One continuous gesture
Hold Option to start, draw and speak together, then release to finish the capture.
Point at the problem. Say what you mean.
Hold Option anywhere on macOS, draw over what matters, and speak while you point it out. Release Option to save the annotation with its optional local transcript.
macOS 14+ · Apple silicon for voice
How does voice annotation work in Assist?
While you hold Option, Assist lets you draw directly over the Mac screen and optionally records your voice. Releasing Option saves the annotated screenshot. On Apple silicon, WhisperKit can create a transcript locally; Assist does not upload or persist the raw recording.
Voice Annotation, without the extra steps.
Image and intent stay together
The marked-up screenshot and optional transcript are saved as one reusable capture.
Local transcription
WhisperKit runs on supported Apple-silicon Macs after a one-time model download. Raw audio is not saved.
Three steps, start to finish.
Hold Option
Start annotation mode over the screen you are already viewing.
Draw and speak
Mark the exact area and explain the context in the same moment.
Release to save
Assist stores the annotation and its optional local transcript in history.
About voice annotation
Does Assist upload my voice recording?
No. Optional transcription runs locally with WhisperKit on Apple silicon. Raw audio remains in memory only until transcription finishes and is never saved by Assist.
Does voice annotation require Apple silicon?
Drawing and screenshots work on supported Macs. The optional local Whisper transcription feature requires Apple silicon and a one-time model download.
How long can a voice annotation be?
Assist records up to 90 seconds of speech during an enabled annotation, then processes the optional transcript locally.