META has formally launched Muse, its end-to-end personal AI agent designed to act autonomously on a user’s behalf. Unlike prior conversational bots, Muse comes with its own dedicated app and web interface and operates without requiring technical know‑how. Users simply state a goal, and Muse breaks it into steps, organises time and resources, and advances each stage in the cloud.
Demonstrations cited in the piece include tasks such as online price comparison, drafting and sending emails, and planning cross‑platform travel itineraries, with Muse able to link across Meta’s own Facebook and Instagram data and, if allowed, bridge third‑party services like Google Workspace, Ticketmaster, OpenTable, Spotify and Apple Health.
Payments are processed via Link (Stripe), using one‑time‑use virtual cards to protect real card details, and Muse is described as the first AI agent covered by Link’s purchase protections.
To address security and privacy risks of handing over control of accounts and finances, Meta has built Muse Secure VM, a dedicated isolated environment where each user’s agent runs separately, with a Sentinel agent supervising outbound actions. Muse does not have visibility into passwords or payment methods, keeps sensitive credentials in secure storage, and requires explicit user confirmation for sensitive actions, while providing a complete audit trail of planned and performed activities.
A forthcoming Muse Confidential VM would encrypt the entire VM and data with a user‑held key, ensuring Meta cannot access the content. Pricing is described as freemium, with core features free and heavier use available on a paid tier, and initial availability in the United States across iOS, Android and the web, with AI glasses support planned.