🔍 Read the full analysis: The Customer Ownership Question Behind AI Shopping on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Meta says more than 10 million people in the U.S. and Canada set up its Muse agent in about six weeks, with shopping its most common use. Walmart is building an approved route for shopping agents, while Amazon has blocked Muse, putting access, customer relationships and transaction credit at the center of the debate.
Walmart is opening a controlled route for AI shopping agents while Amazon blocks Meta’s Muse, bringing a dispute over who controls online purchases into focus. Meta said on its October 21 earnings call that more than 10 million people in the United States and Canada had set up Muse in roughly six weeks, with shopping its most common use; the figures are Meta’s own and do not show how many people completed purchases.
Walmart’s October 21 announcement introduced Agent Profiles, which let customers set spending limits for agents, and a Walmart Agent Gateway based on the Model Context Protocol, or MCP. The gateway is designed to let approved agents search Walmart’s inventory and make purchases through a structured connection rather than browsing the retailer’s website as a person would.
Amazon has taken a different approach with Muse. On September 20, the agent was blocked from the site. Amazon said the agent had not identified itself and that its access breached the company’s Conditions of Use. It also said Muse appeared to capture and store customer credentials. Meta’s launch post disputed that account, saying Muse “has no visibility into people’s passwords or payment methods.”
The immediate stakes are practical as well as commercial. An agent that shops through a browser may be able to reach many stores, but retailers can object that they cannot reliably distinguish the software from a customer. A retailer-run connection gives a business more control over which agents can access its catalog and how transactions work. The available details do not establish that Muse users can complete purchases at every retailer named in Meta’s announcements.
Agentic shopping
Three ways an AI agent buys for you
Every fight between shops and AI agents comes down to which of these three routes the agent uses, and whether the shop agreed to it.
The browser, like a human
The agent opens pages built for people, signs in with your login and clicks through checkout. Works almost anywhere, which is why shops object: they can't tell the agent from you.
Examples: Meta's Muse (from a virtual machine in Meta's cloud), Perplexity's Comet (from your own computer)
The shop's own agent door
Approved agents search and buy through a structured interface instead of clicking through pages. The shop sets the rules.
Examples: Walmart Agent Gateway (MCP, beta), Shopify MCP server and WebMCP checkout tools
Shared protocols
Common rules for who the agent is, what you allowed, and how the checkout and payment are confirmed.
Examples: Personal Agent Protocol (draft), Universal Commerce Protocol, Agentic Commerce Protocol
The protocols competing to be the standard
| Protocol | Backers | Covers | Status |
|---|---|---|---|
| Personal Agent Protocol (Poppy) | Sierra, Meta, Walmart, Shopify, Stripe + 35 design partners | Agent identity, customer permission, one session across site, API and company agent | Draft 0.1 (9 Oct 2026) |
| Universal Commerce Protocol (UCP) | Google, Shopify; site also lists Etsy, Wayfair, Target, Walmart, Amazon, Microsoft | Discovery, cart, checkout, orders; payment mandates via AP2 | Published |
| Agentic Commerce Protocol (ACP) | OpenAI, Stripe | Checkout between agent and merchant | Open source; ChatGPT's own checkout scaled back in March 2026 |
What a careful agent purchase looks like
- You give the task and a limit, for example a $50 cap.
- The agent identifies itself to the shop and gets only the access you approved.
- It prepares the cart and shows you an approval card.
- Payment goes through a token or one-time card, so the agent never sees your card.
- The shop’s normal returns apply if the agent got it wrong.
Sources: Meta (8 Sep 2026), Walmart (21 Oct 2026), TechCrunch on Shopify (28 Sep 2026), Sierra (9 Oct 2026), ucp.dev, agenticcommerce.dev, OpenAI (24 Mar 2026). Purchase steps: Walmart’s agent terms and Meta’s Muse description.
Powered by Thorsten Meyer AI
Who Controls the Shopping Relationship
The debate is about more than whether an agent can find a low price. It concerns who mediates the purchase, what information the retailer receives, and which company can maintain the customer relationship after the transaction. Retailers have incentives to keep shoppers using their own services and assistants; agent makers want their software to be useful across stores.
That tension also affects advertising and referrals. GeekWire reported that Amazon earned more than $68 billion in advertising revenue last year. Meta chief executive Mark Zuckerberg said the company expects to make money by taking a small fee from transactions, or as Meta put it, “a percent of the value” Muse creates. Those statements describe business ambitions, not evidence that either company has settled how agents will be paid across the market.
