AI Time Capture for Solicitors: Recover Hours You Never Billed

Yes. Passive time-capture tools record the emails, documents, calls and meetings you work on, match them to matters and suggest draft time entries for you to approve each day. They recover the short, unrecorded units that leak most, but only if each fee earner reviews the suggestions daily; unreviewed entries turn into write-offs and billing disputes.

How much they recover depends on one question: does your work happen where software can see it? Emails in Outlook, documents in the case management system and calls through a connected phone system are visible. A call on a personal mobile, a corridor conversation at court or twenty minutes thinking about a problem on the train are not. Firms whose fee earners work mostly at their desks recover a lot; firms that spend days in hearings or on site recover less and still need manual entries.

Follow me on Instagram@sagnikteaches

Where the unbilled minutes actually go

Most lost time isn't big tasks forgotten. It's small units never recorded, because recording a six-minute email takes almost as long as writing it. Here's a filled-in picture for an illustrative fee earner in a commercial property practice:

Connect on LinkedInSagnik Bhattacharya

How passive capture decides what to log

Under the marketing, the mechanism is fairly simple. The tool watches activity signals: an email sent or opened, a document edited, a calendar event attended, a call logged. It matches each signal to a matter using whatever links exist, such as the email address of the client, a matter number in the subject line, or the folder the document lives in. It estimates duration from time spent active in that item, pausing when the mouse and keyboard go idle. Then it groups signals into a suggested entry and drafts a narrative from the email subjects and document names.

Subscribe on YouTube@codingliquids

Each step can go wrong, and the review exists to catch it. The matching step is the weakest. A realistic example: a fee earner forwards a thread about one client's lease into a new email about a different client's licence, keeps the original subject line, and the tool files forty minutes of drafting on the second matter against the first. Nothing looks odd until the first client queries the bill. Matter numbers in email subjects, set as a firm habit, prevent most of this.

The habit is small but specific. A subject line such as "Re: Fwd: heads" gives the tool nothing to match on except the recipients, and recipients are ambiguous whenever a client has more than one matter open. "[2291] Revised heads of terms: rent review" matches every time, for the fee earner who writes it and for everyone who replies. Two rules make it stick: put the matter number in square brackets at the start of every new email, and start a fresh email rather than forwarding when the subject moves to a different matter. The forwarded-thread mistake above can't happen if the second client's work starts in a new email with its own number.

Three ways a small firm can start

Capture built into your practice management software

Check this first, because it's often already paid for or available as an add-on. Smokeball's AutoTime runs every night and creates time entries for the previous day's matter activity it can see, including documents, emails, memos, tasks and connected calls. Entries stay pending, so the fee earner can edit date, description, duration and rate before anything is invoiced, and a summary email arrives each morning. It won't create entries for activity with no matter assigned or for matters that are closed. Clio's Manage AI (formerly Clio Duo) suggests unlogged time entries from matter activity and can polish time and expense narratives, highlighting what it changed for review; check with Clio whether it's included on your plan or priced as an add-on.

A standalone passive capture tool

Products such as Laurel (formerly Time by Ping) capture activity across a fee earner's systems and draft narratives in each person's style. They're built and priced mainly for large firms, and they only make sense for a small firm if your practice software has no capture at all. Get a quote and ask what integration your case management system needs.

Reconstructing the day with Microsoft 365 Copilot

If you're on Microsoft 365 and don't want new software, Copilot can help rebuild a day from your mailbox and calendar. Microsoft 365 Copilot Business is $21 per user a month on annual billing. It's reconstruction rather than capture, so it works best as a daily habit before you forget. A prompt that works:

List the emails I sent yesterday and the meetings I attended,
grouped by client or matter where you can tell from the subject,
recipients or attachments. For each item give the time, the
subject and one line on what it was about. Flag anything you
can't assign to a matter. Don't estimate durations.

An illustrative extract of the output:

Matter 2291 (industrial unit lease)
- 09:14 Email to client: "Revised heads of terms"; sent amended
  rent review clause
- 11:30-12:00 Teams: call with client and agent re break clause
Matter 2307 (licence to assign)
- 14:02 Email to landlord's solicitors: "Licence to assign - draft"
Unassigned
- 15:45 Email to [first name]: "Quick question about Friday"

Two corrections are needed. The 11:30 Teams entry is right, but Copilot has no idea you spent another forty minutes afterwards reworking the break clause in Word, so that has to be added. The "unassigned" email turns out to be an internal conversation with a colleague and isn't billable at all. The prompt says "don't estimate durations" because those estimates are guesses; the fee earner supplies them from memory, now with a prompt in front of them.

