AI-drafted time-entry narratives
A legal business where this lands in Finance.
02 — What is actually going wrong
Time gets written up on Friday afternoon from memory. Memory is generous to the client and unkind to the firm: work genuinely done goes unbilled, and the narratives that do get written are thin enough to invite write-downs from the client's billing team.
How it works
The system watches the systems your fee earners already use — calendar, email, document management — and drafts time entries with a narrative describing what was actually done. The fee earner reviews, edits and approves; nothing posts unreviewed.
What good looks like
Write-up time on Friday collapses, recorded hours rise a few percent because the small entries stop falling through, and client write-downs fall because the narratives describe real work in specific terms.
03 — The problem, in money
- Timekeepers10 people101 peoplePlaceholder — not a validated benchmarkStarting point
- Admin hours per timekeeper per week3 hours10 hoursPlaceholder — not a validated benchmarkStarting point
- Recovered billable ratio0.05 ratio0.25 ratioPlaceholder — not a validated benchmarkStarting point
- Billable billable_hourly_rate300 USD780 USDPlaceholder — not a validated benchmarkStarting point
- Share automated0.4 ratio0.5 ratioPlaceholder — not a validated benchmarkStarting point
- Share of freed time redeployed0.5 ratio0.5 ratioPlaceholder — not a validated benchmarkStarting point
04 — The options
05 — The call
Buy it. An off-the-shelf tool wins this one — it gets you live faster, and the math does not justify building from scratch.
06 — How this goes wrong
- Auto-posting without review. A fabricated or duplicated entry on a client bill is a professional conduct issue, not a billing error.
- Narratives that are verbose rather than specific. Billing teams reject padding faster than they reject brevity.
- Counting recovered billables and freed admin hours as separate wins on the same hour — the model splits them deliberately, and you should not add them back together.
- Monitoring email content without telling your people. Get consent and scope it explicitly.
07 — How you would actually do it
- 01Confirm what your practice management system will accept via API before choosing anything.
- 02Scope the activity sources, and be explicit with staff about what is being read.
- 03Pilot with two willing fee earners for a full billing cycle.
- 04Compare their recorded hours against their own prior three months, not against the firm average.
- 05Tune narrative style against what your biggest client's billing team actually approves.
- 06Keep review-before-post permanently. This is not a maturity stage to graduate out of.
08 — Tools worth looking at
Named for orientation, not endorsement. No affiliate arrangements, and nothing here has been paid for.
09 — Run it on your numbers
Start from industry averages, then drag the inputs to match your business. The number moves live.Drag each input to your number — the starting points are neutral midpoints, not industry data. The number moves live.
10 — The evidence · what others report
Reported numbers are third-party claims, source-attributed — not verified by me, and never blended into the model above.
11 — Questions owners ask
Is this your business?
Get your free AI Opportunity Score — the same first pass I run at the start of every Audit.
Get Your Free Score