Meta Launches Muse, an AI Agent That Can Take Action

Meta has launched Muse, a consumer AI agent designed to do more than answer questions. The new service can carry out digital tasks on a user’s behalf, including sending emails, booking travel, filling out forms, organizing plans and making purchases.

Muse reaches beyond a standard chatbot

The agent is initially available to users in the United States through dedicated iOS and Android apps, the Muse website and conversations in WhatsApp. Meta says support for its AI glasses is coming later. People can connect services individually, including email, calendars, shopping, payments and other everyday apps.

When a service offers an API, Muse can use that connection after the user supplies permission and credentials. It can also work through a browser when an API is unavailable. That broader access is what allows an agent to complete a multistep job, but it also gives users more to consider before handing over control.

Muse has a free tier, while heavier use requires a subscription. TechCrunch reports two launch plans: Power at $20 per month and Maximum at $100 per month. For purchases, the service uses Stripe’s Link infrastructure, which can issue a single-use card number rather than passing a customer’s regular card details across multiple sites.

Privacy is central to Meta’s pitch

Meta says each user’s activity runs inside a dedicated virtual machine called Muse Secure VM. A separate Sentinel agent is designed to monitor activity while remaining isolated from the system component that takes action. The company also says Muse does not expose passwords or payment methods to the agent and does not share conversations or connected data with its advertising systems.

Those are company claims that will need independent scrutiny as researchers and customers test the product. The practical appeal of an agent that continues working after a user leaves the app is easy to see. So is the concern: the same integrations that make Muse useful can provide access to some of a person’s most sensitive daily information.

Meta is entering a growing race to turn AI assistants into operators rather than advisers. Muse’s success will depend not only on whether it completes tasks reliably, but on whether users believe its permissions, safeguards and spending controls are understandable. For now, the launch establishes the product’s basic bargain: more automation in exchange for a much higher level of trust.