🔍 Read the full analysis: The Question Retailers Face When AI Becomes The Shopper on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Meta says more than 10 million people in the United States and Canada set up its Muse assistant within about six weeks, with shopping its most common use. Walmart is opening structured access to approved agents, while Amazon has blocked Muse, highlighting a contest over customer relationships, purchasing data and payment credit.
Meta’s Muse shopping agent has exposed a growing divide in how retailers handle AI buyers: Walmart is opening a controlled route for agents to search and purchase, while Amazon has blocked Muse from accessing its site. The dispute matters because an agent can shape what a shopper sees, where they buy and which company receives credit for the sale.
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. The company said shopping was the agent’s most common use and that the average purchase was more than $200. Those figures are Meta’s own metrics; the company has not provided independent verification in the source material.
Walmart announced an Agent Gateway and individual Agent Profiles, which can let approved agents search its inventory and make purchases with customer-defined spending limits. Amazon, by contrast, has barred Muse. In a message to users, Amazon said continued access by an unauthorized AI agent violated its Conditions of Use. Amazon also argued that Muse did not identify itself and appeared to capture and store customer credentials.
A test described by The Verge illustrates the practical limits of agent shopping. Asked where granola bars were cheapest per piece, Muse listed Sam’s Club, Walmart and Costco, then said it could not check Amazon because access was blocked. Meta has said Muse cannot see users’ passwords or payment methods. The source material does not resolve the disagreement over how the agent handled credentials.
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 Journey
The access dispute is about more than whether an agent can load a product page. Retailers want control over how their inventory is shown, how a purchase is completed and what customer information an outside service can use. An agent that compares products across stores could send a customer elsewhere; a retailer that admits only its own assistant could retain more influence over the sale.
There is also a financial question for advertising and referral businesses. The source reports that Amazon earned more than $68 billion from advertising last year, citing GeekWire, and that Meta aims to make money by taking a small fee from transactions. Meanwhile, affiliate publishers may lose commission when an agent reads a review and completes a purchase without using a tracked link. The scale of any such effect is not established in the material.
For shoppers, automation does not remove questions about consent or responsibility. A spending cap and purchase approvals can limit unintended orders, but users still need to know which retailer is the seller, whose return rules apply and how payment information is handled. Those details can differ when a transaction moves between an agent and a merchant.
Products Worth Considering
Bose TV Speaker (2nd Gen) Compact Soundbar with Bluetooth and HDMI Connectivity
Bose SoundLink Flex Bluetooth Speaker (2nd Gen), Portable Waterproof Speaker, Black
DJI Osmo Mobile 8 Standard Combo Phone Gimbal Stabilizer
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retailer Gateways
Muse launched on September 8 as an agent able to browse websites and fill out forms, using a dedicated virtual machine in Meta’s cloud and payments through Link by Stripe, according to the source. Amazon blocked it on September 20. Two days later, Meta announced retail connectors involving Best Buy, Gap, Sephora, Walmart, Wayfair and Expedia; Meta chief executive Mark Zuckerberg said the company expected to earn a small fee from transactions.
Retailers and technology firms are testing several ways to make agent purchases work. One is ordinary browser access, where an agent acts through pages made for people. Another is a retailer-controlled connection, such as Walmart’s gateway or Shopify’s structured tools for checkout. A third is a shared protocol intended to describe how agents, merchants and payment services communicate. These efforts are at different stages, and a published protocol does not by itself show that consumers are widely using it.
There are signs that adoption and access remain uneven. The source cites HUMAN Security data from September in which 81% of agent requests went to product and search pages, while 2.4% went to checkout and payment. Those figures describe requests in that dataset and period, not all agent activity or completed purchases. The source also notes that other companies have raised concerns about outside agents accessing customer accounts.
“Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
— Amazon, in a message shown to Muse users
Adoption, Access and Accountability
The available figures do not show how many Muse users completed purchases, how frequently they return or whether the reported average purchase reflects agent-driven transactions across all users. Meta’s adoption and purchase figures are company-reported, and the supplied material provides no independent measure of their accuracy.
It also remains unclear how widely Walmart’s new tools will be used, which agents will be approved and what spending controls will be available in practice. The source describes a crowded protocol landscape, but does not establish that one approach will become the common standard. The disagreement between Amazon and Meta over credentials is unresolved in the material, and it is not yet clear how responsibility will be divided when an agent makes a mistaken purchase or mishandles a user’s information.
Retailers Test Rules for Agent Buyers
The next test is whether access arrangements move from announcements and technical tools to reliable, completed purchases. Walmart’s Agent Gateway and profiles will show how much control a retailer can retain while allowing outside agents to shop. Other merchants may choose similar gateways, browser access or their own shopping assistants.
Shoppers and publishers will also be watching how agents handle approvals, payment, returns and referral links. For now, the practical picture is mixed: some sites offer structured ways for agents to interact, while others block them or challenge their access. The rules are still taking shape, and the next developments will clarify which systems merchants accept and what protections users receive.
Key Questions
What is the main news about AI shopping agents?
Walmart is opening a controlled route for approved agents to interact with its shopping services, while Amazon has blocked Meta’s Muse. The contrast shows retailers taking different approaches to automated shopping.
How many people have set up Meta Muse?
Meta said more than 10 million people in the United States and Canada set up Muse in about six weeks. This is a company-reported figure, and the supplied material does not include independent verification.
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 disputed the credential concern, saying Muse cannot see users’ passwords or payment methods. The supplied material does not resolve the disagreement.
Can shopping agents complete purchases?
Some tools are designed to support checkout, but access and capabilities vary by retailer. The source cites September data from HUMAN Security showing that 2.4% of agent requests in its dataset went to checkout and payment pages; that is a measure of requests, not a count of completed purchases.
What should shoppers check before using an AI agent?
Shoppers can review purchase approvals and spending limits, check how payment details are handled, and confirm which retailer sells the item and sets its return policy. Agent access does not automatically establish who is responsible for a purchase or return.
Source: ThorstenMeyerAI.com
