Meta Launches Muse, a Personal AI Agent That Books, Buys, and Negotiates for You
Meta has released Muse, a personal AI agent that reads your email, books your travel, fills in forms, negotiates on your behalf, and pays for the results with your card. The company announced it on Tuesday, and it is rolling out in the United States on iOS, Android, and the web, with support for the company’s AI glasses to follow. Muse runs on a model Meta calls Muse Spark, described in the announcement as the company’s most capable model to date, built for real-world agentic work.
What can Muse actually do?
The distinction Meta is drawing is between a chatbot and something that acts. Muse keeps working after the app is closed, makes suggestions nobody asked for, remembers preferences across sessions, and can turn a saved recipe reel into a shopping list. Payments run through Stripe’s Link with purchase protections, and Meta says support for Shop Pay and 1Password is on the way. An agent that fills forms and completes purchases can also fill the wrong form and complete the wrong purchase, at speed, with a saved card, which is why the security design carries so much of the product.
How much does Muse cost?
There is a free tier plus subscriptions at $20 and $100 a month. Meta says most people should be able to do what they need within the free tier, positioning the paid plans for heavier use. There is no advertising inside the product.
How is Muse built to stay safe?
The security architecture is more considered than the usual launch-day assurances. Each user gets an isolated virtual machine holding both the agent and their data, so agents cannot reach each other. A separate system called Sentinel sits between Muse and the internet. It distinguishes read access from write access with time-limited permissions, allows low-risk tasks to proceed, and stops to ask before anything consequential. Meta says credentials are stored so that Muse cannot see passwords or card details, and every action leaves an audit trail.
Later this year, Meta plans a confidential version in which the company itself cannot observe what happens inside the workspace. Until that ships, the advertising promise is a policy rather than a property of the system. Meta says conversations are kept away from its ad machinery, that users can opt out of having their interactions train its models, and that they can tell the agent to forget particular things. Those commitments are real, and they are also revocable, a distinction worth keeping in view given that most of Meta’s revenue comes from knowing what people want.
Why the heavy safety scaffolding?
Consumer agents have had a rough few months in public. A Meta researcher’s own experiment with an agent ended with files deleted, and booking platforms including Resy have told customers not to point automated agents at them. Sentinel is the layer standing between a useful assistant and a very fast mistake, which is why Meta built the permission gates, the isolation, and the audit trail into the product from the start.
What is Meta betting on?
The framing follows an essay Meta’s leadership published this summer about developing personal superintelligence to help people accomplish their goals and pursue their passions. The company has predicted that billions of people will have a personal agent within five years. Muse is the first product built to make that claim testable rather than rhetorical.
FAQ
What is Meta Muse?
Muse is a personal AI agent from Meta that reads email, books travel, fills in forms, negotiates on a user’s behalf, and pays for purchases with a saved card. It runs on a model called Muse Spark and is rolling out in the United States on iOS, Android, and the web, with AI glasses support to follow.
How much does Muse cost?
Muse has a free tier plus subscriptions at $20 and $100 a month. Meta says most people should be able to do what they need within the free tier. There is no advertising inside the product.
Does Muse use your data for advertising?
Meta says conversations are kept away from its ad systems, that users can opt out of having interactions train its models, and that they can tell the agent to forget particular things. A confidential version planned for later this year is meant to prevent Meta from observing what happens inside a user’s workspace. Until then, that separation is a policy rather than a built-in property of the system.
Related coverage
This article summarizes reporting from thenextweb.com.