Meta Muse: personal agent features and architecture explained
Muse is a Meta personal AI agent that can browse, send emails and make purchases, with each agent on its own cloud VM, Stripe one-time cards for payments, and access limited to U.S. adults.
Meta just built an AI that has its own computer, its own credit card and permission to spend your money (Save this) Muse is a personal AI agent that connects to your real apps and acts inside them, rather than just answering questions. It can send emails, book travel, fill forms, make purchases, negotiate bills, and turn recipe Reels into grocery lists. You give it a goal and it plans, opens a browser, and works through the task on its own. The architecture is what makes it different because each agent gets a dedicated cloud machine. Muse Secure VM holds your data and credentials and runs its own browser you can watch, which is why the agent keeps working after you close the app. A separate system called Sentinel runs on that same VM but stays isolated from Muse, approving, blocking, or escalating every action because of that split, Meta says Muse never sees your passwords or payment details. Connections are opt in one service at a time across email, calendar, payments, health, smart home, shopping, and dining. Integrations include Gmail, Spotify, Ticketmaster, Shopify, and OpenTable, plus Instagram and Facebook so it can learn your preferences. Payments run through Link by Stripe using one-time-use cards with purchase protections. If a service has no connector, Muse wires up the API itself and if there is no API it simply uses the browser. Access is US only for now, for adults, via a new iOS and Android app, the web, and WhatsApp, with AI glasses coming soon. It is free with limits, with Power at $20 per month and Maximum at $100 per month, and no ads inside the product.
This post quotes another post: “Zuckerberg unveils Muse, Meta's personal agent”