🔍 Read the full analysis: Retail’s AI Shopping Debate: Access, Control And The Customer 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
Walmart announced tools to let approved AI shopping agents search its inventory and make purchases, while Amazon has blocked Meta’s Muse agent from access. Meta says more than 10 million people in the U.S. and Canada set up Muse within about six weeks, but the scale of agent-driven purchases and the future of competing access standards remain uncertain.
Walmart announced an Agent Gateway and customer spending controls on October 21, offering a route for approved AI agents to search its products and make purchases. The move contrasts with Amazon’s decision to block Meta’s Muse, putting retailers, technology companies and shoppers at the center of a dispute over who controls access, checkout and the customer relationship.
Walmart’s Gateway is based on the Model Context Protocol (MCP), a standard that lets software connect to services. The retailer also announced Agent Profiles, which can give agents defined permissions and spending limits. The announcement came as Meta reported early use of its Muse assistant, which can browse websites and complete tasks for users.
On Meta’s October 21 earnings call, the company said more than 10 million people in the United States and Canada set up Muse in about six weeks. Meta said shopping was its most common use and the average purchase exceeded $200. Those figures are company-reported; the source material does not provide independent verification or explain whether the average refers to completed purchases or a broader measure.
Amazon blocked Muse on September 20. Amazon said the agent had not identified itself, had not been disclosed to the company as a shopping agent, and appeared to capture and store customer credentials. Meta disputed that characterization, saying Muse had no visibility into users’ passwords or payment methods. The companies’ accounts leave the handling of credentials in dispute.
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 AI Checkout
The dispute affects which products an agent can compare, where it can complete a purchase and which company gets the resulting customer relationship. A retailer that allows outside agents may gain another route to shoppers, but may have less control over how its products are presented and how the transaction is mediated. A retailer that blocks them can direct shoppers toward its own tools, though users may still seek alternatives.
There is also a potential cost for publishers and affiliate businesses. Amazon Associates commissions generally depend on a customer using a tracked link. If an agent gathers information from a review and completes a purchase without that link, the publisher may not receive credit. The source material describes this as a risk; it does not quantify any lost commissions.
For shoppers, agent access is not the same as consumer protection. Spending caps, purchase approval and payment controls can limit what an agent does, but users still need to know which merchant handles returns, disputes and order support. The agent may simplify the transaction without becoming the party responsible for it.
Products Worth Considering
Dell UltraSharp 27 Thunderbolt Hub Monitor (U2726DER)
Canon IVY 2 Mini Photo Printer Bundle, Pure White
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
The current conflict follows a fast sequence of product and standards announcements. Meta launched Muse on September 8, describing an agent that can open a browser and fill out forms. It runs in a dedicated virtual machine in Meta’s cloud and uses Link by Stripe, which the source says issues a one-time-use card. Amazon restricted its access 12 days later.
Meta announced retail connections with companies including Best Buy, Gap, Sephora and Walmart on September 23. Shopify followed with structured tools for browser agents to complete checkout. In October, Meta and Sierra introduced a draft called the Personal Agent Protocol, or Poppy, with Walmart, Shopify and Stripe among its supporters. The source also identifies Google and Shopify’s Universal Commerce Protocol and OpenAI and Stripe’s Agentic Commerce Protocol as competing approaches. They cover different parts of shopping and are at different stages of development.
These approaches reflect three ways an agent can shop: by using a regular browser, by connecting through a retailer’s dedicated interface, or by following a shared protocol. Browser-based agents can reach sites designed for people, but retailers argue they can be hard to distinguish from human users. Dedicated interfaces give merchants more control over access. Shared standards aim to make connections more consistent, but the protocols have not settled into one common system.
“Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
— Amazon, in the message shown to Muse users
Access, Sales and Standards Remain Unsettled
It is not yet clear how many Muse sessions result in completed purchases, how Meta calculated its reported average purchase value, or how much revenue retailers may gain or lose as agents become more common. The source figures about Muse are company statements, not an independently audited account of shopping activity.
The access dispute also remains unresolved. Amazon’s claims about Muse’s identification and credentials are contested by Meta, and the source material does not report a resolution between the companies. Nor is there evidence yet that shoppers will consistently prefer retailer-approved agents over browser-based tools.
Several protocols are being developed, but their adoption, compatibility and commercial terms are still developing. A published or draft standard does not show that retailers and agent providers will use it broadly. How responsibility will be divided when an agent makes an incorrect purchase, or when a return or payment dispute arises, is also not established in the material.
Retailers Test Agent Access
The immediate next step is whether retailers and agent providers put their announced connections into regular use. Walmart’s Gateway and Agent Profiles will offer a test of whether merchants can permit agent shopping while setting limits on what an agent can access or spend. Further adoption of Poppy, UCP or ACP would show whether companies can agree on ways to identify agents and pass purchase instructions between systems.
Shoppers and publishers will be watching the practical outcomes: whether agents can find accurate prices and inventory, whether they complete purchases reliably, and how orders, returns and commissions are attributed. The announcements establish competing approaches, but the scale and terms of everyday AI shopping have not yet been settled.
Key Questions
What did Walmart announce?
Walmart announced an Agent Gateway using MCP and Agent Profiles that can set permissions and spending limits for shopping agents. The announcement was made on October 21, 2026.
Why did Amazon block Meta’s Muse?
Amazon said Muse had not identified itself as an agent, had not been disclosed as a shopping tool and appeared to capture and store customer credentials. Meta disputed the credential claim, saying Muse could not see passwords or payment methods.
How many people have set up Muse?
Meta said on its October 21 earnings call that more than 10 million people in the United States and Canada set it up in about six weeks. This is a company-reported figure; it does not establish how many users completed purchases.
Does an AI agent complete every purchase automatically?
No. Agents may research products without reaching checkout, and access can be blocked or limited by a retailer. Walmart’s announced profiles also offer a way to set permissions and spending controls.
What remains uncertain about AI shopping?
The volume of completed agent purchases, the business terms for retailers and publishers, and which technical standards will gain broad use remain unclear. The source material also does not establish who will handle disputes when an agent makes a purchase error.
Source: ThorstenMeyerAI.com
