case management systems connect easily; others mean working through Microsoft 365 instead.
• Tidy before you automate. If every department opens matters differently, agree one way first.

For a second opinion, our free automation audit is up to two hours with Ross going through how your firm works, then a written report of what to automate first, roughly what it would save, and the data-protection points. You can see the format in our sample report, which uses a fictional Cardiff accountancy practice. There’s more on our AI for law firms page.