What we listen to.
What we don’t.

Audio is captured only while speaktobrowse is toggled on. It streams directly from your browser to our realtime voice provider over a peer-to-peer WebRTC connection. We do not record, proxy, or retain your audio. When you toggle off, the stream ends and the buffer is discarded.

Our server is involved only at session start, to mint a short-lived (≈60 second) token that authenticates that single connection. The audio itself never touches our servers.

To act on the page, the extension reads the visible DOM and sends a structured summary to the model along with your spoken command. We send the minimum needed to identify what you want to click, fill, or read.

Sensitive form fields are excluded by default. Password inputs, payment fields, and any input marked autocomplete=“cc-*” or autocomplete=“current-password” are never read or transmitted.

On install, the extension generates an anonymous identifier and stores it locally in Chrome. That identifier is the only thing we use to count your free minutes. We do not ask for your name, email, or payment details for the free tier.

The API key that powers the voice model is held on our servers, not in your browser. The extension never sees it. When you start a session, our server mints a short-lived token (about 60 seconds) that the extension uses to open one direct connection to the voice provider.

The extension does not run analytics, telemetry, session replay, or fingerprinting. There are no third-party trackers. We do not know when you use it, how often, or on which sites.

Microphone. Used only while speaktobrowse is toggled on. Chrome shows the recording indicator the entire time the mic is live.

Active tab. Used to read and interact with the page you are currently on. The extension does not run on tabs you have not activated it on, and it does not run in the background.

Storage. Used to keep your API key and your preferences on your device.

The only third party involved in normal operation is OpenAI, which processes your audio and the page context as our realtime voice provider. Their handling of that data is governed by their own privacy terms.

We do not share data with anyone else. We do not sell data. We do not have data to sell.

When this policy changes in a material way, the date at the top changes and we post a note in the extension’s release notes. Continued use after a change means you accept it.