Meta’s Muse Spaces turn the chatbot into a project board that keeps updating itself

TechCrunch reports that Meta has added Spaces to its Muse agent, days after launching Muse Voice with video chat and spoken commands. Spaces live in the Muse sidebar as breakout conversations dedicated to a single goal: a vacation planner, a recipe collector, a holiday gift tracker, a wedding board or a DIY project. You describe the idea in chat and Muse builds the space. If you connect other apps, Meta says the space keeps itself current, adding concert dates, price changes or savings progress without another prompt.

Mark Zuckerberg posted that he has already built a barbecue game, an event tracker and a home-control dashboard. On the October earnings call he said Muse reached 10 million users in the U.S. and Canada less than two months after its debut.

Chat threads are poor project managers. Every new conversation means re-explaining context: the dates, the budget, the kids’ allergies. Spaces acknowledge that. A persistent room with an agent working toward a goal is closer to how people actually finish a trip plan or a wedding list: they reopen the same board. Rival agents like OpenAI’s Dots and Grokbot have competed on cartoon personality. Meta is adding structure, which is a more durable advantage.

It also fits how quickly Meta is building out the Muse stack, from the frontier Muse Wonder model (launch coverage) to the low-cost Muse Ember for agent execution (our analysis) and open-source gadget kits for building Muse hardware (details).

Background sync across third-party apps is the useful part and the risky part. A price drop or new concert listing appearing unprompted is delightful until the wrong calendar entry or bank balance shows up in a space that looks shareable. Meta’s trust record makes this sharper. When Muse launched, it asked people to connect email, calendars, payments and apps just days after an $18 billion multistate settlement over social-media harms (Will consumers trust it?).

If Spaces stay personal productivity tools, the risk is contained. If they become the default place families keep plans and payments, the agent stops being a novelty and becomes infrastructure, with the lock-in and data gravity that implies. With 10 million users, any messy failure will be visible quickly.

The better path is agents that remember the project and pull live context, rather than another empty chat box. That is what Spaces promises. The test is whether permissions stay clear as the spaces update themselves, and whether users can easily see and revoke what each one can access. More in our Meta coverage and agents coverage.