Meta Launches Muse Personal AI Agent to Handle Everyday Tasks Autonomously
Meta has launched Muse, a new personal AI agent designed to do more than answer questions. The company says Muse can take action on behalf of users, manage everyday tasks, work toward longer-term goals and continue working in the background.
The new AI agent is powered by Muse Spark, Meta’s latest model built specifically for real-world agentic work. Users can interact with Muse through a dedicated app or WhatsApp, with Meta aiming to make the technology accessible without requiring technical knowledge.
Muse can perform tasks such as sending emails, booking travel, filling out online forms and helping users organize plans. It can also coordinate multiple steps of a task and return to the user when an approval is required, such as before sending an email or completing a purchase.
One of Meta’s demonstrations highlights how Muse can connect information across its ecosystem. For example, the agent can identify a recipe from a previously saved Instagram Reel, work out the required ingredients and prepare a grocery list. It can also remember relevant preferences when helping users organize events.
The company has also designed Muse to operate inside a dedicated Muse Secure VM, a virtual computer intended to isolate the agent and users’ connected data. A separate safety system called Sentinel monitors Muse’s actions and can require user approval for sensitive activities.
Privacy and security are likely to be among the biggest tests for Muse. Because the agent can interact with email, shopping, payments and other personal services, mistakes or unauthorized actions could have consequences beyond those associated with a conventional chatbot. Reuters reported that internal testing had uncovered reliability and security concerns before the launch.
Meta says users remain in control of which applications Muse can access and can revoke permissions. The company also says users can opt out of having their interactions used to train Meta’s AI models. An encrypted Muse Confidential VM, designed so that even Meta cannot access the contents, is planned for later this year.
Muse is initially rolling out in the United States on iOS, Android and the Muse website, with integration planned for Meta’s AI glasses. Meta says a basic version will be free, while subscription options will be available for users who want greater usage.
The launch marks a major step in Meta CEO Mark Zuckerberg’s broader push toward what the company calls “personal superintelligence”—AI systems that do not simply provide information but increasingly carry out tasks and make progress toward users’ goals.
