Meta Claims Millions Shop With Muse. Executives See 'Hundreds' of Orders

Meta says more than 10 million people in the US and Canada have activated Muse, its agentic AI tool for automating online tasks — yet DoorDash, one of its most natural partners, says orders from Muse and similar agents number only in the "hundreds." That gap between Meta's usage figures and the trickle of commerce companies actually report defines the current state of agentic shopping (AI agents that browse, compare, and buy on a user's behalf).

Meta has claimed Muse is becoming a major force in shopping, with millions of people using it to make purchases averaging more than $200. Wall Street fears the opposite trajectory: analysts worry that proactive shopping assistants will erode brand loyalty, ignite price wars, and squeeze corporate profits. The stakes are substantial — if agents mediate purchases at scale, they reshape how brands reach customers and who controls the transaction.

But executives across roughly 50 companies — spanning travel, internet providers, food, and self-storage — have shrugged off immediate disruption, both on recent earnings calls and in conversations with WIRED. Many insisted they would mostly benefit from any real widespread adoption of agents. The claims of disruption, so far, outrun the receipts.

What did companies actually report?

The numbers corporate leaders volunteered were uniformly small:

• DoorDash: orders from Muse and other agents are in the "hundreds."
• T-Mobile: "We haven't seen anything of significance" from agents.
• Hilton: bookings from agents are "insignificant."
• Ally Financial: agent activity is "quite low."
• eBay: agent-driven activity is "very, very small."
• Life Time Group: "seen zero" cancellations from agents.
• Eternal (India's delivery app giant): agentic commerce is "close to zero."

The fitness chain's data point matters because financial experts have suspected users would ask Muse to cancel unused subscriptions and memberships. Life Time, Public Storage, and Hartford Insurance Group all reported essentially no such activity. The subscription-cancellation threat, at least for now, exists more in analyst models than in customer behavior.

These figures leave an open question: if 10 million people activated Muse, where is the money going? Zuckerberg said on Meta's earnings call this month that over 25 percent of Muse users deploy it for shopping — more than any other task and above the company's own expectations. Usage, in other words, is real and measurable; conversion into completed transactions that companies can see is not.

Meta declined to comment beyond the earnings call remarks. The company pointed to partnerships with 10 retailers, including Walmart, Best Buy, and Gap, intended to simplify shopping with Muse. A caveat: many of those retailers last reported earnings before Muse launched and have not shared data about it. The partner list is longer than the evidence.

Why are companies preparing anyway?

Across the board, executives acknowledged agents are at an early stage. But public anxiety about AI has pushed companies to extract lessons from the outset rather than wait.

Chipotle is already optimizing its website for agents. Chief digital officer Arlie Sisson told reporters this week that the chain discovered agents struggling to complete purchases because of hiccups in applying loyalty discounts or ordering after-hours. Muse's debut increased agentic orders by a single-digit percentage, she said. Chipotle is now tidying up its website code to help agents better navigate and satisfy the hungry humans behind them.

Similar work is underway at DoorDash — even though CEO and Meta board member Tony Xu said the present hype about agents might be "delusional." That is a notable formulation from someone who sits on Meta's own board: the technology's long-term promise and its present-day noise are separable judgments.

Costco and Extra Space Storage offer a parallel case. Both said sales originating from AI chatbots such as Gemini and ChatGPT are surging as they optimize their online content for those tools, and they plan to do the same for AI agents. Their bet: agent-readiness follows the same playbook as search-readiness, and the traffic already justifies the work.

What will Muse cost businesses?

Reliance on agents could carry a price. Zuckerberg said Meta plans to charge businesses a fee for transactions facilitated by Muse, and he predicted this revenue would eventually top the company's massive advertising business. That is a striking claim: Meta's ad business generated the overwhelming majority of its income, and transaction fees overtaking it would make Meta one of the world's dominant commerce tollbooths.

Not everyone is ready to sign. InterContinental Hotels CEO Elie Maalouf is taking a cautious approach in ongoing talks with Meta. "We haven't come to any conclusions on what an eventual collaboration could look like, if any," he said. The "if any" carries the skepticism the rest of his industry has expressed in numbers.

What about security and errant orders?

Security issues and errant orders remain top of mind. This week, Bank of America became the latest big bank to warn customers about giving login information to AI assistants. "If you do share your credentials, keep in mind that you're responsible for activity resulting from that access," chief client care officer Ashley Ross wrote in an email. The warning shifts liability onto customers — a position banks have taken as agents request access to sensitive accounts.

There is also the possibility that agents encourage comparison shopping in a way that hurts profits, since an agent optimizing purely on price strips away the storefront, the branding, and the impulse buy. Executives at many companies dismissed the notion. They suggested product quality, customer service, and brand trust would still factor into purchase decisions despite less human involvement. That claim is untested at scale; the current data is too thin to confirm or refute it.

Are the industry's positions shifting?

Stances on agents are evolving quickly — faster, in some cases, than the underlying usage numbers.

Amazon blocked Muse last month, contending that users would prefer its own Alexa shopping agent. This week, the retail giant acknowledged it anticipates working with outside agents. A month reversed a platform position.

Insurance marketplace Insurify also blocked Muse, but changed its mind this week after introducing technology to ensure agents could present accurate offers. Its reversal came with a control mechanism attached — a pattern other companies are likely to study as they weigh access against accuracy.

The pattern across the roughly 50 companies is consistent: dismiss the current volume, prepare for the future traffic, and negotiate hard over who pays when the volume arrives. Zuckerberg's claim that agent transaction fees could exceed advertising revenue sets the ceiling for what Meta believes this becomes. The executives counting hundreds of orders set the floor for what it is today. Both cannot describe the same quarter — and the distance between them is where the next year of agentic commerce will be decided.