Meta has released Muse, a personal AI agent that can read your email, book travel, fill in forms, negotiate on your behalf and pay for the results with your card, so routine errands get handled while you spend attention elsewhere. It is rolling out in the United States on iOS, Android and the web, with support for Meta’s AI glasses to follow. Meta says the conversations you have with the agent are kept away from its advertising systems.<\/p>\n
The company positions Muse as something that acts rather than a chatbot that only replies. It keeps working after the app is closed, makes suggestions you did not request, remembers preferences across sessions and can turn a saved recipe reel into a shopping list.<\/p>\n
Muse is built to complete tasks end to end. It reads email, books travel, fills in forms and negotiates on a user’s behalf, then pays for the outcome. Payments run through Stripe’s Link with purchase protections, and Meta has promised support for Shop Pay and 1Password. The agent runs on a model the company calls Muse Spark, which Meta describes as its most capable model to date, built for real-world agentic work.<\/p>\n
The security design carries much of the product’s promise, and it goes further than typical launch-day assurances. Each user gets an isolated virtual machine that holds both the agent and their data, so one person’s agent cannot reach another’s. A separate system called Sentinel sits between Muse and the internet. It distinguishes read access from write access using time-limited permissions, allows low-risk tasks to proceed, and stops to ask before anything consequential happens.<\/p>\n
Meta says credentials are stored so that Muse cannot see passwords or card details, and every action the agent takes leaves an audit trail. That combination lets a user review what happened and keeps sensitive login and payment data out of the agent’s direct view.<\/p>\n
Muse offers a free tier alongside subscriptions at $20 and $100 a month. Meta says most people should be able to do what they need within the free tier, which frames the higher plan as aimed at a smaller group of heavier users.<\/p>\n
Meta says there is no advertising inside the product and that conversations are held apart from its ad machinery. Users can opt out of having their interactions train the company’s models, and they can tell the agent to forget particular things. Those are stated commitments, and they remain revocable, a distinction worth keeping in mind given that the company earns most of its revenue from knowing what people want.<\/p>\n
A confidential version is meant to arrive later this year, in which Meta itself cannot observe what happens inside the workspace. Until that ships, the separation between Muse and advertising rests on policy rather than on the architecture of the system. Meta frames the wider effort as a step toward a personal superintelligence that helps people accomplish their goals and pursue their passions, and the company has predicted that billions of people will use a personal agent within five years.<\/p>\n
The safeguards give users a check on an agent that moves quickly and spends real money. An assistant that can fill forms and complete purchases can also fill the wrong form or complete the wrong purchase at speed, using a saved card. Consumer agents have run into trouble in public: one 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 that pauses before a consequential action, which is what separates a useful assistant from a fast, expensive error.<\/p>\n
Muse is a personal AI agent from Meta that can read email, book travel, fill in forms, negotiate on your behalf and pay for results with your card. It keeps working after the app is closed, remembers preferences across sessions and can turn a saved recipe reel into a shopping list.<\/p>\n
Muse has a free tier plus subscriptions at $20 and $100 a month. It is rolling out in the United States on iOS, Android and the web, with support for Meta’s AI glasses to follow. Meta says most people should be able to do what they need on the free tier.<\/p>\n
Meta says there is no advertising in the product and that conversations are kept away from its ad systems. Users can opt out of having interactions train Meta’s models and can tell the agent to forget specific things. A confidential version, in which Meta itself cannot observe the workspace, is planned for later this year.<\/p>\n