Meta Platforms launched Muse on Tuesday, Sept. 8, a personal artificial intelligence agent for U.S. adults that can send emails, book travel, fill out forms and make purchases. The launch moves Meta beyond chatbots and into software that can act across other services on a consumer’s behalf.
Muse is rolling out on iOS, Android and muse.ai, and people can also message it through WhatsApp, Meta said. The agent is powered by Meta’s Muse Spark model and is designed to keep working after a user closes the app, returning when a task changes or requires approval.
Meta is building an agent around everyday transactions
Meta is positioning Muse as a general-purpose assistant rather than a search or writing tool. The company says a user can give it a long-term goal, such as planning a project or managing a training schedule, and let the agent coordinate steps over time. For sensitive actions, including sending a message or completing a purchase, Muse is designed to ask the user for permission.
Payments are part of the product from launch. Muse can check out with Stripe’s Link, which uses one-time card credentials and extends purchase protections to eligible agent transactions. Meta said Shop Pay and 1Password support are planned. Those integrations give the agent a path from recommendation to transaction, not merely from question to answer.
Security architecture is central to the pitch
Meta says each Muse runs in a dedicated cloud computer called Muse Secure VM. A separate Sentinel agent reviews outbound actions before they reach the internet and can require user approval. Credentials and payment details are stored separately so Muse can use them without displaying them, according to the company.
Users can choose which services Muse connects to, limit what it may do, review an audit trail and disconnect access. Meta also says people may opt out of having their interactions used to train its AI models, and that conversations and VM data are not shared with its advertising systems.
The Associated Press independently confirmed the U.S. launch and adult-only availability. TechCrunch highlighted the central commercial question: whether consumers will grant Meta enough access to make the agent useful. The security and privacy architecture described at launch remains a company claim that has not yet been independently tested at scale.
Why it matters for commerce and customer experience
If consumers allow agents to plan, compare and buy, the interface between companies and customers begins to shift away from individual websites and apps. Product data, availability, pricing, return policies and fulfillment reliability become inputs an agent must interpret before it recommends or completes a transaction.
That raises the stakes for machine-readable commerce information and consistent customer policies. It also gives Meta a chance to sit between discovery, communication and payment across WhatsApp, mobile apps and eventually AI glasses. For marketers, the emerging contest is not only for a person’s attention, but for an agent’s trust.
Investors responded quickly. Meta shares rose 6.5% Wednesday as the launch reached the market, according to an AP market report. Adoption is still unproven, however. Muse’s business significance will depend on whether people accept its access model and whether it can complete real-world tasks reliably enough to earn repeated use.