Hold Ctrl+Win, talk the way you think, let go. Saylo drops the ums, fixes your corrections and puts tidy text wherever your cursor is, in about a second.
Works in any app you can type in
Saylo sits quietly in your tray. It only listens while you hold the keys.
Click into any text box, then hold the shortcut.
Ramble, pause, change your mind halfway. Talk like you would to a friend.
Clean, punctuated text appears where you were typing.
Not a word-for-word transcript. Saylo edits as you speak, without adding anything you didn't say.
Ums, uhs, “like”, “you know” and repeated words are gone.
Change your mind mid-sentence and Saylo keeps only the final version.
Say “first… second… third…” and get a numbered list.
Email, chat, docs, code editors, browser forms. If it has a cursor, Saylo can type there.
Typical time from letting go to text on screen is under a second and a half.
Past dictations are saved only on your PC, and you can turn that off. Saylo has no server of its own.
Every build comes straight from the public source code on GitHub.
Windows 10 and 11. Installs in seconds, starts with Windows, lives in the tray.
or the portable .exe · all releases
The Saylo bubble shows up above your keyboard in any app: hold it and talk. Works with Gboard or any keyboard. Android 8 and newer.
Google Play listing coming soon
Yes. Saylo uses Groq's speech and language models, and you plug in your own Groq API key, which is free for personal use within Groq's limits. A paid version that needs no key is coming in the next update.
Sign in at console.groq.com/keys, click “Create API Key” and paste it into Saylo during setup. It takes about a minute. The key is stored only on your computer.
Saylo records only while you hold the shortcut. The clip goes directly from your PC to Groq to be turned into text and cleaned up; Saylo has no server in between. Your history (the text, plus each recording for 30 days so you can replay it) is kept on your PC only, and you can switch it off or delete it.
That warning appears for new apps that aren't code-signed yet. Click “More info”, then “Run anyway”. The installer is built automatically from the public code in the GitHub repository.
English for now.
Not yet. On Windows it's Ctrl + Win, held down while you speak.