Meta’s Muse can now do more than answer a question. It can put together an order.
On October 3, Meta added a shopping function to Muse, its AI assistant. A user tells Muse what they want, a brand, a price ceiling, a stock requirement, and the assistant browses retail sites, selects the items and stages the order for confirmation. The payment remains the user’s to make. Meta announced partnerships with Walmart, Gap, Sephora, Expedia, Wayfair and Best Buy, among others.
The arrangement is a study in where Meta wants to draw the line. The assistant does the searching and the choosing, but it does not spend the money. Meta has kept the final click in the user’s hands, a design choice that reads as a concession to the anxiety that has trailed every announcement about AI agents that can act on a person’s behalf.
Shopping, a Meta spokesperson said, has become one of Muse’s most-used features. The statement suggests that people are already leaning on the assistant for tasks that end in a transaction, and that Meta sees the habit as a durable one. The company is betting that commerce, once it enters the assistant, will make Muse a destination rather than a conversation partner.
For the retailers, the pitch is straightforward: a new way to be found when a customer asks an AI for help instead of typing into a search box. Walmart, Gap, Sephora and the rest are agreeing to let Muse browse their catalogs and build carts, a bet that AI-mediated shopping will grow without cutting them out of the relationship with the customer. The risk, which none of them has resolved, is that the assistant becomes the customer’s real point of contact and the retailer is reduced to a warehouse that fills the order.
The worry that hovers over the whole exercise is price. If an AI agent knows what a shopper is willing to pay, the fear runs, it can be used to raise prices for the people most likely to say yes. Meta has not answered that concern; it has only acknowledged that consumers raised it. Whether AI-driven dynamic pricing pushes costs up or down is likely to determine how far the shopping feature spreads.
The feature is also a step in a longer campaign. Meta has been pushing Muse across its apps for months, positioning the assistant as the layer that connects a user to the rest of the internet. Commerce is the most direct test of that ambition, because it is the one activity where a successful assistant produces money rather than conversation. If Muse can reliably save shoppers time without spending their money unwisely, Meta will have a case that the assistant is worth the attention it has poured into it.
The move puts Meta into direct competition with the companies that already own online shopping. Amazon has its own AI shopping assistant, and Google has spent years turning search into a place where a query ends in a purchase. Meta’s advantage is reach: its apps hold billions of users who already message, scroll and share inside them. If even a fraction of those users start their shopping inside Muse, the assistant becomes one of the largest storefronts on the internet without Meta having to hold a single item in inventory.
That reach is also what makes the pricing worry acute. Meta can see more about what a user wants, and what they are willing to pay, than most retailers can. Critics of AI pricing have argued that the combination, a machine that knows your budget and a merchant that wants your money, is an invitation to charge different people different prices for the same item. The concern has surfaced in congressional hearings and consumer surveys, and it will not be settled by a product announcement.
The partners have their own reasons to proceed anyway. Retailers have watched Amazon and Google capture the opening moments of online shopping for two decades, and they have little appetite to sit out the next shift. Joining Muse early, on terms they can still shape, looks safer than waiting until the assistant is established and the terms are set without them.
What Meta has not said is how it will make money from the feature itself. The partnerships could eventually carry referral fees or commerce commissions, but the company has not disclosed terms. For now, the shopping function looks like a way to deepen the habit of using Muse, with the economics to be worked out once the habit is set.
The test will come at checkout. If shoppers complete more orders through Muse than they abandon, the feature will expand, and the line between assistant and agent will keep moving. If they hesitate at the moment of payment, Meta will have learned something about how much control people are willing to hand over, and how much they insist on keeping. The cart, for now, belongs to the machine. The wallet still belongs to the user.
