What time will you get here?
I'll be there at 2:30 and bring the keys
No final period · “at 2.30” became “at 2:30”
Speak. The text lands at your cursor.
You think in whole sentences and type one letter at a time. By the time your fingers catch up, the thought has moved on. Say it out loud.
Hold ⌘ /.
Speak. Let go.
The text appears where your cursor is, in any app and any text field.
or ⌘ / on your keyboard
Friday
Demo The site doesn't recognize speech. The phrase is prepared in advance, and the app does the real recognition.
Anywhere there's a cursor.
Messengers, email, notes, AI chats, code editors. Each app gets its own rules.
To: Anna
Hi Anna,
the supply agreement is ready to sign. All the edits are in, and I'm waiting for your comments.
Punctuation by voice: “comma”, “new line”
So, like, an idea for the site: show the text appearing right at the cursor.
Filler words removed where they carry no meaning
Cut this text down to three bullet points and keep all the numbers as they are.
Enter
Sent with Enter right after pasting
41// add a "token" check before the request
42if (!token) return
Lowercase first letter · straight quotes · no period
And everything around dictation.
Text cleanup
Three levels in the dictation settings. “Light” runs on your computer: it removes filler words and fixes spaces and capitals without touching the meaning.
So, um, Let's meet at the entrance.
“Strong” rewrites the text with AI using your own key. History shows what changed.
Indicator
It sits at the bottom of the screen and never steals focus. You pick the style and the color.
Punctuation by voice
“comma”, “new line”
Say the mark and it lands in the text. Turn it on for the apps where you want it.
Workflows
“Start my day”
- Open Mail
- Open Calendar
- Open Slack
- Music: resume
One phrase, and Glyvo opens your apps and sites in order, waits, and starts the music.
Dictionary
grand olaGranola
use offuseAuth
Fixes the words the model keeps getting wrong.
History with audio
Every dictation is kept with its recording, so you can find it, play it and transcribe it again. It stays on your computer.
Esc keeps the phrase
Esc cancels the recording, and the phrase stays in History. If you change your mind, press “Restore”.
Snippets
“my billing details”
Say a trigger phrase and Glyvo pastes your saved text in its place.
Speak your language.
Get English.
You said Coordina el lanzamiento con el equipo de diseño antes del viernes.
Pasted Coordinate the release with the design team before Friday.
A second shortcut, ⌘ ', translates your speech into English. Translation can run on your computer too, with the Whisper medium model (about 540 MB).
Demo The site doesn't translate. The app does.
Calls, written down.
And kept with you.
Glyvo transcribes the call as it happens and keeps “Me” separate from the other speakers. It tells voices apart right on your Mac, and the audio stays on your computer.
- Live call translation: a translated line under every phrase
- A summary with key points, decisions and tasks (with your AI key)
- Search across all your meetings
And on iPhone.
Glyvo keyboard
A keyboard of its own, with a mic and “Paste last”. Recognition runs on the phone, with a model of about 460 MB.
Action button
Press it, speak, and the text is on your clipboard. It works from the lock screen too.
Together with your Mac
Dictionary, snippets and history sync over Wi‑Fi, with no account and no cloud.
Your voice never leaves
your computer.
Microphone
Parakeet on Mac
Text|Cursor
No server needed
Parakeet on Mac
Text|Cursor
No server needed
On your computer
Recognition runs on your Mac. The NVIDIA Parakeet model, about 670 MB, downloads once, right inside the app.
No account
No account, no server, no telemetry. There's nothing to sign up for.
Cloud only if you want it
Your voice stays on your computer unless you turn on cloud recognition with your own key.
You speak.
Glyvo writes.
The link will appear with the first release.
Mac for now. A Windows version is in the works. You can change the shortcut in settings.