Meta's Muse agent gets its own computer — and a warden

Meta's answer to "why would you trust an agent with your credit card" is shaped like hardware: every Muse user gets a Muse Secure VM, a dedicated cloud machine holding both the agent and their data, walled off from every other user's. A separate Sentinel agent lives on that same machine, isolated at the system level, and nothing Muse does reaches the internet until Sentinel signs off.

The agent runs on Muse Spark, Meta's new model built for agentic work. You talk to it like a person — in the Muse app or straight inside WhatsApp — and it opens browsers, fills forms, books travel, negotiates on your behalf, and keeps working after you close the app. Checkout goes through Link by Stripe, and Muse never sees your passwords or card numbers: credentials are stored separately and used without being exposed. Emails and purchases need your approval, every action lands in an audit trail, and Meta says none of it feeds its ad systems.

Live now in the US on iOS, Android and the web, free at the basic tier with paid plans above it. AI glasses come later. Shop Pay and 1Password support are queued.

Why it matters: routing an agent with a spending budget through WhatsApp puts one in front of billions of people who never went looking for it.