Meta launched a new personal AI agent called Muse, designed to handle tasks for users rather than just answering questions. Muse can make purchases, book travel, send emails, fill out forms, and negotiate on behalf of users.
Once tasked, Muse creates a personalized plan and works independently, even after the app is closed. It will notify users for approval before carrying out sensitive actions, such as sending emails or making purchases. Muse uses Stripe’s Link payment service for online purchases, ensuring users’ payment details remain private by generating single-use virtual cards.
The AI remembers past information to make unsolicited suggestions. Operating through a Secure VM, Muse interacts with websites and services. A separate AI agent, Sentinel, governs Muse’s internet access, deciding if actions can proceed, be blocked, or need user approval.
Users can select the services Muse can access and define its actions. Meta ensures Muse cannot view passwords or payment methods and keeps an audit trail of its activities. Conversations and data in Muse’s virtual machine are not shared with Meta’s advertising systems, and users can opt out of having their data used for AI training.
Initially available to users aged 18 and older in the U.S., Muse can be accessed via a dedicated app or WhatsApp, with plans to support Meta’s AI glasses. While most uses are free, subscription plans for heavier usage cost $20 and $100 per month.
