You can now dictate a task

Tap the microphone next to the composer or in chat search and speak. The words appear in the field as they are recognized. Nothing is sent until you press send.

Recognition follows the interface language: Russian or English. It works best in Chrome and Edge, on HTTPS or localhost. The audio does not go to RootPilot’s servers — the browser handles speech-to-text.

This is dictation, not hands-off voice control of the server. After that the task is a normal message: you see the steps, and risky commands wait for your OK.