The ten-minute end-of-day review

1. Open the day's suggested entries, sorted by matter.
2. Delete anything that wasn't chargeable work: internal chat, admin, a document opened by mistake.
3. Correct the matter on anything mis-filed. Check forwarded threads in particular.
4. Adjust durations. Suggested times can run long (a document left open over lunch) or short (reading on paper).
5. Add what the tool couldn't see: mobile calls, thinking time, meetings outside the calendar.
6. Edit narratives so each says what was done, specifically enough to survive a client query.
7. Approve. Anything not reviewed within two working days should be flagged to a supervisor rather than approved in bulk.

An illustrative Tuesday's queue for the commercial property fee earner shows where the edits go:

The suggested total was 2.9 hours; the reviewed total is 2.2 hours. Capture over-recorded the idle document and missed the paper and phone work, which is the usual pattern. Approving the suggestions unread would have recorded 1.1 hours that weren't worked and still missed 0.4 hours that were, a net 0.7 hours over in a single day's queue.

Calls that cover two matters need a rule the whole firm follows. When a client rings about both the lease and the licence in one 24-minute call, the tool will attach it to whichever matter the phone system associates with the number, often the most recent. Split it by what was actually discussed (say 0.3 hours to one matter and 0.1 to the other), note "call covered both matters" in each narrative, and never record the full duration against both.

The review replaces most of the time fee earners previously spent reconstructing timesheets, but it doesn't disappear. Treat it as part of the day, not something to batch on Friday afternoon, when the memory it relies on is gone.

Narratives a client will pay without querying

AI-drafted narratives are built from subject lines and document names, which makes them either usefully specific or uselessly generic. Here's an illustrative before and after for the same half-hour of work.

Before (auto-drafted): "Email correspondence; document review."

After (reviewed): "Considering landlord's solicitors' amendments to licence to assign; drafting response on guarantor provisions; email to client explaining amendments."

The second one tells the client what they're paying for. If you use AI to improve narratives, give it the facts and forbid padding:

Rewrite these draft time narratives so each states what was done
and why, in under 20 words. Use only the information given. Don't
add work that isn't mentioned, and don't use "extensive",
"detailed" or "thorough".

Banning inflating adjectives matters because they are exactly what a client circles when querying a bill.

The subtler inflation is in the verbs. Given the draft "Email to client re amendments", an assistant running that prompt might return, illustratively: "Advising client on implications of landlord's amendments to licence." No banned adjectives, under 20 words, and it reads well. But the email simply forwarded the amendments with a line asking the client to call, and "advising on implications" describes work that hadn't happened yet. If the client later asks what the advice was, there's nothing to point to. The reviewer changes it to "Email to client forwarding landlord's amendments; arranging call to discuss", and it's worth adding "don't upgrade the type of work, such as turning 'email' into 'advising'" to the prompt.

The arithmetic for a five-fee-earner practice

Here's how the numbers might work for an illustrative five-fee-earner commercial property practice with a blended rate of $180 an hour, working about 200 billable days a year.

• Recovered time: if passive capture plus the daily review surfaces an extra 25 minutes of genuine chargeable work per fee earner per day, that's about 0.4 hours x 5 people x 200 days, or roughly 400 hours a year recorded that previously weren't.
• Recorded value: 400 hours at $180 is about $72,000.
• What actually gets billed and paid: not all recorded time is billed, and not all billed time is paid. If the firm realises 75% of recorded time, that's about $54,000.
• The cost side: the tool (for Copilot, 5 x $21 is $105 a month, about $1,260 a year; practice-software capture varies by plan), plus review time. At ten minutes a day, review is about 167 hours a year across the team, though much of that replaces time already spent on timesheets.

These are illustrative, and the 25 minutes is the assumption everything rests on. Measure it in your own pilot rather than trusting a vendor's figure; firms with disciplined time recording already will recover less. The figure that matters is billed and collected fees, not recorded hours. For fixed-fee work, the value is different: captured time tells you what each matter type actually costs you, which you can compare against your fees using profit-per-matter tracking.

A wills and probate practice working mostly on fixed fees shows the other side of it. Capture rarely changes the bill, but after three months the practice might see that estates with more than three beneficiaries take twice the recorded time of smaller ones, while both carry the same fixed fee. That finding is worth more than any recovered unit.

