In short
Meta says its Muse AI shopping agent has been activated by more than 10 million people in the U.S. and Canada, but most companies say the real-world impact is still minimal. Even so, merchants are already adjusting websites, policies and partnerships to prepare for AI-driven commerce.
- Meta says more than 10 million people in the U.S. and Canada have activated Muse.
- Executives at about 50 companies told WIRED current agent-driven activity is minimal or early-stage.
- Some merchants are already changing code and policies to make their sites easier for AI agents to use.
- Meta says it plans to charge businesses a fee for transactions completed through Muse.
Meta’s Muse shopping agent is getting limited real-world use, but it is already pushing companies to rethink how their websites, checkout flows and policies handle AI-driven commerce. According to reporting from WIRED, Meta says Muse has been activated by more than 10 million people in the U.S. and Canada, yet executives across about 50 companies say the current impact is still tiny or early-stage.
What Meta says Muse is doing so far
Muse is Meta’s attempt to make shopping more conversational and more automated, letting people browse, compare and complete purchases with help from an AI assistant. The company says more than 25% of Muse users have used the tool for shopping, a sign that the product is being tested in real consumer behavior rather than only in demos or lab settings.
That reach sounds large on paper. But the companies on the receiving end of those interactions describe something much smaller in day-to-day business terms. In the accounts gathered by WIRED, merchants and service providers repeatedly said agent-driven traffic is not yet moving the needle in a meaningful way.
By the numbers
| Metric | What the fact sheet says | Why it matters |
|---|---|---|
| 10 million+ | Muse activations in the U.S. and Canada | Shows the product has real distribution, even if usage is uneven |
| 25%+ | Share of Muse users who have used it for shopping | Suggests shopping is one of the early use cases |
| Roughly 50 companies | Executives discussed by WIRED across sectors | Provides a cross-industry snapshot of reaction to agentic commerce |
| Hundreds | Orders from Muse and other agents at DoorDash | Indicates some commercial activity, but not a major share |
| Single-digit percentage | Increase in agentic orders at Chipotle after Muse debuted | Shows small but measurable lift in one brand’s order flow |
| 10 retailers | Partnerships Meta points to, including Walmart, Best Buy and Gap | Signals that Meta is trying to build merchant support, not just user adoption |
Why it matters
The immediate story is not disruption on a mass scale. It is preparation for a shift that may arrive more slowly than the hype suggests. Companies that once optimized for search engines and chatbots are now starting to think about a new audience: AI agents that can take action on behalf of a customer, not just answer questions.
That matters because an agent that can compare prices, assemble a cart and place an order could change how businesses compete for attention. Brands worry that automated shopping may increase price pressure and weaken loyalty if the software is simply hunting for the best deal. The fact sheet says companies cited in the story generally do not see that threat materializing yet. Still, several are already changing systems and rules in anticipation.
The market implication is straightforward: even if traffic from Muse remains small today, merchants may have little choice but to make their sites legible to AI tools. If a growing share of purchases starts with an agent rather than a browser, the companies that are easiest for those agents to use could have an edge. That is why the current numbers matter less as a revenue story than as a signal of where ecommerce software is heading.
How are companies reacting?
The answers from businesses are mixed, but the broad pattern is cautious adaptation rather than panic. Some companies say they have seen almost nothing. Others are already cleaning up websites, adjusting back-end code or testing ways to make offers easier for agents to parse accurately.
Near-zero impact at many firms
Several companies told WIRED they have not seen much at all from AI agents. T-Mobile said it has not seen anything significant from agents. Hilton called agent-driven bookings insignificant. Ally Financial described agent activity as quite low, while eBay said it is very, very small. Life Time Group said it has seen zero subscription cancellations from agents.
Public Storage and Hartford Insurance Group reported similar near-zero effects. Eternal, which operates in India, estimated agentic commerce there is close to zero. Those comments suggest the phenomenon is still early and concentrated, not yet broad enough to reshape consumer behavior across categories.
“We haven’t seen anything of significance”
That blunt assessment from T-Mobile captures the tone of the responses from many of the companies surveyed in the reporting.
Some brands are already making changes
Not every company is dismissive. Chipotle said Muse increased agentic orders by a single-digit percentage and that it is cleaning up website code to help agents complete orders. Costco and Extra Space Storage said sales from AI chatbots are rising, and both plan to optimize for agents as well.
Those moves matter because they show businesses are no longer treating AI shopping as a novelty. Even if the volumes are modest, the operational response is getting concrete: fix code, improve compatibility, and make sure an automated assistant does not get stuck at checkout.
Some companies first blocked, then relented
Meta’s push also appears to be forcing a policy debate inside companies. Amazon initially blocked Muse, then later said it expects to work with outside agents. Insurify first blocked Muse as well, then reversed course after adding technology meant to ensure accurate offers.
That sequence is important. It suggests businesses are not just deciding whether to welcome AI assistants; they are deciding what kind of technical safeguards and business rules are needed before those assistants are allowed in at all.
What Meta is betting on
Meta has positioned Muse as part of a broader commerce strategy, not just a consumer convenience feature. The company says it has partnerships with 10 retailers, including Walmart, Best Buy and Gap. Those relationships give Meta a way to point to merchant support even as the wider industry remains cautious.
