Managing matter email with AI: filing, summaries and follow-ups
By the ChamberOne team 4 minute read
Ask a fee earner where the important email is and the answer is often a shrug. It is in an inbox, a folder with a half-remembered name, or a forwarded copy that nobody can date. Email is where much of the real record of a matter lives, and it is the least organised part of it. AI can help in three ways: filing messages to the right matter, summarising long chains and suggesting follow-up tasks. Each has a failure mode, and each touches privacy. This article takes them in turn.
Key points
- Let the tool suggest which matter an email belongs to, and keep a one-click human confirmation.
- Summaries of long chains should be dated, linked to messages and checked for instructions, deadlines and figures.
- Propose tasks from emails, but show the source sentence and require acceptance; keep court dates in a separate human process.
- Limit what the tool can read, respect restricted matters and check privacy terms before connecting mailboxes.
Why email is the weak point in the file
Matter files often hold the formal documents in good order while the correspondence that explains them sits in personal mailboxes. When a colleague goes on leave, or a client disputes what was said, the missing emails cause trouble. Regulators and insurers also expect a complete record of what advice was given and when.
The reason for the mess is simple: filing an email takes effort at the moment when the person is busiest. Anything that reduces that effort helps, as long as it does not file things in the wrong place.
Filing: let the tool suggest, let a person confirm
A tool can compare an incoming email with the firm's matters, using the sender, the subject line, reference numbers and the content, and suggest the most likely matter. For routine correspondence with a clear reference, the suggestion will usually be right. For a client with several matters, or a generic subject line, it will often be wrong or uncertain.
The risk of a wrong match is more than untidiness. An email filed on the wrong matter may be seen by people who should not see it, or be missed when it matters. So keep a confirmation step: the tool proposes, the person accepts with one click, and anything low-confidence goes to a queue for a human decision.
- Show the suggested matter and the reason, such as a matching reference number.
- Make correcting a suggestion as easy as accepting it.
- Never auto-file onto a restricted matter without a person's decision.
- Keep the original email unchanged, and record who filed it and when.
Summaries of long chains
Long email chains are where context goes to die. A summary of thirty messages can save a colleague half an hour. It can also leave out the one message in which the client changed instructions, or flatten a disagreement into apparent agreement.
Ask for summaries that are organised by date, name who said what and link each point to the message it came from. Ask for decisions, commitments and open questions to be listed separately. Then check the items that matter: instructions, deadlines, figures and promises. Do not paste a machine summary into the file as if it were a note of advice. Label it as a summary, with the date and the chain it covers.
Follow-up tasks
A good use of AI on email is spotting what needs to happen next: a request for a document, a deadline mentioned in passing, a promise to call back. A tool can propose tasks from an email or a chain, with a suggested date and owner.
Treat the proposals as a reading aid. People write deadlines ambiguously, and a model may mistake a date in a quoted earlier message for a new one. A proposed task should show the sentence it came from, and a person should accept it before it enters anyone's task list. Court and limitation dates deserve a separate, human-controlled process regardless.
Privacy limits
Email carries the most candid material on a file, often including personal information about third parties. Before connecting a tool to mailboxes, ask which messages it can read, what is sent to an outside provider, whether it is kept and who can see the results. A tool with access to a whole mailbox sees private and unrelated messages as well as matter email.
Limit scope. Prefer tools that act only on emails a user chooses to file or forward, over those that scan everything. Exclude folders such as personal and HR. Respect restricted matters, so that summaries and suggestions about them are visible only to the matter team. Check your firm's policy and the guidance from your regulator or bar on confidentiality and technology.
Habits that make the tool work
Technology helps most when people also change small habits.
- Put the matter reference in the subject line of outgoing emails and ask clients to keep it.
- Use one thread per topic, and change the subject line when the topic changes.
- Copy the matter's shared mailbox or filing address on important messages, if the firm has one.
- File as you go, in short bursts, instead of once a week.
- Mark emails containing advice or instructions so they are easy to find.
Secretaries and fee earners: who does what
Decide who is responsible for filing and checking. In some firms, secretaries confirm suggestions and fee earners look at summaries and tasks. In others, each fee earner handles their own. Either works if it is clear. What does not work is assuming the tool's suggestion is the record.
Review a sample each month: count wrongly filed emails, tasks that were wrong or missed, and summaries that omitted something important. Adjust the rules and tell the team what you found. A small, regular check keeps confidence honest.
- 1The tool suggests a matter, task or summary
- 2A person confirms or corrects it
- 3Review a sample each month
- 4Count wrong filings, wrong tasks and omissions
- Then back to step 1.
Getting started in a small firm
Start with one team and one function, such as suggested filing. Run it for a month with confirmation on every item. Measure how often suggestions were right and how long filing took before and after. Add summaries for long chains next, then task suggestions. Throughout, keep the confirmation step. The aim is a complete, accurate record with less effort, not a record that nobody has looked at.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
Onboarding a new client matter: an AI-assisted step-by-step checklist
Walk through every step of opening a new matter — from conflict check to task assignment — showing where AI saves time and where only a lawyer can decide.
6 minute readUsing AI matter briefs to pick up an unfamiliar file quickly
How a generated brief speeds up handovers and cover, and how to check it against the file before acting.
5 minute readAI task suggestions and matter phases: standardising how work starts
How matter-type templates and suggested tasks help new files start consistently, and why each suggestion needs review.
5 minute read