Rules that keep recovered time honest

• Only bill work that was done. A suggestion is a prompt, not evidence. If the fee earner can't say what the time was, it doesn't get billed.
• No rounding up to fill gaps. Idle time recorded as activity (a document left open) must come out.
• Follow your fee agreement and your regulator's rules on billing. Captured time changes how you record, not what you're allowed to charge.
• Tell staff what's captured. Explain what the tool sees and how data is used, and take advice before enabling anything that goes beyond matter activity.
• Keep the audit trail. If a bill is challenged, you want the original suggestion and the reviewed entry, so keep both.

On telling staff, a short written notice does more than a mention in a team meeting. An illustrative version for the pilot, to be checked by your employment or data-protection adviser before use:

"From [date], time capture will run for [names of pilot fee earners]. It records which matter emails, documents, calendar events and desk-phone calls you work on, and for how long they are active. It does not read personal email, browsing or messages outside the firm's systems. Suggested entries are visible only to you until you approve them. The data is used for time recording and to understand what our matter types cost us to deliver; it will not be used for performance reviews. Questions to [practice manager]."

If any sentence in that notice isn't true of the tool you've chosen, for example if it can see browser activity, change the tool's settings or change the notice before the pilot starts.

A four-week pilot with two fee earners

Pick two fee earners who record time reasonably well already, so the comparison is fair. For two weeks before the pilot, note their recorded chargeable hours per day. Then switch on capture (or the Copilot routine) for four weeks with the daily review. Compare three numbers: recorded chargeable hours per day, time spent on timesheets, and how many captured entries needed correcting. If hours rise, timesheet time falls and corrections are dropping by week four, extend it.

Illustrative results for the property practice's pilot pair: fee earner A went from 5.1 recorded chargeable hours a day in the two baseline weeks to 5.5 in weeks three and four, and fee earner B from 4.6 to 5.2. Friday timesheet reconstruction, which had taken each of them about 45 minutes, disappeared, replaced by the ten-minute daily review. Corrections fell from roughly one suggested entry in three during week one to one in eight by week four, mostly because both started putting matter numbers in subject lines. An extra 0.4 to 0.6 hours a day is in line with the assumption in the arithmetic above, which is the point of piloting: the firm now has its own figure rather than a vendor's.

Running your first AI pilot covers the pilot mechanics; for the approach outside legal practice, see AI time tracking for small teams without timesheets. If you're weighing capture against other legal AI spend, legal AI pricing for small firms helps put the numbers side by side, and it's worth checking the AI already in your practice software before buying anything new.

AI time capture: questions fee earners ask

Is it lawful to track staff activity this way?

Usually it can be, but employee-monitoring and data-protection rules apply, and they vary. Tell staff exactly what is captured and why, limit it to work systems and matter activity, keep the data only as long as billing needs it, and don't use it for performance surveillance without saying so. Take advice from your employment or data-protection adviser before switching it on firm-wide.

Does passive capture help on fixed-fee matters?

Yes, though not for billing. On fixed fees, captured time tells you what each matter type actually costs you to deliver, which is what you need to price the next one. Many firms find a fixed fee they thought was profitable is being eaten by unrecorded emails and calls. Keep capture running on fixed-fee work and review it monthly by matter type.

Will clients object to AI-written narratives?

Clients object to vague or inaccurate narratives, not to how they were drafted. An AI-suggested narrative that says exactly what was done, reviewed by the fee earner who did it, is usually clearer than one typed from memory on Friday. The risk is padding or generic wording, so the reviewer should edit anything that overstates the work or couldn't be explained if queried.

Further reads

• AI for Small Law Firms: What to Automate First — Whether time capture should come before other automations.
• Can Lawyers Use AI Note Takers in Client Meetings? — Meeting notes that also feed accurate attendance time.
• Copilot in Teams Meetings: Recaps, Actions and Privacy Settings — Teams recaps and the privacy settings to check first.
• Is Microsoft 365 Copilot Worth It for a Small Business? — Whether a Copilot licence pays back beyond time capture.
• How to Measure Time Saved After Rolling Out AI in a Small Firm — How to measure the change honestly after rollout.
• AI Implementation Plan for a Small Law Firm: The First 90 Days — A day-by-day 90-day AI plan for a six-person law firm: policy first, two scored pilots, shadow testing, an error log and a go/no-go decision.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Smokeball Support Hub, AutoTime Basics; Clio Help Centre, Manage AI articles on time entry suggestions and billing narratives; Laurel company information; Microsoft 365 Copilot Business pricing.