🔍 Read the full analysis: When AI Shops, The Customer Relationship Gets Complicated on ThorstenMeyerAI.com
Get the latest gadgets 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 US and Canada set up its Muse shopping agent within about six weeks, but access to retailers is uneven. Walmart is building tools for approved agents, while Amazon has blocked Muse, bringing questions about customer data, payments and referral revenue into focus.
Meta says more than 10 million people in the United States and Canada set up its Muse agent in roughly six weeks, as retailers take sharply different approaches to AI-led shopping: Walmart is creating an access route for agents, while Amazon has blocked Muse. The split puts control of the shopping relationship, transaction access and purchase data at the center of a fast-developing retail dispute.
Meta disclosed the adoption figures during its October 21, 2026, third-quarter call, according to the source report. The company said shopping is the most common use of Muse and that the average purchase made through it is more than $200. Those are Meta’s own figures; the report does not provide independent verification or additional detail about how the figures were calculated.
The access dispute became visible when The Verge tested Muse by asking where granola bars cost less per piece. The agent returned Sam’s Club, Walmart and Costco, then said it could not check Amazon because the retailer had blocked it. Amazon said Muse had not identified itself as an agent and raised concerns about credential handling. Meta disputed that characterization, saying Muse cannot see users’ passwords or payment details.
Walmart announced Agent Profiles and a Walmart Agent Gateway on October 21. The gateway is designed to let approved agents search inventory and make purchases through a structured connection, rather than acting through ordinary web pages. Agent Profiles add controls such as spending limits. The announcements show different retailer policies, but do not establish how widely the Walmart tools are available or how many customers are using 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 Transaction
An AI agent can become the shopper’s point of contact, deciding what to compare and where to buy. If the agent directs a customer to one retailer or completes a purchase through a retailer’s own assistant, the shop may have less direct contact with that customer. That affects brand relationships, product discovery and customer data, as well as who can influence the final choice.
The economics also reach beyond retailers. Meta has said it expects to earn a small fee from transactions, while the source report says Amazon generated more than $68 billion in advertising revenue last year, citing GeekWire. Those figures describe different business models and do not establish what either company will earn from agent shopping. They illustrate why control over the path to purchase matters: agents may change how shoppers find products and how companies pay to reach them.
There is a separate issue for publishers and affiliate businesses. The source report says Amazon Associates commissions depend on customers following tracked links. If an agent reads a review but completes a purchase without using that link, the referral may not receive credit. The scale of any such effect is not established by the figures provided, but it raises a practical question about attribution as shopping moves from clicks to agent-mediated transactions.
For shoppers, convenience does not remove the need to know who is authorized to spend, how much an agent can spend, and which retailer’s rules apply. Purchase approval, spending caps and payment controls can help limit the consequences of an agent acting on a user’s behalf. The source report also describes Meta’s use of Link by Stripe for one-time-use cards, though it does not independently assess the security of that arrangement.
Popular right now
Ninja DualZone Smart Air Fryer, 10 Qt, 6-in-1, DZ950
LG 85-inch QNED75B Mini LED Smart TV Bundle with HDMI and CPS Protection Plan
TCL 85 Designer Series QD-Mini LED TV (Model 85A400U) with Google Smart TV
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
The access conflict is part of a short sequence of product and policy moves described in the source report. Meta launched Muse on September 8, 2026, presenting it as an agent able to browse, fill in forms and act for users. Amazon blocked the agent on September 20, saying it violated the retailer’s Conditions of Use. Meta’s response was that Muse had no visibility into passwords or payment methods.
Meta later announced retail connections involving Best Buy, Gap, Sephora, Walmart, Wayfair and Expedia. Other companies have pursued structured ways to let software interact with commerce systems. Shopify has tools for agent checkout, while Walmart’s October announcement introduced an agent gateway. These approaches differ from an agent using a normal browser, where the retailer may have difficulty distinguishing the software from a human customer.
Companies are also developing shared protocols to describe how agents identify themselves, interact with stores and handle checkout. The report describes Personal Agent Protocol (Poppy) as a draft effort involving Meta, Sierra, Walmart and Shopify, among others. It also lists Google’s Universal Commerce Protocol and OpenAI and Stripe’s Agentic Commerce Protocol. Their existence does not mean retailers or agents have adopted a common standard: they cover different parts of shopping and remain at different stages of development.
Early usage data also calls for caution about claims that agents are already completing most online purchases. The source report cites HUMAN Security data from September showing 81% of agent requests went to product and search pages, compared with 2.4% to checkout and payment pages. That describes the observed requests in that data set, not the overall share of purchases made by AI agents.
Adoption, Access and Accountability
The reported 10 million Muse setups are company-reported, and the source material gives no independent measurement of active users, completed purchases or repeat use. Meta’s claim that shopping is Muse’s most common use and its reported average purchase value also lack a stated methodology in the supplied material.
It is not yet clear how many retailers will permit outside agents, what conditions they will impose, or whether Walmart’s gateway will become broadly available. Amazon’s stated concerns and Meta’s rebuttal remain competing accounts; the supplied material does not independently establish how Muse handles credentials in practice. The status of competing protocols is also unsettled, and no common standard is confirmed.
There is not enough evidence here to determine whether agents are reducing affiliate revenue, changing advertising spending or increasing purchases overall. The cited request data measures activity on particular page types, not the full market. Consumer protections, responsibility for mistaken purchases and the treatment of returns also remain open questions in the material provided.
Retailers Set the Next Rules
The next developments to watch are whether Walmart’s gateway moves from announcement to wider use, whether Meta adds more retailer connections, and whether Amazon changes its policy on outside agents. The source report gives no rollout timetable for the Walmart tools or a date for any change by Amazon.
Protocol work may clarify how an agent identifies itself, what a shopper has authorized and how a purchase is handed to a retailer. But the competing initiatives still need merchant participation and workable payment and checkout arrangements. For shoppers and businesses, the near-term practical test will be whether agents can complete transactions reliably while making permissions, payment terms and returns responsibilities clear.
Key Questions
What happened between Meta’s Muse and Amazon?
Amazon blocked Muse in September 2026. Amazon said the agent had not identified itself and raised concerns about credential handling; Meta said Muse cannot see users’ passwords or payment details.
How is Walmart approaching AI shopping agents?
Walmart announced Agent Profiles and an Agent Gateway. The gateway is intended to let approved agents search inventory and make purchases through a structured connection, with profiles that can include spending limits.
How many people are using Muse?
Meta said more than 10 million people in the United States and Canada set up Muse in about six weeks. The figure is company-reported, and the supplied material does not say how many were active users or completed purchases.
Are AI agents already making most online purchases?
The evidence cited does not show that. HUMAN Security data reported in the source material found 81% of observed agent requests went to product and search pages and 2.4% to checkout and payment pages in September. Those figures concern requests in that data set, not the overall share of purchases.
What should shoppers check before letting an agent buy?
Shoppers should check what the agent is allowed to purchase, whether it requires approval, any spending limits, the payment method and which retailer’s return policy applies. The source material does not provide a single set of protections that applies across services.
Source: ThorstenMeyerAI.com
