Meta rolls out customizable Muse Spaces for persistent app experiences

By Elena Vasquez

Meta’s current Muse model scores 47% on its Personal Superintelligence Index, up from 38% for the previous model, the company reported.

On November 6, 2026, Meta announced that Spaces were available everywhere Muse was available. Users can describe an idea to Muse and keep refining it into a saved, customized experience.

Meta says Spaces can support routines, goals, interests, practical work, or entertainment. Its introductory examples include a holiday gift tracker, a daily word game, and a planning board for a do-it-yourself project.

Meta’s examples put the feature across shopping, education, personal planning, and small-business work. The common thread is a reusable interface that keeps working after the initial conversation ends.

What can a Muse Space do?

Creation starts with a plain-language description supplied to Muse. Meta treats the first result as a starting point rather than a finished product.

Users can ask Muse to add missing elements, try a different appearance, or change how the space functions. Each revision remains inside the app, ready for another visit.

Meta says that persistence is the point. A Space can gradually take shape as its purpose becomes clearer without losing the work already done.

The examples show how broad that purpose can be. A daily word game turns a simple prompt into an entertainment tool, while the do-it-yourself board supports a practical project from planning onward.

Users also control where a Space appears. They can place it in the sidebar or main tabs, hide spaces they do not currently need, and schedule when selected spaces appear.

One Meta example keeps a weekly planner prominent during the workweek, then replaces it with weekend plans. The feature therefore manages both a Space’s contents and its visibility inside Muse.

How do connected Spaces stay current?\n

Connecting apps lets a Space update in the background, according to Meta. New concert dates, price changes, and savings progress can appear without the user requesting each refresh.

People can also work directly inside a Space. They can check off items, add notes, and make edits, and Meta says that work remains when they return.

Together, those controls combine conversational creation, hands-on interaction, and changing information from connected services. A Space can start as a prompt, become a working surface, and keep receiving fresh data.

Meta’s shopping example combines holiday and birthday gift ideas in one view. It can pull wishlists from Instagram saves, Marketplace listings, and buyer messages into the same hub.

The school dashboard example searches school emails for relevant events and adds them to a calendar. Meta says this reduces the need to assemble the schedule manually.

Other Meta examples cover music releases, upcoming shows, wedding RSVPs, vendor invoices, trip reservations, and daily itineraries. The music tracker follows releases and shows, while the wedding and trip planners collect records people would otherwise scatter across services.

For business users, Meta suggests an order tracker that brings together new orders, customer messages, and items needing shipment. A freelance client hub can hold projects, deadlines, and invoices in one place.

These examples matter because they move Muse beyond a question-and-answer tool and into recurring administrative work. Meta’s release makes persistence, scheduling, and background data part of the product rather than leaving each task inside a fresh chat.

What changed in Meta’s models?

Meta AI Research introduced specialized versions of the company’s existing models for people using Muse and Meta AI. Meta says the versions were trained using Muse Wonder.

The specialized models include Active Memory, which Meta intends to retain relevant earlier information. They also distinguish information that is temporary, permanent, current, or stale.

They include Active Focus, which Meta says is intended to keep work moving toward a goal when unexpected obstacles arise. Its proposed responses include trying harder, recovering, or finding alternatives.

Meta reported four measurements for the current Muse model:

• 47% on the Personal Superintelligence Index, compared with 38% for the previous model.
• Correct use of critical context in 90% of Active Memory tasks, compared with 76% previously.
• A 2.4x reduction in forgetting critical information needed to help users achieve their goals.
• A 67% improvement in completing everyday tasks when unexpected issues arise.

Meta reports the figures through its own announced evaluations. The 90% result measures whether the model used critical context correctly, while the 2.4x figure covers its reduction in forgetting information needed for a user’s goals.

The models are designed for several forms of ongoing work. Meta says they can handle multitasking, decide when to pursue goals independently or collaborate live, and adapt to steering and clarifications during running tasks.

For Meta AI, Meta says the corresponding changes should break complex problems into steps, produce sharper follow-up questions, and accommodate changes of direction during a conversation.

That makes context retention part of the interface itself. Meta links the model work to Spaces, where earlier information can shape later actions instead of ending with one response.

What remains unfinished?

Sharing and collaborative Spaces remain under exploration, Meta said. The immediate release centers on individual customization rather than multiplayer or shared Spaces.

The broader change combines a reusable interface, connected information that stays current, and models intended to retain context and pursue goals. Meta describes continuity across visits as central to whether Spaces become durable parts of a user’s routine.

For users, the practical test is whether background updates and retained context feel dependable enough to replace a collection of separate trackers. For Meta, the release tests whether specialized models can make that continuity work across both Muse and Meta AI.

The immediate release leaves collaboration unfinished. Meta’s next challenge is turning its reported 90% context-use rate and 2.4x reduction in forgetting into dependable daily behavior, with shared Spaces still under exploration.