« All work
[ Case study ] — Legal

Voice notes that become billable time for solo lawyers

A lawyer speaks for 30 seconds after a hearing or call; it becomes billing-ready time entries, tasks and deadlines in Clio.

VocketOwn productMobile appAndroid beta · iPhone in progress2026
Vocket — product screenshot
Vocket product screenshottryvocket.com
ClientOwn product
IndustryLegal
ProductMobile app
My roleFounder & product owner
[ 01 / 07 ] — About the client

Who it’s for

Solo and small-firm lawyers in the US bill by the hour — but on average only about 2.9 hours of an 8-hour day make it onto the bill. The rest is real work that never gets written down. Most of them run their practice on Clio.

[ 02 / 07 ] — The challenge

What needed to change

Rebuilding the day from memory at night loses hours, and the deadlines promised on calls live only in someone’s head. Lawyers needed to capture work in seconds, from the parking lot — without an AI ever filing something they didn’t approve.

[ 03 / 07 ] — The build

What I built

  1. [01]
    One-press voice capture that works offline — a recording is never lost
  2. [02]
    A capture engine that turns speech into time entries in tenth-hours, matter notes, tasks, deadlines with reminders, calendar entries and expenses
  3. [03]
    New clients created by voice, with a duplicate check that can’t be skipped
  4. [04]
    “Never guess”: an unknown matter or amount becomes a tap-to-answer question instead of a guess
  5. [05]
    Review, edit and confirm — big items start unticked, and nothing is filed without the lawyer’s OK
  6. [06]
    Clio integration that files each item exactly once, reconciles anything stuck and offers undo
  7. [07]
    Deadline reminders at a sensible local hour, an evening close-out and a full audit trail
  8. [08]
    Privacy by design: audio deleted after transcription, encrypted Clio access, export and delete-everything, and a “hallway mode” that hides transcripts in public
[ 04 / 07 ] — How it works

How it works, step by step

  1. [01]

    Press and speak

  2. [02]

    Speech becomes text, tuned to the lawyer’s own matters

  3. [03]

    AI proposes entries — unclear items become questions

  4. [04]

    The lawyer reviews and confirms

  5. [05]

    Each item is filed to Clio exactly once

[ 05 / 07 ] — Results

Results so far

$0.0074

AI cost per capture

~10 s

from speaking to proposed entries

28

end-to-end checks against real Clio

[ 06 / 07 ] — Key decisions

Decisions that mattered

  1. [01]
    A niche over a general assistant: solo lawyers were chosen after scoring nine professions.
  2. [02]
    Accuracy over cost: the stronger AI model won because a wrong date is unforgivable in a deadline product — prompt caching brought the cost back down.
  3. [03]
    A filing instrument, not an AI assistant: the lawyer confirms everything, and only their own voice is captured.
[ 07 / 07 ] — Under the hood

The technology

React Native (Expo)Over-the-air updatesNode.js on RailwaySupabase (PostgreSQL, row-level security)Speech-to-text + Claude with prompt cachingClio API (OAuth, encrypted tokens)Automated evals

For the technically curious. You don’t need to know any of this to work with me.

[ Get started ] — Book a call

Have something to build?

Tell me what’s slowing your business down. I’ll tell you honestly what software can fix — and what it can’t.

[01]Project

Product build

From idea to launched product. Clear scope, weekly releases.

[02]Ongoing

Ongoing partner

Your part-time technical lead: new features, fixes and advice every month.

[03]Add-on

Add AI to your business

An AI assistant, automation or private AI added to what you already run.