Meta Launches Muse AI Assistant for Shopping, Email and Trip Planning

Meta's first productivity-focused AI agent takes on everyday chores like emails and travel booking, but early hands-on time suggests the results can feel unsettling.
Meta has released Muse, its first AI assistant built for productivity rather than social feeds or chat. The company says the agent can handle tasks such as online shopping, drafting emails and planning trips, positioning it as a direct answer to similar tools from Google and OpenAI. It is Meta's clearest move yet into the crowded market for AI that does work on a user's behalf.
The pitch is straightforward: hand off the small stuff. Meta describes Muse as a way to "take the busywork off your plate," a phrase that puts the product in the same category as Google's Gemini extensions and OpenAI's Operator, both of which chase the same promise of an agent that clicks, types and books for you. Unlike those rivals, Muse arrives attached to Meta's existing accounts, which gives it instant reach across Facebook, Instagram and WhatsApp but also raises questions about how much of that data the agent can touch.
A first hands-on run through the tool, reported by The Verge, found that Muse does complete some assignments. It can move through shopping flows and email drafting well enough to be useful, though the experience left the reviewer uneasy rather than impressed. That reaction matters because trust, not raw capability, is shaping whether people let an agent act with their credentials and payment details.
Meta has not disclosed pricing or a full rollout schedule for Muse, and it has said little about the safeguards around an agent that reads messages and makes purchases. Competing assistants have already stumbled on exactly this ground, with early agent products misreading pages, stalling mid-task or asking for confirmation at the wrong moment. How Meta handles permissions and recovery from mistakes will likely decide whether Muse is treated as a convenience or a liability.
The launch also signals where Meta wants its AI spending to go. After pouring billions into models and data centers, the company needs consumer products that justify the outlay, and productivity agents are one of the few areas where users may eventually pay. Muse gives Meta a foothold in that market while its Llama models continue to compete on the developer side.
What to watch next is whether Meta expands Muse beyond an early test group and how it explains the agent's access to personal accounts. The company has not said whether Muse will run on its own app or live inside existing ones, and it has not addressed what happens when the agent gets a booking or a purchase wrong. Those answers will shape whether users see Muse as help or as a stranger with their passwords.