Dictating instead of typing

The microphone beside the message box turns your speech into text in the draft. Click it to start, click again to stop.

It is worth having: the requests that work best are the ones with context in them, and people type those short and speak them long.

Where it runs

Dictation uses your operating system's speech recognition, not kyte's model.

On Windows, online speech recognition has to be switched on in the system settings. If it is off, kyte says so and offers a button that opens the right settings page.

What it does with the text

Nothing. The words land in the draft as if you had typed them; nothing is sent until you press send. Correct the text first if the recognition swallowed a name — it usually does, with company names.

The same rules apply

A dictated message is a message. Personal data in it is noticed the same way, see personal data in a message, and the level in the switch still decides what may happen afterwards.