What Makes an AI Time Narrative "Client-Ready"?
An AI can write a time entry in a second. Writing one your client will actually pay for is a completely different problem. Here’s the anatomy of a great narrative.
A large language model can produce a time-entry narrative in about 400 milliseconds. Producing one your client will actually pay for is a completely different problem.
Most AI timekeeping tools get the first part right and the second part wrong. They produce grammatically-correct, plausible-sounding narratives that look fine on a review screen but fail the first time they land on a GC's desk at a Fortune 500 client. Here's why — and what we've learned making Esqio's narratives hold up in the real world.
A good narrative does four things
When a client in-house counsel reviews your bill, they're implicitly asking four questions about every entry:
- What was done? A specific action, not a category. "Drafted response to plaintiff's motion to compel" — not "Legal research."
- Why was it necessary? The connection to the matter's strategy or the client's objective. "...in anticipation of tomorrow's hearing on discovery scope."
- Who did it? Implicit from the timekeeper, but level of seniority should match the complexity of the task.
- How long did it take? And was that time proportionate to the task?
A narrative that satisfies all four reads naturally, bills out cleanly, and builds trust. A narrative that skips even one triggers a flag, a question, or — worst case — a silent write-down.
What most AI narratives get wrong
The common failure modes we see from naïve AI generation:
- Generic verbs. "Worked on" and "reviewed" where a specific action is available. This signals the timekeeper wasn't paying attention.
- Missing context. The narrative describes the task but not its purpose within the matter.
- Block-billing hidden as prose. "Reviewed discovery, attended call with client, drafted motion" — three distinct activities smushed into one entry. Most OCGs explicitly forbid this.
- Over-clever paraphrasing. The AI rewrites the same two activities five different ways across a week, making the timesheet look padded.
- Mismatched register. Overly-casual or overly-formal for the firm's voice and the client's industry.
How Esqio writes narratives differently
We've spent the last two years working with firms to figure out what separates a billable narrative from a rejected one. The system we've built has four distinctive properties:
It's grounded in specific artifacts
Every Esqio narrative is generated from actual work product — specific emails, specific documents, specific meeting invites. The narrative can reference the "fifteen-page expert report" because it saw the fifteen-page expert report.
It's calibrated to your firm's voice
The first two weeks after deployment, Esqio is watching your team edit. Every edit — a verb swap, a phrase added, a passive voice rewritten — trains a firm-specific voice model. By week three, you stop editing.
It knows your clients' OCGs
Upload a client's outside counsel guidelines once. Esqio enforces them automatically — block-billing prohibitions, minimum narrative detail, non-billable categories, phase-code restrictions, all the rest.
It doesn't pad
If you worked 0.3 hours, the narrative reflects 0.3 hours of actual work. Esqio will never stretch an entry to look like more than it was. That's the single fastest way to lose client trust, and it's a rule the system is built around.
A before-and-after example
Naïve AI output:
"Reviewed documents and drafted email to client regarding case status." · 0.6 hours
Esqio after two weeks of learning your firm's voice:
"Reviewed most recent expert report (sixteen pages) prepared by Dr. Halpern; drafted email to client summarizing key findings and recommending next steps for engagement of responsive expert in connection with upcoming Daubert motion." · 0.6 hours
Same six minutes. Dramatically different outcome at the moment the client reviews the bill.
The bottom line
AI narratives are going to become the default for professional services billing within two or three years. The firms that win this transition are going to be the ones whose AI produces narratives clients actually want to pay for — not the ones that cost the least, or ship the fastest, or demo the best.
That's the single standard we measure Esqio against. See what it looks like against your firm's real data.
Stop reading. Start billing better.
See a 30-minute demo tailored to your firm. We'll show you exactly how Esqio captures time in the tools you already use.