Earlier this month, Meta launched Muse across iOS, Android and the web, where it quickly climbed the US App Store charts. The premise was already ambitious: an assistant designed not merely to answer questions, but to work across the services people use every day.¹

Now Meta has taken that pitch to the desktop. Muse is available on Mac, where it can interact with files, messages, calendars, notes and mail inside their native apps. Meta says users opt in to the access they grant, and that the agent must seek approval before carrying out sensitive actions.¹

Mark Zuckerberg framed the release as a rapid expansion of that vision. “Muse for Mac is out today! It works across apps, files, calendar, notes, and messages on your computer,” he wrote, adding: “You control what it can access. The team is shipping fast.”Mark Zuckerberg (finkd)²

The Mac version adds practical tasks to the sales pitch: Meta says Muse can organize files, fill out forms and pull information from Messages, Calendar and Notes.³ The company has also pushed new connectors for Granola meeting notes and Notion documents, signaling that Muse’s usefulness will depend on how deeply it can plug into the work tools already on a user’s screen.²

That is where the stakes widen. Consumer-facing AI agents are increasingly being cast as a potential successor to today’s app-first computing model—putting pressure on software subscriptions, mobile apps and, eventually, the devices that host them. Meta’s safeguards address the obvious concern over handing an agent access to personal data and actions. But the competitive question is more immediate: whether Muse can become a trusted operator before rivals do.¹