Meta announced Muse on September 8: a personal AI agent it says “keeps working after people close the app.” It can open its own browser, fill out forms, send email, book travel and make purchases, and Meta describes it as turning “long-term goals into action plans” rather than answering one question at a time. It is rolling out in the U.S. on iOS, Android and muse.ai, reachable in its own app or directly inside WhatsApp, with AI glasses “coming soon.”
The one guardrail Meta states plainly: Muse “checks with the person before sensitive actions like sending an email or making a purchase.”
What is new is not the chat. It is the persistence. A user can hand an agent a goal that takes weeks — find me a place, compare it over time, act when the moment comes — and the agent keeps the goal alive while the person does something else.
For the valley: a visitor deciding where to stay, a family comparing schools, or a company weighing a second office may now be represented by an agent that reads a business’s website, its listings and its reviews on their behalf, over months, before a human ever sees the page.