Publishers and affiliate businesses could also lose credit for recommendations. Amazon’s Associates program pays commission on purchases made after a customer follows a tracked link. If an agent reads a review but buys directly through a retailer or its own assistant, that link may not record the sale. How common that outcome is remains unclear, but the changing path from recommendation to checkout could affect how product information is funded.
Products Worth Considering
Dell XPS 13 9350 AI Professional Laptop 13.4 120Hz FHD+ Display
Dell XPS 13 9350 Laptop, 13.4 FHD+ Display, Intel Ultra 7 256V, 16GB DDR5, 1TB SSD
Dell UltraSharp 27 Thunderbolt Hub Monitor (U2726DER)
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
The shift has happened quickly. Meta launched Muse on September 8, describing it as a personal agent able to open a browser, fill out forms and act for a user. It runs in a dedicated computer environment in Meta’s cloud and uses Link by Stripe, which can issue a one-time-use payment card. On September 23, Meta announced retail connectors involving Best Buy, Gap, Sephora, Walmart, Wayfair and Expedia.
Other companies are building more structured ways for agents to interact with shops. Shopify has offered checkout tools for browser agents, while Walmart’s gateway is intended to connect approved agents to its services. Several protocols are also being developed, including the Personal Agent Protocol backed by Meta, Sierra and partners, and Google’s Universal Commerce Protocol. The Personal Agent Protocol draft was published in October; these efforts are still developing and do not yet amount to one universally adopted system.
There has already been a setback for agent-led checkout. OpenAI introduced Instant Checkout in ChatGPT using the Agentic Commerce Protocol in September 2025, then said in March 2026 that it was stepping back from that version to let merchants use their own checkout experiences while focusing on product discovery. That change suggests that connecting agents to stores is not the same as persuading retailers to hand over checkout.
““has no visibility into people’s passwords or payment methods.””
— Meta, in its Muse launch post
Adoption and Purchase Data Remain Limited
Meta’s reported 10 million setups do not establish how many users made a purchase, how often they use Muse, or how the figure was measured. Meta also said the average purchase was over $200, but the supplied information does not specify the measurement period or provide independent confirmation. Those numbers should not be read as a measure of completed agent transactions across retailers.
There are also unanswered questions about the rules for Walmart’s gateway, which agents will be accepted, and how customers can resolve disputes when an agent makes a mistake. The source material describes a planned, controlled access route but does not set out its full operating terms. Nor does it establish how Amazon’s restrictions may change or whether other retailers will permit browser-based agents.
One study cited in the source, from HUMAN Security in September, found that 81% of agent requests went to product and search pages and 2.4% to checkout and payment. Those figures describe the requests observed in that study; the supplied material does not detail its full sample or establish that they represent all shopping agents. The broader question of whether agents will research products far more often than they complete transactions remains open.
Retail Access Rules Take Shape
The next test is whether Walmart’s gateway moves from announcement to routine use, and which agent providers connect to it. The Personal Agent Protocol and other commerce standards are also still being developed, so their reach and compatibility will need to be demonstrated in practice rather than assumed from partner lists.
For shoppers, the immediate step is to check what an agent is allowed to do before connecting an account or payment method. Purchase approval, spending limits and payment controls can reduce the risk of an unwanted order. Customers should also check which retailer’s return policy applies and who handles problems when an agent places the order. The next phase will show whether agent shopping becomes a broadly supported service or remains limited by retailer-by-retailer access rules.
Key Questions
What is the main development in AI shopping?
Walmart announced an agent gateway and spending controls as Meta reported early Muse usage figures. Amazon, by contrast, has blocked Muse from its site.
How many people have set up Meta Muse?
Meta said more than 10 million people in the United States and Canada set it up in about six weeks. That is a company-reported setup figure, not a count of completed purchases.
Why did Amazon block Muse?
Amazon said Muse was an unauthorized agent that did not identify itself and appeared to capture and store customer credentials. Meta disputed the credential concern, saying Muse could not see users’ passwords or payment methods.
What is Walmart’s Agent Gateway?
Walmart describes it as an MCP-based connection through which approved agents can search inventory and buy without navigating the store’s website like a human visitor. Walmart also announced Agent Profiles for customer-set spending limits.
Can shoppers prevent an AI agent from buying without approval?
Controls vary by service. The source material describes Walmart spending limits, but does not establish the full controls available across agents. Users should check approval settings, spending caps and payment protections before authorizing purchases.
Source: ThorstenMeyerAI.com
