Skip to content

About Keebye

On-device voice dictation for builders running parallel AI-agent workstreams. Hold a hotkey, speak, and the text lands in whatever app has focus — speech-to-text runs locally.
Who it is for
Builders running parallel AI-agent workstreams on Mac and Windows
Problem it solves
Repeatedly typing prompts, messages, and instructions can interrupt the flow of work. Dictation also needs to handle technical vocabulary and insert text reliably into the tools people already use, including terminals.
The solution
Hold a key, speak, and the text lands in whichever app has focus, including your terminal. Speech-to-text runs entirely on your computer, so no audio leaves it.
What makes it unique
Speech-to-text runs on-device with no cloud transcription and no wake word. The network is used only to download models, check for updates and confirm your plan, never to send audio.

Features

  • On-device speech-to-text (no audio leaves your computer)

  • Hold-to-talk or tap-to-toggle hotkey

  • Inserts text into the focused app by paste or synthetic typing

  • Optional multilingual engine covering 25 languages

  • Local, text-only dictation history with 30-day auto-delete

  • Four things it does on your computer

    No wake word, no cloud transcription, no model deciding what you meant. Every step below runs locally.

Details

Platforms
  • macOS
Pricing
  • Freemium
  • Subscription
  • One-time purchase
  • Free trial
Built for

Builders running parallel AI-agent workstreams on Mac and Windows

Building since

2026

Technology

Keebye product screenshot

1 / 1

Website performance

Recorded mobile lab measurements and finalized weekly standings from IndieTools Speed.

Keebye website speed overview

96
Performance scoreGood

In IndieTools' mobile lab test on November 9, 2026, the Keebye website scored 96 out of 100 (good). Largest Contentful Paint took 1.67 s, rated good.

Measured (UTC)

LCPLargest Contentful Paint
1.67 s
Good
TBTTotal Blocking Time
69 ms
Good
CLSCumulative Layout Shift
0.000
Good
FCPFirst Contentful Paint
1.67 s
Good

Not ranked in a finalized week yet.

Stored lab observations from controlled test runs, not real-user Core Web Vitals or a guarantee of search rankings. Weekly standings remain attached to their finalized edition.Method psi-mobile-two-sample-mean-v1, 2 samples per reading. Lab results can vary between runs.

DR4 (primary model)
2

First DR4 reading

DR3
17

First DR3 reading

Model difference−15. DR4 is 15 points lower than DR3.

Measured · keebye.com

What do DR4, DR3 and the model difference mean?
DR4:
Ahrefs' newer model. It gives more weight to links from sites with organic search traffic and rates a subdomain on its own links.
DR3:
The previous model, shown in the Ahrefs app until 1 December 2026.
Model difference:
DR4 minus DR3 for the same domain. The models calculate differently, so this is not growth or a drop; the “since” lines compare each model with itself.

Related products

Others in the same categories.

Never hear "hey, you're on mute" again. One-press mute, camera, and share controls. Same buttons, same place, every meeting app you use.