🔍 Read the full analysis: How Retailers Are Responding To AI Shopping Agents on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Retailers are taking different approaches to AI agents that can search for products and make purchases: Walmart has announced tools for approved agents, while Amazon has blocked Meta’s Muse from accessing its site. Meta says Muse attracted more than 10 million users in the US and Canada in about six weeks, but available data suggests agents still do far more product research than checkout.
Walmart is opening a route for approved AI shopping agents, while Amazon has blocked Meta’s Muse, the companies’ contrasting moves show retailers beginning to set different rules for software that shops on a person’s behalf. The dispute affects whether agents can compare prices and complete purchases across stores—and which company controls the customer relationship when they do.
Walmart announced its Agent Gateway and Agent Profiles on October 21. The Gateway uses the Model Context Protocol, or MCP, to let approved agents search Walmart inventory and make purchases through a structured connection rather than navigating the retailer’s consumer website. Profiles are intended to let shoppers set permissions, including spending limits. The announcement describes a way for agents to interact with Walmart; it does not establish that all agents or all purchases will be supported.
Amazon has taken a more restrictive approach to Meta’s Muse. According to the source material, Amazon cut off Muse on September 20 and displayed a notice saying that continued access by an unauthorized AI agent violated its Conditions of Use. Amazon said Muse had not identified itself as an agent and appeared to capture and store customer credentials. Meta, in its launch post, said Muse had no visibility into users’ passwords or payment methods. Those are opposing company statements; the materials provided do not independently resolve the dispute.
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, and that shopping was its most common use. Meta also reported that the average purchase was more than $200. These are Meta’s figures; the source material gives no independent verification or breakdown of how many users completed a purchase, how often they used the agent, or what products they bought.
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
Control of Shopping Moves to Agents
The immediate issue is practical: an agent can only compare or buy from retailers it can access. In one example reported by The Verge, Meta’s Muse compared the price of granola bars at Sam’s Club, Walmart and Costco, then said it could not check Amazon because Amazon had blocked it. That leaves shoppers with a partial comparison when a major retailer excludes the agent.
The larger commercial question is who gets the customer relationship. If a shopper asks an agent to find a product, the agent may shape the shortlist before a retailer’s own site or app is opened. The source material says Amazon earned more than $68 billion in advertising revenue last year, citing GeekWire, while Meta has said it expects to earn a small fee from transactions. Those figures and statements point to competing business interests, but the material does not establish how agent use will affect either company’s revenue.
Independent retailers and publishers may also be affected. HUMAN Security’s September data, as cited in the source, found that 81% of agent requests went to product and search pages, compared with 2.4% to checkout and payment. The figures describe the requests measured in that dataset, not all shopping agents or all online shopping. They suggest that agents are currently used more for research than for completing transactions.
Products Worth Considering
Dell XPS 13 9350 Laptop, 13.4 FHD+ Display, Intel Ultra 7 256V, 16GB DDR5, 1TB SSD
Dell XPS 13 9350 AI Professional Laptop 13.4 120Hz FHD+ Display
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Three Routes Into Online Stores
Retailers and agent makers are pursuing three broad technical approaches. The first is browser-based shopping: an agent navigates pages as a person would, sometimes using a shopper’s login. Meta says Muse runs in a dedicated virtual machine in its cloud and uses Link by Stripe for one-time-use payment cards. Amazon’s objection to Muse includes its handling of credentials; Meta disputes that concern. Browser agents can reach sites built for people, but stores may find it harder to distinguish an agent from a customer.
The second route is a retailer-provided connection. Walmart’s MCP-based Gateway is an example. Shopify has also offered tools for browser agents to interact with checkout through structured functions, rather than relying only on screen-reading and simulated clicks. These connections can give a retailer more control over which actions are allowed, though their availability depends on retailer participation and implementation.
A third route is shared protocols intended to let agents and merchants communicate across services. The source identifies Personal Agent Protocol (Poppy), backed by Meta, Sierra, Walmart, Shopify and Stripe, among others, as a draft published in October. It also describes Google and Shopify’s Universal Commerce Protocol and OpenAI and Stripe’s Agentic Commerce Protocol. The approaches cover different parts of shopping and are at different stages; their existence does not mean that retailers have adopted one common standard.
The recent sequence helps explain the fast-moving debate: Meta launched Muse on September 8; Amazon blocked it on September 20; Meta announced retail connectors on September 23; and Walmart announced its Gateway and Profiles on October 21. The source also says OpenAI stepped back in March 2026 from its initial ChatGPT Instant Checkout approach, saying it would focus on product discovery while merchants used their own checkout experiences. That example underscores that building agent connections does not by itself settle how purchases will be completed.
“Muse has no visibility into people’s passwords or payment methods.”
— Meta, in its Muse launch post
Adoption, Access and Payments Remain Open
It remains unclear how many shoppers will rely on agents to complete transactions rather than research products, and whether reported early usage will persist. Meta’s user and purchase figures are company-reported, while the HUMAN Security percentages cover a particular dataset and request period. Neither source, as presented here, provides a full measure of agent-led retail sales.
The practical reach of Walmart’s Gateway and Agent Profiles is also not yet clear from the supplied material: it does not specify a complete list of participating agents, rollout timing, or all supported purchase and payment cases. Nor does it establish whether major retailers will agree on common protocols. Amazon’s concerns about Muse and Meta’s response remain disputed company positions, not independently resolved findings.
There are questions for shoppers as well as retailers. The source recommends setting purchase approval requirements and spending caps, and knowing which seller’s return policy applies. It does not provide details on how responsibility for refunds, mistaken orders, or disputes will be divided between an agent provider and a merchant across the different systems.
Retailer Integrations Will Test Adoption
The next test is whether the announced connections move from plans and protocols into reliable, broadly available shopping. Watch for Walmart to provide details on Gateway access, supported agents and how shoppers manage Agent Profile permissions. Retailer participation will matter as much as the technology: an agent cannot offer a complete comparison if some stores block it or keep their inventory and checkout systems closed to outside tools.
For shoppers, the near-term question is not only whether an agent can find a lower price, but what it can do without approval and how a purchase is handled afterward. Spending limits, confirmation steps, payment controls and clear return terms will be important signs of how each service protects consumer choice. The source material does not give a common timetable for wider adoption or a final outcome in the dispute between Amazon and Meta.
Key Questions
What is Walmart’s Agent Gateway?
Walmart announced an MCP-based connection intended to let approved AI agents search its inventory and make purchases through structured tools. The supplied material does not list every supported agent or describe the full rollout.
Why did Amazon block Meta’s 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. The source material does not independently settle the disagreement.
Are AI shopping agents already completing many purchases?
The available figures cited in the source suggest more agent activity happens during product research than at checkout: HUMAN Security reported that 81% of measured requests went to product and search pages and 2.4% to checkout and payment. Those figures apply to its September dataset and are not a measure of every agent or retailer.
The source recommends using purchase approvals, spending caps and one-time-use cards where available, and checking which retailer’s return policy applies. Specific controls vary by agent and merchant; the material does not establish that every service offers each option.
Source: ThorstenMeyerAI.com
