← co-agent

The whole policy, in one line:

Last updated July 20, 2026 · Co-Agent is a Source4AI product

What we collect, and why

• Your email address — when you buy, subscribe, or sign up. We use it to run your account and to send occasional emails about our products and offers. Every email has an unsubscribe link that works.
• Operational metering — token counts and session counts tied to your license, so we can pay our AI bill and catch abuse. Counts, not content.

What we deliberately do not collect

• Your voice recordings. We never store them, and in the app they don't even pass through our servers (see below).
• Your screenshots or screen contents.
• Your files, messages, browsing, or anything your agent works on.
• Analytics or ad-tech: this site and the app ship zero trackers, zero ad pixels, zero third-party analytics.

Where your voice actually goes

When you hold the talk keys in the app, the audio of what you say — and one screenshot of your screen for that question — is sent directly from your computer to Google's Gemini API over an encrypted (TLS) connection to generate the response. It does not route through our servers, so we couldn't read it even if we wanted to.

One exception, clearly bounded: during a step-by-step walkthrough you asked for (a guided tutorial where the app points at what to click), the app also takes a screenshot after each of your clicks so it can see where you are and place the next step accurately. Those screenshots follow the exact same direct-to-Google path — never our servers — and it only happens while a walkthrough you started is active. Pressing the talk keys (or Escape) cancels the walkthrough instantly, which also stops those screenshots.

• Our server's only role is minting a single-use, short-lived session token so you don't need your own API key. The token expires in minutes and works once.
• We use Google's paid API, which per Google's terms is not used to train their models.
• Release the keys and the microphone lane is closed. No wake words. Silence sends nothing.
• Prefer a direct relationship with Google? The lifetime license runs on your own Gemini key.

Blind Private Relay

On by default for subscribers, your voice reaches Google through our relay server instead of connecting directly. This splits knowledge so no one holds both halves:

• Google processes what you said, but sees only our relay's IP address — never yours, never your location.
• We know which license connected (to bill and rate-limit), but the relay only forwards encrypted bytes it cannot read — it never decrypts your audio. We log no content.
• It's a dumb pipe by design: a CONNECT tunnel that can only reach Google's Gemini endpoint and nothing else.
• One toggle in Settings → Privacy turns it off for slightly lower latency (still encrypted, just direct to Google).

Your agent, your machine

Real work — files, research, business context — is handled by your own agent (Hermes, Claude Code, Codex…) running under your accounts on your machine. None of it routes through us. Actions on your computer require your explicit on-screen approval and are receipted locally, for you.

The cloud Agent Computer

Hosted desktops are disposable by design: created when needed, wiped after idle. We don't inspect what happens on them; the visual stream goes to your app over TLS. Don't store anything precious there — that's a feature.

This website's live demos

The voice demo streams your held-button audio through our server to Google's Gemini API and is not stored by us. Demo computers are deleted automatically minutes after creation. We keep coarse rate-limiting counters by IP for a few hours to prevent abuse.

Payments

Card processing is handled by Stripe; card numbers never touch our systems.

Deletion & questions

Email [email protected] to delete your account and email address, unsubscribe, or ask anything about this policy. A human answers.