Muse gives a personal AI agent its own computer

Meta’s Muse launch caught my attention. It looks promising: a personal AI agent you can message through its own app or WhatsApp, with a dedicated computer where it can keep working on your tasks.

That computer is a persistent, isolated Linux environment with a browser. Meta says Muse can work across connected apps, fill out forms, and continue after you close the app. You give it a task or a longer-term goal, and it comes back with progress, suggestions, or a request for approval.

The controls are part of the product. You choose which apps it can access, review its activity, and approve sensitive actions such as sending an email or making a purchase. Meta says passwords are stored separately from the main agent, which can use them without seeing them.

I haven’t tried it yet. What interests me is the combination: a familiar messaging interface, a computer that stays available, and a way to check what the agent is doing on your behalf.