🔍 Read the full analysis: Why The AI Shopping Agent Debate Matters To Amazon And Walmart on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Meta’s Muse shopping agent has run into Amazon’s access restrictions, while Walmart is building tools for approved agents to search and buy. The dispute matters because control of the shopping agent could affect customer data, advertising, commissions and the checkout experience, but most agent activity still appears to be research rather than purchases.
Walmart is opening a structured route for approved AI shopping agents, while Amazon has restricted Meta’s Muse from accessing its site, setting up a dispute over who controls the path from product search to purchase. The clash matters to retailers and shoppers because an agent could become the customer’s main interface for comparing products, while determining which merchants, ads and checkout systems it can reach.
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 inventory and make purchases through a structured connection rather than navigating ordinary web pages. Agent Profiles are intended to set permissions, including spending limits. The announcement positions Walmart as a retailer seeking to make its services available to external agents under defined conditions.
Amazon took a different approach with Meta’s Muse. The Verge reported that Muse could not check Amazon prices after Amazon blocked its access. Amazon said Muse had not identified itself as an agent, that Meta had not told Amazon it would shop on the site, and that the system appeared to capture and store customer credentials. Meta’s launch post said Muse could not see users’ passwords or payment methods. Those statements describe the companies’ positions; the available material does not independently resolve the disagreement.
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 said the average purchase was above $200. These are company-reported figures, not independent estimates. The same day Walmart announced its tools, the figures underscored the potential audience for agents while leaving open how often users complete purchases through them.
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 Journey
The commercial stakes extend beyond whether an agent can find a lower price. If shoppers increasingly ask an AI to compare products and choose where to buy, the agent may influence which retailers get considered—and which offers or ads users see. Amazon’s restrictions protect its ability to set terms for access to its marketplace and customer accounts. Walmart’s gateway offers a way to participate while requiring agents to connect through retailer-approved tools.
The shift could also affect advertising and referral revenue. The source material says Amazon earned more than $68 billion from advertising last year, citing GeekWire. If an agent mediates product discovery, retailers may have to rethink how sponsored placements fit into recommendations. Meta chief executive Mark Zuckerberg said Meta expected to make money by taking a small fee from transactions. That is a stated business aim, not evidence that a particular fee structure is in place across retailers.
Publishers and affiliate businesses face a separate risk. Amazon Associates commissions rely on a customer using a tracked link. If an agent reads a review and then completes a purchase through another route, the publisher may not receive credit. Whether agents will cause a material decline in referrals or commissions is not established by the figures provided, but the change could make attribution rules a point of negotiation among retailers, platforms and publishers.
For shoppers, the convenience of delegating research or purchases comes with questions about spending permissions, payment handling and returns. An agent’s recommendation does not by itself clarify who is responsible if it buys the wrong item, applies a discount incorrectly or selects a seller with different return terms. Spending caps and purchase approvals can limit exposure, but the terms will depend on the agent and merchant.
Products Worth Considering
Amazon Kindle Paperwhite (12th Generation, 2024), 16 GB, 7-inch Display, Jade
Bose S1 Pro+ Portable PA Speaker
Bose TV Speaker (2nd Gen) Compact Soundbar with Bluetooth and HDMI-ARC, Black
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
The current dispute reflects different ways for an agent to shop. A browser-based agent can interact with pages as a person would, potentially using the shopper’s credentials. This approach can work across many sites, but makes it harder for retailers to distinguish an authorized agent from a human customer or to control how its systems are accessed. Meta says Muse runs in a dedicated virtual machine in its cloud and uses a one-time card through Link by Stripe for payments.
A second approach is a retailer’s agent-specific connection. Walmart’s gateway and Shopify’s tools give participating agents structured ways to search or check out. A third is a shared protocol intended to let agents and merchants exchange information under common rules. The source material describes several efforts, including the Personal Agent Protocol, backed by Meta, Sierra and others, and Google’s Universal Commerce Protocol. These projects differ in scope and status; their existence does not mean retailers and agents have settled on a common standard.
The timeline has moved quickly. Meta launched Muse on September 8, then announced retail connectors that included Walmart and other retailers on September 23. Amazon blocked Muse on September 20, according to the supplied account. Walmart announced its agent gateway and profiles on October 21. A separate example shows the limits of the technology: OpenAI and Stripe launched an Instant Checkout feature in ChatGPT in September 2025, then OpenAI said in March 2026 it would let merchants use their own checkout experiences while focusing its efforts on product discovery.
““[Muse] has no visibility into people’s passwords or payment methods.””
— Meta, in its Muse launch post
How Much Agents Actually Buy
It remains unclear how much shopping-agent use turns into completed transactions. The source material cites HUMAN Security data from September showing 81% of agent requests went to product and search pages, while 2.4% went to checkout and payment. Those figures describe the requests measured in that dataset and period; they do not establish the share of all online shoppers or purchases involving agents.
Other open questions include how Walmart will approve agents, how its spending limits will work in practice, and whether Amazon will allow access through any third-party agent under different conditions. The source material also describes competing protocols at different stages, but does not establish which, if any, will become widely adopted. Meta’s usage and purchase figures are company-reported, and the basis for comparing them with other shopping services is not provided.
Responsibility for errors, refunds, disputed purchases and customer data may also vary by agent, retailer and payment provider. The available information does not lay out a single set of rules governing those issues across services.
Retailers Set the Access Rules
The next test is whether agents begin using retailer-approved connections at scale, rather than relying mainly on browser interactions. Walmart’s gateway will show how a major retailer handles agent identity, permissions and checkout when outside services request access. Amazon’s policies will indicate whether it intends to keep third-party agents out or define a narrower route for approved systems.
Shoppers and publishers will also be watching for clear terms on purchase approval, payment, returns and referral credit. For now, the central development is a contest over access and control—not proof that AI agents have replaced ordinary online shopping. The share of agent requests that reach checkout, along with the terms retailers set for participation, will help show whether the new tools change buying behavior.
Key Questions
What did Walmart announce?
Walmart announced an Agent Gateway using MCP and Agent Profiles with permissions such as spending limits. The tools are intended to let approved agents search Walmart inventory and make purchases through a structured connection.
Why did Amazon block Meta’s Muse?
Amazon said Muse had not identified itself as an agent, that Meta had not told Amazon it would shop on the site, and that Muse appeared to capture and store customer credentials. Meta said Muse could not see users’ passwords or payment methods. The supplied information does not independently settle those claims.
Are AI agents already making many purchases?
The evidence cited here suggests much recorded agent activity is product research. HUMAN Security’s September data said 81% of agent requests went to product and search pages and 2.4% to checkout and payment. Those figures apply to that dataset and period, not necessarily to all shopping.
How could AI shopping agents affect publishers?
Affiliate publishers may lose referral credit if an agent reads their content but completes a purchase without using a tracked link. The scale of any effect is not yet established.
What should shoppers check before using an agent?
Shoppers should check what purchases require approval, whether they can set a spending cap, how the agent handles payment, and which seller’s return policy applies. The exact protections depend on the service and retailer.
Source: ThorstenMeyerAI.com
