Voice control
Drive Pact hands-free — your speech is transcribed in the browser and parsed into structured, audited commands.
Voice control lets you operate Pact by speaking. You say what you want, and Pact turns it into a sequence of structured actions that run against your workspace.
Opening it
Press ⌘⇧V on desktop, or tap the opt-in mic button on mobile.
Your audio stays on your device
Your speech is transcribed in the browser using the Web Speech API. Only the final text transcript is sent to Pact — the audio never leaves your device.
Audio never leaves your device
Transcription happens locally in your browser. Pact only ever receives the final text, never the recording.
From speech to commands
The transcript is parsed into an ordered list of structured commands:
- Move a deal's stage.
- Set a deal's close date.
- Add a note.
- Create a task.
Relative dates like "next Friday" resolve against today.
How commands run
A deterministic, workspace-scoped dispatcher runs each command in order and writes to the record timeline and an audit log, so every action is traceable.
When the mic is live
The mic is only active while the panel is open or the button is held. Close the panel or release the button and listening stops.
Gating
Voice control is gated by a workspace feature flag. Settings live at /settings/voice.
AI runs only when configured
A live Claude model parses the transcript into commands. It runs only when AI is configured for your workspace.