🔍 Read the full analysis: AI Agents Are Shopping: Amazon And Walmart Pick Different Paths 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 has announced tools intended to let approved AI agents search and buy from its store, while Amazon has blocked Meta’s Muse agent and says outside agents must identify themselves. The split highlights competing approaches to who controls AI-mediated shopping, though current data suggests agents mostly research products rather than complete purchases.
Walmart is opening a structured route for approved AI shopping agents to search its inventory and make purchases, while Amazon has blocked Meta’s Muse from accessing its site. The opposing approaches, announced or described in a six-week burst of developments, put retailers and AI companies in a contest over how automated shopping works and who controls the customer relationship.
Walmart announced Agent Profiles and the Walmart Agent Gateway on Oct. 21. The gateway is designed to let approved agents interact with Walmart through a structured connection based on the Model Context Protocol, rather than navigating the store’s pages as a person would. Agent Profiles are intended to let customers set permissions, including spending limits. The announcement does not, by itself, show how widely the tools are available or how many shoppers are using them.
Amazon took a different position after Meta’s Muse began shopping through websites in a browser. Amazon blocked the agent on Sept. 20. It said Muse had not identified itself as an agent and appeared to capture and store customer credentials. Meta’s launch materials disputed the credential concern, saying Muse has no visibility into users’ passwords or payment details. The accounts describe a dispute between the companies, not an independently established finding about how Muse handled credentials.
Meta said on its Oct. 21 earnings call that more than 10 million people in the United States and Canada had set up Muse in about six weeks, that shopping was its most common use, and that the average purchase was above $200. Those are Meta’s figures; the company did not provide an independent audit in the source material. Muse can browse and fill forms in a cloud-based virtual machine, while Walmart’s gateway offers a more direct, retailer-approved route.
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 Shopping Route
The choice of route affects retailer access, customer data and accountability. A browser agent can act on pages built for people, but a retailer may have difficulty distinguishing that activity from a customer’s own browsing. A structured gateway can make an agent identifiable and give the store a defined way to set permissions, but it also gives the retailer more control over which agents can connect and what they can do.
The dispute also touches advertising and referrals. The source reports that Amazon earned more than $68 billion in advertising revenue last year, citing GeekWire, while Meta chief executive Mark Zuckerberg has said Meta expects to take a small fee from transactions. Those business interests make agent access more than a technical question: if an agent steers a purchase, the retailer, platform or publisher may seek a share of the value.
For shoppers, the practical issue is whether an agent can buy without clear approval and which company’s protections apply. Spending caps and purchase-by-purchase approval can limit risk, but the source does not establish a common standard for refunds, returns, disputed orders or responsibility when an agent makes a mistake.
Popular right now
Samsung 42-Inch Class OLED S90F 4K Smart TV (2025)
De'Longhi Linea Classic Espresso Machine with Adjustable Milk Frother
Ninja Luxe Café Mini Plus Espresso and Drip Coffee Machine with AutoFroth, Stainless Steel, ES351
As an affiliate, we earn on qualifying purchases.
How Retailer Access Took Shape
The latest announcements followed a rapid sequence. Meta launched Muse on Sept. 8, describing it as a personal agent that could open a browser, fill out forms and act for a user. On Sept. 20, Amazon restricted Muse. Three days later, Meta announced retail connectors with Walmart and several other companies, including Best Buy, Gap, Sephora and Wayfair, and Zuckerberg discussed transaction fees as a possible business model.
Walmart’s gateway is one approach among several efforts to standardize agent commerce. Shopify has offered structured checkout tools for browser agents. Meta and Sierra announced the Personal Agent Protocol, or Poppy, with Walmart, Shopify and Stripe, then published a draft on Oct. 9. A separate Universal Commerce Protocol is associated with Google and Shopify, while OpenAI and Stripe developed the Agentic Commerce Protocol. The existence of competing protocols does not mean they are interchangeable or widely adopted.
There is also a gap between shopping research and completed orders. According to HUMAN Security data for September, 81% of agent requests went to product and search pages, while 2.4% went to checkout and payment. Those figures describe requests in that data set, not all shopping agents or all online purchases. They suggest that much current activity is product discovery rather than automated checkout.
“Muse has no visibility into people’s passwords or payment methods.”
— Meta, in its launch materials
Adoption and Purchase Safeguards
It remains unclear how many Muse users have completed purchases, how often Walmart’s gateway will be used, or whether customers will adopt agent shopping beyond early experimentation. Meta’s reported user and purchase figures lack an independent audit in the source material, and the cited traffic data covers a particular security-data set rather than the entire market.
The companies’ disagreement over credentials and agent identification is also unresolved here. The material does not establish whether Amazon’s concerns were confirmed by an independent technical review, or how Walmart will verify agents and enforce customer-set limits in practice. Nor does it specify a shared process for returns, payment disputes, unauthorized orders or errors made during checkout.
Finally, the competing protocols remain at different stages. Poppy was published as a draft, while other protocols have separate designs and backers. It is not clear whether retailers, AI platforms and payment providers will converge on a common system or continue to support multiple, incompatible routes.
Retailer Rollouts and Protocol Tests
The next test is whether Walmart’s announced gateway and profiles become available to shoppers and outside agents at meaningful scale. Further details about eligible agents, permission settings, payment handling and returns would show how the system works beyond its initial announcement. Walmart has not supplied adoption figures in the source material.
Meta and its partners are likely to face continued scrutiny over how Muse identifies itself, handles credentials and gets authorization before spending. The competing protocol groups will also need to show that their systems work across retailers and payment providers, rather than only in announcements or limited integrations. Until those details emerge, the confirmed development is a strategic split: Walmart is building a defined access route, while Amazon is restricting an outside agent and promoting its own approach.
Key Questions
What did Walmart announce?
Walmart announced Agent Profiles and an Agent Gateway on Oct. 21, 2026. The gateway is designed to let approved AI agents search inventory and purchase through a structured connection, while profiles are intended to give customers controls such as spending limits.
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 include an independent technical finding settling the dispute.
Are AI agents already completing many online purchases?
The cited September HUMAN Security data indicates that 81% of agent requests went to product and search pages and 2.4% to checkout and payment. Those measurements cover that data set, not all agents or online shopping, and suggest that research is more common than checkout in the observed activity.
What should shoppers check before allowing an agent to buy?
Shoppers can review whether the agent requires approval for each order, supports spending limits and provides clear payment and return information. The source material does not establish one standard for who is responsible when an agent makes an error or a purchase is disputed.
Source: ThorstenMeyerAI.com
