Turn voice into finished writing
Give a spoken instruction, optionally include selected text or a screenshot, and let a local or cloud writing model produce the result.
How Command works
- 1
Optionally select text in the active app.
- 2
Press Option–Shift–Space and say the outcome you want, such as “make this clearer and more concise.”
- 3
Your speech is transcribed locally. The selected writing model receives the instruction and only the context enabled for that request.
- 4
Review, copy, paste, or refine the result with another voice or text turn.
Selected text, screenshots, and attachments
- Send selected text uses Accessibility permission to include the text highlighted in another app.
- Command with screenshot uses Screen Recording permission and is available only with a vision-capable model.
- Attachments and earlier conversation turns can provide additional reference material.
- Treat attached and selected content as untrusted reference material and review the generated result before using it.
Follow-up turns and safe replacement
Continue a Command conversation by voice or text to make the result shorter, warmer, more formal, or otherwise refined.
Safe refinement replacement replaces the previous result only when the exact text is still unchanged. This helps avoid overwriting edits you made after the first result.
Personal context
Add up to 20,000 characters of reusable context, such as your name, role, company, preferred tone, sign-off, and recurring facts. WhisperInk includes it with every Command request.
Personal context stays on your Mac with local models. If you select an OpenAI model, it is sent as part of the Command request.
Useful Command settings
- Send selected text
- Copy result to clipboard
- Automatic paste
- Safe refinement replacement
- Result notification
- Advanced Command instructions