Muse

A personal AI experience that helps carry a task through.

Meet Muse.

Muse is Meta’s personal AI experience: you give it a goal and it uses available information, software and connected services to help complete the task. Muse is also the name of the wider family of models and tools. Using the personal product differs from building with a model API—an interface for software to call a model—or setting up Glimmer yourself.

What Muse is good at

Keep a goal moving

Muse can continue background work while the app is closed and return when there is progress or a decision to make.

Use context you control

Connect selected services and review or edit saved memory as your needs change.

Create a useful output

Artifacts are interactive outputs such as an itinerary or dashboard that you can review and use.

Review the work

Check the activity history and approved permissions, and decide on consequential actions.

Turn your trip bookings into a clear itinerary.

Organize the information you provide, flag conflicts, and review any proposed changes.

Illustrative scenario

Your workspace

1
Request
Provide your trip bookings and notes, and ask for an itinerary with any conflicts flagged.
2
Prepare
Muse reads the permitted documents, organizes dates and locations, and identifies gaps or overlaps.
3
Approve
In this scenario, ask it to pause before changing a booking or sending a message. Available connections and permission settings determine what can proceed.
4
Result
Review the proposed itinerary and unresolved questions against your original bookings.

Illustrative workflow. Actual access depends on configuration.

Where it fits in Muse

Muse is the connected experience. Models provide intelligence, tools make work possible, and permissions set the boundaries around actions.

Explore further

This guide does not yet include a dedicated example for this product. Start with a task to see how the workflow could fit your needs.

Explore personal tasks

Technical architecture

The model reasons about the task; the surrounding software supplies tools, context, permissions and an execution environment. Available connections and approvals determine which actions can be taken. Goals can continue while the app is closed. You can review activity and edit saved memory; interactive outputs such as itineraries and dashboards are called Artifacts.

Your request passes to Muse / Spark for reasoning, context and planning, then browser, file and code tools, and result and review. Connected services require permissions. Review feeds back into planning. — Conceptual flow. The model proposes work; software executes tools and enforces configured controls. Steps may repeat.

Specifications and limitations

Role: Personal AI agent experience
Access — checked September 29, 2026: Meta describes availability in the US and Canada. Check current account and regional access through the official footer resources.
Consumer pricing — checked September 29, 2026: Free for most uses, with optional subscriptions for higher usage. Consumer plans are separate from Muse Code subscriptions and metered API billing.
Workspace: Secure VM and connected tools
Control: Permissions and approvals

Keep in mind

Capabilities depend on the tools and services available to the account.
Important actions and final results need human review.
Safety controls reduce risk but cannot eliminate every problem.