Meta’s free Muse AI app can do more than answer questions, and you can use it through WhatsApp too

Meta has introduced Muse, a new personal AI agent designed to do more than respond to prompts. Instead of simply answering questions, Muse can take action on a user's behalf — from sending emails and booking travel to managing projects, planning long-term goals and making purchases online.

Built around Meta's Muse Spark AI model, Muse is designed to work more like a personal assistant that can actually complete tasks rather than just suggest what users should do.

What Is Meta Muse?

Muse is a personal AI agent that can understand a person's goals, create plans and carry out tasks across the apps and services they use.


Users can communicate with Muse much like they would message another person, either through the dedicated Muse app or directly through WhatsApp. Once given a task, Muse can open a browser, fill out forms, interact with websites and continue working even after the user closes the app.

For larger projects, Muse can turn a broad goal into a personalized action plan and coordinate the necessary steps over time. It can also return to the user when circumstances change or when approval is required before taking a sensitive action.

Muse Can Actually Take Action

Meta says Muse is designed to handle both everyday tasks and more ambitious goals.

It can help with tasks such as:

• Sending emails
• Booking travel
• Filling out online forms
• Making eligible purchases
• Managing projects
• Creating plans around long-term goals
• Adjusting plans when circumstances change

Muse can also remember information users have previously shared. For example, it could turn a recipe saved from Instagram into a grocery list, suggest a menu for a dinner party and remember guests' dietary restrictions when helping prepare invitations.

That persistent memory is intended to make Muse more proactive rather than requiring users to explain the same context repeatedly.

Muse Uses a Secure Virtual Computer

One of the biggest differences Meta highlights is the technology behind Muse.

The AI agent operates through Muse Secure VM, a dedicated virtual machine in the cloud that houses the agent, user data and securely stored credentials. Meta designed the system so that Muse operates inside its own isolated computing environment.

A separate security system called Sentinel controls Muse's access to the internet. According to Meta, Muse cannot perform internet actions unless Sentinel approves them, with user permission required for certain sensitive activities.

Muse also isn't supposed to directly see passwords or payment details. Credentials can be stored securely so the agent can use them without exposing the underlying information to Muse.

Users Stay in Control

Privacy and user control are central to Meta's design for Muse.

Users can decide which apps and services Muse can access and determine what it is allowed to do. For example, someone could allow Muse to read email without giving it permission to send messages.

Muse also asks for approval before sensitive actions such as sending an email or making a purchase. Users can view an audit trail showing what the agent has done and what it plans to do.

Meta says users can disconnect services or change permissions at any time, and they can also tell Muse to forget specific information it has remembered.

The company says Muse conversations and data stored inside the user's virtual machine aren't shared with Meta's advertising systems.

Muse Can Buy Things Online

Muse is also designed to handle online purchases.

Meta says Muse can check out using Link by Stripe, with an agent-specific wallet generating a one-time-use card so a user's actual payment card details aren't exposed to the websites where purchases are made.

Meta says Muse is the first AI agent covered by Link's purchase protections, which can include protections related to damaged or lost items, eligible price drops, returns and other qualifying purchases.

Shop Pay is expected to be added later, while 1Password support is also planned to allow Muse to use existing logins.

Meta Plans More Private AI

Meta says it will introduce Muse Confidential VM later this year. The planned system will encrypt the entire virtual machine, including user data and conversations, with an encryption key controlled by the user.

The goal is to create an environment where even Meta cannot access the contents of the user's Muse VM.

This reflects a broader challenge for personal AI agents: the more autonomy an AI receives, the more sensitive information it potentially needs access to. Giving an agent the ability to browse websites, send messages and purchase products requires considerably stronger security than a conventional chatbot that only generates text.

Where Is Meta Muse Available?

Muse is rolling out in the United States on iOS, Android and muse.ai, with support for AI glasses planned in the future.

Meta says Muse will be free for most everyday use, while subscription plans will be available for people who want additional capabilities.

The larger idea behind Muse is straightforward: instead of AI simply telling people what to do, Meta wants an AI agent that can increasingly do it for them.