What's this about?
Muse is Meta's always-on personal agent. It gets a persistent Linux VM and browser, connects to email, calendar and other apps, works after the client closes, builds tools when needed, and asks for approval before sending messages or making purchases.
What exactly makes it an “agent” rather than a chatbot?
A chatbot answers and waits. An agent plans several steps, operates websites and apps, keeps state and continues in the background. Muse is closer to giving a remote assistant its own computer than opening another chat tab.
What's the community saying?
HN 334 points and 336 comments.
- The largest objection was trust: many users would not give Meta email, calendar, browsing and purchase access regardless of technical protections.
- Commenters saw the shopping-heavy demo as evidence that the agent may optimise commerce and advertising rather than the user's interests.
- Others called it the first polished mass-market version of an OpenClaw-style personal agent, with support that non-technical users can actually use.
- Comparisons with Facebook M — the human-assisted Messenger agent shut down in 2018 — highlighted how long Meta has pursued this product shape.