use case · email
Say the email. Send it sooner.
Replying is where the day goes — another message, another few paragraphs. Most people speak three to four times faster than they type, so speak them: the words land right in the draft, ready to fix and send.
reply at talking speed
Put the cursor in the reply box — Gmail or Outlook in the browser, or any mail app on your desktop — press Ctrl ` , and talk. Your words appear as ordinary editable text: fix a name, cut a sentence, then send it yourself. Nothing goes out until you decide.
Spoken sentences arrive with detours and "ums". A voice mode tidies the grammar and drops the filler before the words land, so a quick spoken reply still reads like a written one — free on the built-in local model or your own Ollama server, and Pro can route it through your own OpenAI or Claude account.
when the email is about something on screen
A wrong line on an invoice, an order that looks off, a page that won't load — some emails exist only to describe a thing you're looking at. Press Alt ` , drag over that part of the screen, circle the line while you say what's wrong, and paste into the draft. The marked-up screenshot and your transcribed words arrive together — no saving a file, no hunting for the attach button, no typing up what the picture already shows.
private by default
Email is full of things that are nobody else's business, and dictating it doesn't change that. Your speech becomes text on your device — audio isn't uploaded to anyone, and the app tracks nothing about you. What reaches your outbox is exactly what you chose to put in the draft, nothing more. The full network-use table →
Windows 10/11 today — macOS and Linux builds are on the way. Also see writing by voice and voice for support teams.