100% on-device · nothing uploaded

Talk, and it becomes finished work.

Hold a key, speak, and get clean text — a commit message, a ticket, a polished paragraph — right where your cursor is. It all runs on your machine, so you can finally dictate at work. Mac first; Windows on the roadmap.

No account needed to use it · Bring-your-own cloud key is optional and off by default
🎙️ you said:
→ BoloMic, locally:
Why this doesn't exist yet

Smart dictation, or private dictation. Rarely both.

The tools that turn talk into finished text upload your audio — and some screenshot your screen for "context." The ones you can run locally mostly just transcribe and leave the thinking to you. BoloMic does the smart part and keeps it on your machine.

☁️ Cloud dictation tools

  • Uploads your audio to a server
  • Some screenshot your screen for context
  • Blocked by most work / NDA policies
  • Stalls on slow Wi-Fi; $15+/mo forever

🔒 BoloMic

  • Audio never leaves your machine
  • Finished work, not just a raw transcript
  • Runs where cloud tools are banned
  • Mac first · no forced subscription
More than "remove the ums"

Dictation that ends in finished work

It doesn't just transcribe. Say what you want, and a local model does it — offline.

🔒

Fully local

A local speech model transcribes and a small on-device model cleans and formats. No network call unless you choose one.

Mac first

Native and lean — fast cold start, light on memory. Windows is on the roadmap; vote for it with your signup.

Agentic commands

"Make this a commit message." "Turn these notes into a ticket." "Summarize what I highlighted." Locally.

⌨️

Paste anywhere

Push-to-talk, then text lands at your cursor in any app — editor, terminal, browser, chat.

📓

Your vocabulary

Custom terms, names and replacements so it spells your stack and your project right.

🔑

BYO cloud key

Want a bigger model for a hard task? Plug in your own API key. Your call, off by default.

How it works

Four steps. Zero servers.

Hold the hotkey and talkPush-to-talk from anywhere on your system.
Transcribed on-deviceA local speech model turns your audio to text, then discards it.
A local model cleans & transformsPunctuation, formatting, or a command like "make this a ticket."
It pastes at your cursorFinished text, in whatever app you're in. Nothing left your machine.
Help us price it

Which would you actually pay?

We're allergic to subscriptions for tools that run on your own hardware. Tap the one you'd choose — it's how we decide.

Pay once
$49 one-time
  • ✓ Local transcription + AI, forever
  • ✓ Mac first (Windows on the roadmap)
  • ✓ Free updates for a year
I'd pick this →
Free + Pro
$0 + $5/mo Pro
  • ✓ Core dictation free, local engine open
  • ✓ Pro: agentic commands + sync
  • ✓ Cancel anytime
I'd pick this →

Not charging anything today — this is a one-tap signal so we build the right model.

Builder's note: BoloMicbolo is Hindi for "speak." Whatever else changes, the promise won't: your voice stays on your machine.