🔍 Read the full analysis: What Happens When AI Shopping Agents Meet Retail’s Rules? on ThorstenMeyerAI.com
Get the little things that make your day 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 United States and Canada set up its Muse agent within about six weeks, with shopping its most common use. Walmart is building a gateway for approved agents, while Amazon has blocked Muse and requires agents to identify themselves. Access, payment controls and how sales are credited remain unsettled.
Retailers are setting competing rules for AI shopping agents as Meta says more than 10 million people in the United States and Canada set up its Muse agent within roughly six weeks. Walmart has announced a gateway and user profiles for approved agents, while Amazon blocked Muse, making access to products and checkout a central dispute in the emerging market.
Meta reported its Muse figures during its October 21 earnings call, saying shopping was the agent’s most common use and the average purchase was more than $200. Those are company-reported figures; the material does not provide an independent breakdown of completed purchases or explain how the average was calculated. Muse can use a browser to search and buy on a user’s behalf, but its access depends on each retailer’s rules.
Walmart’s October 21 announcement introduced Agent Profiles, which can set spending limits, and an Agent Gateway using the Model Context Protocol, or MCP. The gateway is intended to let approved agents search inventory and make purchases through a structured connection rather than behaving like a person clicking through a website. Walmart’s announcement points toward controlled access, not open entry for every agent.
Amazon took a different approach. On September 20, users attempting to use Muse on Amazon saw a notice saying access by an unauthorized AI agent violated the retailer’s Conditions of Use. Amazon said Muse had not identified itself and appeared to capture and store customer credentials. Meta’s launch post said Muse had no visibility into users’ passwords or payment methods. Those statements describe opposing accounts of the access and security concerns; the supplied material does not independently resolve 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
Retailers Control the Agent Checkout
The dispute concerns more than whether a shopping bot can load a product page. Retailers want to control who accesses their inventory, what data an agent receives and how a purchase is completed. A retailer’s own assistant or an approved technical connection can preserve that control. A browser-based agent, by contrast, may interact with a shop much like a human customer, making it harder for the retailer to distinguish automated activity from ordinary browsing.
There is also a commercial question: who owns the customer relationship when software compares products and places an order? Meta chief executive Mark Zuckerberg said the company expected to earn “a small fee from transactions.” The source also reports that Amazon earned more than $68 billion from advertising last year, citing GeekWire. For retailers, a third-party agent could influence which products shoppers see and where they buy; for Meta, transaction fees could make shopping a revenue source. The eventual arrangements, fees and influence over recommendations are not established.
Publishers and affiliate businesses may also be affected. The source says Amazon Associates commissions depend on a customer clicking a tracked link. If an agent reads a review but completes a purchase through a different route, a publisher may not receive credit under that model. How new agent systems will attribute referrals or compensate publishers remains open.
Products Worth Considering
Hisense 85 U6 Pro Series Mini-LED ULED 4K UHD HDR Gaming AI Smart Fire TV (85U6SF Pro)
TCL 85-Inch QM6K Series QLED Mini LED 4K HDR Smart Google TV
Levoit Core 400S-P Smart Air Purifier for Large Rooms
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
The recent developments have moved quickly. Meta launched Muse on September 8, describing an agent that can open a browser, fill out forms and negotiate for users. The service 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. On September 23, Meta announced retail connectors involving Best Buy, Gap, Sephora, Walmart, Wayfair and Expedia.
Other companies are testing different ways to connect agents and shops. Shopify introduced structured tools for browser agents to complete checkout, while Walmart’s gateway is designed for approved connections. Proposals for shared protocols include the Personal Agent Protocol, known as Poppy, and Google and Shopify’s Universal Commerce Protocol. OpenAI and Stripe’s Agentic Commerce Protocol is another approach. These systems aim to standardize parts of discovery, checkout, identity or permission, but their functions and adoption differ.
Protocols alone have not settled how agent shopping works in practice. The source notes that OpenAI launched Instant Checkout in ChatGPT in 2025 and later stepped back from that model, saying its initial version lacked the flexibility it wanted and that it would focus on product discovery while merchants used their own checkout experiences. A September HUMAN Security analysis cited in the source found 81% of agent requests went to product and search pages, compared with 2.4% to checkout and payment. Those figures describe the requests covered by that analysis, not all agent shopping activity.
“Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
— Amazon, in the notice shown to Muse users
Access, Safety and Sales Credit
How widely retailer gateways will be available is not yet clear, nor whether major retailers will accept outside agents on common terms. The source describes several protocols at different stages, including a draft of Poppy, but does not establish which will gain broad adoption or how they will work together.
It is also unclear how often agents complete purchases, how Meta calculated its reported average purchase value, and whether the sales figures represent transactions initiated or completed through Muse. The cited request data suggest many agent interactions focus on product pages rather than checkout, but do not establish the conversion rate across services or retailers.
Security and responsibility remain unresolved in the material. Amazon and Meta offered conflicting statements about Muse’s handling of credentials, and the source does not report an independent finding on that disagreement. Consumers may also need to determine which retailer’s return, payment and customer-service rules apply when an agent places an order.
Retail Rules Will Shape Adoption
The next test is whether retailers make agent access operational: which agents they approve, what information those agents can retrieve and what limits customers can set before an order is placed. Walmart’s announced profiles and gateway offer one model; Amazon’s block and identification requirement show another. The source material does not give a date for broad rollout or a schedule for further policy changes.
For shoppers, the practical safeguards are straightforward even as standards develop: check whether an agent asks for purchase approval, set a spending cap where available and understand which payment method is used. Confirming the retailer’s return policy and keeping track of the order can help clarify who handles a problem after an automated purchase.
For publishers and merchants, the coming issue is whether technical integrations preserve referral credit and give businesses a say in checkout. Further announcements from retailers, payment firms and protocol developers may clarify those terms. Until then, the confirmed picture is a fast-growing consumer tool facing uneven access, competing commercial interests and unresolved rules.
Key Questions
What is Meta’s Muse shopping agent?
Muse is an AI agent that Meta says can browse websites and take actions for a user, including shopping. The source says it runs in a virtual machine in Meta’s cloud and can use a one-time-use payment card through Link by Stripe.
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 said Muse could not see users’ passwords or payment methods. The supplied material does not independently settle the disagreement.
How is Walmart handling AI shopping agents?
Walmart announced an Agent Gateway for approved agent connections and Agent Profiles that can set spending limits. Its approach is designed to give agents a structured way to access inventory and make purchases.
Do AI agents usually complete purchases?
Not based on the cited data alone. A September HUMAN Security analysis in the source found 81% of the agent requests it examined went to product and search pages, while 2.4% went to checkout and payment. The figures do not establish the checkout rate for all agents or retailers.
What should shoppers check before using an agent?
Check for purchase approval settings, spending limits and payment details. Also review the retailer’s return policy and keep the order confirmation so it is clear where to go if an order needs attention.
Source: ThorstenMeyerAI.com
