Meta has launched Muse in the U.S., introducing a personal AI agent that can carry out tasks on behalf of users, including sending emails, booking travel, filling out forms and making purchases.
Muse is now rolling out across iOS, Android and muse.ai in the U.S. Meta said the service will be free for most use cases, while subscription plans will be available for users who want access to additional capabilities. The company also plans to bring Muse to its AI glasses.
The launch marks Meta’s push to turn its AI assistant into a system that can do more than answer questions. Muse is designed to take action for users and continue working even after they close the app.
Users can give Muse a task, and the agent can develop a plan around it, coordinate the resources it needs and adjust its actions when circumstances change. When it reaches a step that requires the user’s approval, such as sending an email or completing a purchase, it can return to the user for confirmation.
READ: Meta to develop an AI agent platform Hatch for task completion (August 25, 2026)
“People just tell Muse what needs to get done, and it takes action,” Meta said in a statement.
Meta has also built security controls around Muse as it begins rolling out the agent to U.S. users. The system runs on Muse Secure VM, a dedicated virtual machine that contains the agent and the user’s data. Muse has its own browser and can interact with services that users connect to it.
A separate Sentinel agent operates within the same environment and controls Muse’s access to the internet. “Nothing Muse does reaches the internet unless the Sentinel approves it,” Meta said.
Muse does not directly see users’ passwords or payment information. Those credentials are stored separately in secure storage, allowing the agent to use them without exposing the underlying information.
The agent also requires approval for sensitive actions and keeps an audit trail showing what it has done and what it plans to do. Users can decide which services Muse can access and set individual permissions, including whether the agent can only read their emails or can also send messages on their behalf.
Muse can retain information from conversations and use that information when handling future tasks. Meta said users can ask the agent to forget specific information and can opt out of having their interactions used to train its AI models.
READ: Meta launches Pocket, an AI app for creating games with text prompts (July 3, 2026)
For purchases, Muse can use Link, Stripe’s payment service, through a wallet designed for AI agents. The wallet generates a one-time-use card for transactions. Meta said Muse is the first AI agent covered by Link’s purchase protections.
Support for Shop Pay and 1Password is also planned.
Meta said it plans to introduce Muse Confidential VM later this year. The feature will encrypt the entire virtual machine, including user data and conversations, with a key held only by the user.
Muse is powered by Muse Spark, which Meta described as its most capable model to date and specifically designed for agentic tasks.
For now, the launch is focused on the U.S., where Meta is making Muse available to users across its mobile platforms and the web as it expands its push toward AI agents that can take real-world actions rather than simply generate responses.