Meta has launched Muse, a personal AI agent that can take action for users instead of only answering questions. The Meta Muse AI agent gives the company a new consumer service to test across shopping, travel, email and long-term planning. It also shows that the company sees trust and payment controls as central to making agentic AI useful in everyday life.

The Associated Press reported that Muse is available to people 18 and older in the United States. Users can communicate with it in a separate Muse app or through WhatsApp. The agent can handle simple tasks such as sending an email or booking travel, while Meta says it can also help turn larger goals into plans and carry out work over time.

Meta’s official newsroom announcement says Muse is available on iOS, Android and the web through muse.ai. The company said the service is free for most everyday use, with subscription plans for people who want to do more. Meta also said Muse is expected to come to its AI glasses, extending the service beyond phones and computers.

The product is built around a dedicated environment called Muse Secure VM. Meta says the virtual machine houses the agent, the user’s data and connected credentials. The company says a separate Sentinel agent controls whether Muse can reach the internet and asks the user for permission when needed. AP also reported that Meta is stressing safety and privacy as it introduces the service.

That control layer is important because Muse is designed to open a browser, fill out forms and negotiate on a user’s behalf. Meta says people can decide which apps Muse can access and whether it can only read information or also act for them. The company also says users receive an audit trail of what the agent has done and what it plans to do.

Payments are another major part of the rollout. Meta said Muse can check out using Link, the payment service built by Stripe. Link can create a one-time-use card so a user’s real card details are hidden from the merchant. Meta says eligible purchases can receive protections for damaged or lost items, price drops, no-fee returns and a return guarantee, while Shop Pay and 1Password support are planned for a later stage.

The payment design gives Meta a way to reduce one of the biggest barriers to automated shopping: the fear that an agent will spend money without enough oversight. Meta says Muse will ask for approval before sensitive actions such as sending an email or making a purchase. Whether those prompts feel helpful or disruptive will influence how often people allow the agent to work without close supervision.

For Meta, the service also creates a new consumer-business model around recurring AI use. A free tier can encourage people to try the product, while subscriptions may support heavier users who want more tasks completed. The company will need to show that Muse saves enough time to justify continued use, especially when customers are asked to connect email, shopping and payment services.

AP described AI agents as a step beyond chatbots because they can use software and complete a process rather than simply return information. Muse is entering that market with Meta’s large consumer distribution, but adoption will depend on reliability, privacy and the quality of its approvals. An agent that pauses at the right moments can build confidence; one that makes an unexpected change can quickly lose it.

Meta said a more private Confidential VM is planned for later this year, with encryption keys controlled by the user. The Muse rollout therefore starts with convenience, but its longer-term business test is trust. Usage, subscription conversion and repeat purchases will show whether consumers treat a personal AI agent as a useful service or just another assistant they try once.