Meta is pushing its Muse AI agents beyond conversation and into construction. The company has introduced Spaces, a feature that turns a plain-language request in the chat window into a functioning personal tool — a tracker, a planner or a project board — that lives in the sidebar of the Muse interface. Meta calls these creations “custom experiences,” and they mark a shift from chatbot to something closer to a do-it-yourself app factory driven by natural language.
- Prompt-to-tool: Users describe what they need and Muse assembles the space, whether that is a holiday gift tracker, a wedding planner, a recipe collection for a family gathering or a DIY project board.
- Background syncing: Spaces can connect to third-party apps so concert dates, price changes and savings progress refresh automatically without the user asking.
- Fast-growing base: Mark Zuckerberg said on Meta’s October earnings call that Muse agents reached 10 million users in the U.S. and Canada in under two months.
How Spaces actually works
The mechanics are deliberately simple. Each Space is a topic-focused conversation that sits in the sidebar, and clicking it reopens the thread where Muse keeps working toward a specific goal on the user’s behalf. Meta’s own framing is blunt: “Describe an idea in chat, and Muse builds a space for you, whether that’s a tracker, a planner or a project board.”
The more consequential piece is integration. Meta says that once users connect the services they already rely on, “your spaces keep themselves up to date in the background, so new concert dates, price changes, or savings progress show up without you having to ask.” That turns a static list into a living dashboard — and it is precisely where the data-privacy questions begin, since keeping a space current means feeding it calendar, shopping and financial signals.
Zuckerberg demonstrated the feature on X, writing that he had been “playing around with spaces in Muse lately” and that he had built “a BBQ game, an event tracker, and a home control dashboard.” His summary of the workflow: “Tell Muse what you need and it builds it.”
A rapid-fire release cadence
Spaces arrives barely a day after Meta unveiled Muse Voice, which lets users hold video calls with their agent and issue spoken commands. The back-to-back launches signal that Meta is racing to turn Muse into a daily habit rather than an occasional novelty, and it is doing so in a crowded field of personality-driven agents that includes OpenAI’s dots and SpaceXAI’s Grokbot.
The growth numbers give that urgency context. Fewer than two months after the agents debuted, Zuckerberg told investors that Muse had crossed 10 million users across the U.S. and Canada. Meta has also been upgrading the underlying models, with specialized versions trained using learnings from Muse Wonder now powering both Muse and Meta AI, according to the company’s research arm.
The caveats worth watching
Meta has not disclosed which countries or platforms Spaces is available on, whether it is open to every Muse user, or which third-party apps are supported. That leaves the rollout only partially specified. The company’s recent track record also counsels caution: Meta previously had to pause Muse’s conversation-sharing feature after conversation titles ended up in public DNS records, a reminder of how much personal context these agents accumulate.
There is a transparency dimension too. Meta no longer open-sources its strongest AI models, so what happens inside a Muse space is increasingly opaque to outside scrutiny. For users, the trade-off is straightforward: a space that updates itself is far more useful than a chat you forget once you close it, but that convenience depends on handing over the calendar, purchase and finance data that make the automation possible. Spaces is a strong hook for daily engagement; whether it becomes a trusted home for personal planning will depend on how carefully Meta handles the information flowing through it.