Click the mic on any task description or reply and dictate it instead. On a computer, a Whisper model transcribes it inside your browser, so your voice never leaves the machine. On a phone, the mic uses the phone's own dictation.
Live capture from the AgentRQ app — the mic button's entry point into dictation.
Whisper in the browser on your computer, the phone's own dictation on mobile.
Every description and reply field has a mic button next to it. Click it, grant microphone access once, and start talking.
On a computer, a Whisper model downloads once (with visible progress) and runs in a background Web Worker, so the page never stalls. On an iPhone, iPad or Android phone, the phone's built-in speech recognition does it, with nothing to download.
The transcript is inserted at your cursor in whatever field you were dictating into (on a phone, word by word as you speak). Keep editing it like anything you'd typed yourself.
On a computer there's no speech-to-text API call: no audio uploaded to a transcription service, no third party hearing what you said. The model runs in your browser, on your hardware, and the only thing that leaves the Web Worker is the text it produced.
That matters for the same reason auto-title generation runs locally too: task descriptions and replies often contain internal hostnames, file paths, or context you wouldn't want leaving your machine just to get transcribed.
Running Whisper in a mobile browser tab crashed phones, so phones and tablets never run it. The mic there uses the same speech recognition as the keyboard's dictation key, and a phone with dictation turned off shows no mic button at all.
That audio is handled by the phone's own speech service, which Apple or Google may process on their servers, as with any dictation on the device. AgentRQ only receives the text. If you need dictation that never leaves the device, use it on a computer.
Each workspace has a speech language setting, so you can dictate in the language you actually think in, not just English. On a computer it picks Whisper's language; on a phone it goes to the phone's dictation, which covers every language the phone dictates in and keeps your region (German on an Austrian phone dictates as de-AT).
On a computer, like auto-title generation, the model is downloaded once and cached by the browser, so every dictation after the first starts instantly. On a phone there's no model to download at all.