Agent Brain

Agent Brain

+ Capture

Sign in

Your captures, understood — and available to the AI agents you choose.

Understand captures with Anthropic's Claude?

To work out what you saved and why, Agent Brain sends your captures (text, links, images and text recognized in them) to Anthropic's Claude. Voice notes your phone couldn't transcribe are transcribed by Cloudflare Workers AI. Anthropic doesn't use this data to train its models.

Captures containing the sensitive details you choose in Privacy (like one-time codes, passwords and card numbers) are never sent. You can turn this off anytime in Settings → Privacy; captures are then organized with simple rules instead.

Connect an AI agent

Create a connection and paste its URL into Claude, ChatGPT, or any MCP client.

Permissions: Full access

Write back (resolve loops, record actions, report progress) Sensitive captures — one-time codes, card numbers, medical View original images

Anyone with this link can access what you allowed — treat it like a password. You won't see it again; you can regenerate it anytime.

Apps that support sign-in can instead add and approve access in the browser.

Connections

Each agent sees only what its connection allows. Change or revoke access anytime.

Projects

Projects give intent inference context ("this is competitor research for Mosey").

Privacy

Account