Before the launch: Meta had spent billions trying to regain ground in the AI race, while its consumer products continued to carry the residue of privacy controversies and high-profile missteps. That history matters because Muse is not another chatbot: it is designed to act inside the services people rely on every day.¹

Tuesday: Meta unveiled Muse in the US as a chat-based personal agent for iOS, Android and the web. Users can ask it to send emails, book travel, shop, fill out forms and plan projects; it can also connect with Meta apps, WhatsApp and third-party services including Gmail, Spotify, Ticketmaster and OpenTable.² Zuckerberg’s pitch was deliberately expansive: Muse “understands your goals and works 24/7 to get things done for you.”Mark Zuckerberg (finkd)³

Meta is offering a free tier, alongside $20-a-month Power and $100-a-month Maximum plans. Chief AI officer Alexandr Wang said most people should be able to do what they need on the free tier, with subscriptions helping cover compute costs for heavier users.⁴ Zuckerberg cast the launch as part of a longer drive to deliver personal superintelligence “to everyone over time.”Mark Zuckerberg (finkd)⁵

The test begins now: Meta says Muse works in a dedicated cloud virtual machine and uses a separate system, Sentinel, to decide whether actions are allowed, blocked or require approval. It says users can limit read and write permissions, opt out of model training, and eventually use a confidential workspace that even Meta cannot inspect.¹

Yet the product’s utility is inseparable from its demand for access. Muse may learn from conversations and link to calendars, payments, health, home and shopping services; critics of the broader agent trend see precisely that intimacy as the risk.⁶ Meta’s promise is a tireless assistant. Its burden is proving that convenience does not become another privacy trade-off.⁷