Meta has launched Muse, a personal AI agent designed to complete tasks, manage projects and turn long-term goals into practical action plans. Powered by Muse Spark, the company’s most capable model for agentic work, Muse operates through its dedicated app and WhatsApp.
Unlike traditional chatbots, Muse can take actions on behalf of users. It can send emails, book travel, open websites, complete forms and negotiate tasks, while continuing to work even after users close the app.
Muse takes action for users
For larger projects, Muse can create personalized plans, organize time and resources and continue working until it needs more information or approval. It can also remember useful details from conversations and apply them to future tasks.
For example, Meta says Muse could use a saved Instagram recipe to create a shopping list, plan a dinner menu and remember guests’ dietary preferences.
Meta focuses on security and privacy
Muse operates inside a dedicated cloud computer called Muse Secure VM, which isolates the agent, connected data and credentials. A separate Sentinel agent monitors Muse’s activity and controls its internet access.
Users must approve sensitive actions, including emails and purchases. Meta says Muse cannot directly access passwords or payment details, while users control service permissions and can disconnect accounts or delete stored information.
Muse supports Link by Stripe for purchases through one-time payment cards. Meta also plans support for Shop Pay and 1Password, alongside Muse Confidential VM, which will encrypt stored data using a user-controlled key.
Muse is rolling out across the US on iOS, Android and muse.ai. Meta plans AI glasses support later, while most users will receive free access with paid plans offering higher limits.