The company also has a clear monetization plan. Meta says it intends to charge businesses a fee for transactions completed through Muse. Zuckerberg has said that fee revenue could eventually exceed Meta’s advertising business. That is an ambitious claim, and one that depends on a much larger volume of transactions than the current figures appear to show.
For now, the gap between the vision and the present reality is the central tension. Meta is describing a future in which AI agents are a meaningful channel for commerce. Merchants are describing a much smaller present in which the software is still finding its footing.
Background: how we got here
Agentic AI tools are designed to do more than answer prompts. They can browse, compare and complete online tasks, including shopping and booking. That makes them different from the chatbot experiences many consumers already know. A chatbot can suggest a product; an agent can theoretically buy it.
That shift helps explain why brands have spent years optimizing for search engines and, more recently, for chatbots. The next step is learning how to be visible to software that acts on behalf of the user. If an agent reads a site differently from a human shopper, then the site’s design, code and data structure become commercial issues, not just technical ones.
Companies also have a reason to be wary. An agent that prioritizes price and speed may reduce the influence of branding, loyalty programs and merchandising. The fact sheet says brand owners worry that agents could increase price competition and weaken loyalty, even though the companies cited in the reporting say actual usage is still low.
Timeline of recent developments
| When | Event | Meaning |
|---|---|---|
| This month | Zuckerberg said more than 10 million people in the U.S. and Canada had activated Muse | Meta highlighted scale and early momentum |
| Last month | Amazon blocked Muse | Shows merchant resistance at the platform level |
| This week | Bank of America warned customers about giving login information to AI assistants | Signals growing concern about credential and authorization risks |
| This week | Amazon said it anticipates working with outside agents | Suggests a possible shift toward accommodation |
| This week | Insurify reversed its Muse block after adding offer-accuracy technology | Illustrates the role of safeguards in opening access |
| This week | Chipotle said it is tidying website code to help agents complete orders | Shows merchants are making technical adjustments now |
What the companies are actually saying
The reporting paints a broad but consistent picture: AI shopping traffic is present, but not yet transformative. DoorDash said orders from Muse and other agents are in the hundreds. That is a real number, but not the kind that suggests a wholesale shift in commerce.
At the same time, the response from executives shows that companies are watching carefully. If enough customers begin delegating purchases to AI, the economics of search, discovery and checkout could change quickly. Brands would need to decide whether to optimize for visibility, block access, or build safeguards that preserve pricing and accuracy.
The issue is not only sales. It is also data quality and trust. If an agent sees the wrong price, the wrong inventory status or an inaccurate offer, the customer experience can break down instantly. That is why some companies are not just allowing agents but actively reworking technical systems to make sure the machine gets the same answer a human would.
“delusional”
That was how Tony Xu, DoorDash’s chief executive and a Meta board member, described one part of the conversation around AI agents, underscoring the skepticism that still runs through the sector.
Why it matters for readers, merchants and the wider market
For readers, the practical takeaway is that online shopping may increasingly include an AI layer between the buyer and the store. That could make purchasing faster and more convenient, but it also raises questions about who controls the transaction and what information the assistant is allowed to share.
For merchants, the issue is less about today’s tiny traffic levels than about readiness. Companies that wait until agentic commerce becomes obvious may find themselves scrambling to fix site structures, rewrite rules and adjust support systems later. Those that prepare early may be better positioned if the channel grows.
For the market, Muse is a test case for whether commerce software can become a meaningful distribution channel. Meta’s stated plan to charge fees on completed transactions shows that the company is not viewing this as a side project. It is trying to build a new business line on top of consumer intent.
What happens next?
The next phase will depend on adoption, merchant cooperation and trust. The fact sheet points to several open questions that will shape the story from here.
- Whether Muse adoption will grow enough to materially affect online commerce.
- How Meta will structure and enforce any transaction fee for businesses.
- Whether more major retailers and service providers will open or block access to Muse.
- Whether agents shift shopping behavior toward lower prices at the expense of brand loyalty.
- How banks and merchants will balance convenience with credential and authorization risks.
Those questions matter because the current data cuts both ways. Meta can point to millions of activations and retailer partnerships. Merchants can point to minimal disruption and small order volumes. The truth may be that both are right: consumer interest exists, but the business impact has not yet caught up.
What is clear is that companies are now adapting ahead of the curve. Even where the numbers are low, the policy changes, code cleanups and partnership discussions show that AI shopping agents have already moved from speculation to operations.
Frequently asked questions
What is Meta’s Muse AI shopping agent?
Meta’s Muse is an AI shopping agent designed to browse, compare and complete online shopping tasks for users. The fact sheet says more than 10 million people in the U.S. and Canada have activated it, and more than 25% of users have used it for shopping.
Is Meta’s AI shopping agent already disrupting ecommerce?
No, not yet. Companies surveyed by WIRED largely described the impact as minimal or early-stage, with several saying they have seen little to no meaningful activity from agents. A few brands, however, are already preparing their sites and policies for more AI-driven orders.
How are retailers responding to AI shopping agents?
Retailers and service providers are reacting in different ways. Some initially blocked Muse, some later opened access, and others are adjusting website code or offer systems so AI agents can complete transactions more accurately.
Will Meta charge businesses for AI shopping transactions?
Yes, Meta says it plans to charge businesses a fee for transactions completed via Muse. Zuckerberg has said that fee revenue could eventually exceed Meta’s advertising business, although the current level of activity appears far smaller than that long-term ambition.
