While Meta's previous AI products focused on answering questions, Muse is built to execute workflows across various services, such as converting social media recipes into grocery lists or managing smart home devices.
The introduction of Muse represents a major shift toward the "agentic era," where AI moves beyond conversation to take autonomous action on behalf of the consumer.
To facilitate these actions, Meta is asking users for deep access to their personal data, including calendars, health apps, and financial credentials.
To address potential security concerns, Meta claims the agent operates within a "Muse Secure VM," a dedicated virtual machine—an isolated digital environment—that reportedly prevents the AI from seeing passwords or sharing interaction data with Meta’s advertising systems.
Muse is currently available as a free service on the web, iOS, Android, and WhatsApp, with plans to integrate it into Meta’s AI glasses.
Although the base version is free, the company is introducing a subscription model for heavy users, with "Power" and "Maximum" tiers priced at $20 and $100 per month, respectively.
Users must opt into each app connection individually and can personalize the agent’s name and appearance to build a more direct relationship with the software as it learns their preferences over time.