🔍 Read the full analysis: Who Gets The Customer When An AI Agent Shops? 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
Meta says more than 10 million people in the U.S. and Canada set up its Muse agent in about six weeks, with shopping its most common use. Walmart is building tools for approved agents, while Amazon has blocked Muse; the competing approaches leave open who controls the purchase and receives credit for it.
Meta says more than 10 million people in the U.S. and Canada set up its Muse agent in roughly six weeks, making shopping its most common use, while Walmart is opening dedicated tools to approved agents and Amazon has blocked Muse. The split puts retailers, technology companies and publishers in a dispute over who controls the customer relationship—and how a purchase is attributed—when software does the shopping.
At a Meta third-quarter earnings call on October 21, the company said Muse users’ average purchase was worth more than $200. Those are Meta’s own figures; the company did not provide independent verification in the supplied report. The same week, Walmart announced Agent Profiles, which can set permissions such as spending limits, and an Agent Gateway for connecting approved agents to its shopping services.
Amazon is taking a different approach. A Verge reporter’s test of Muse found that it compared granola-bar prices at Sam’s Club, Walmart and Costco, but said Amazon had blocked it from checking prices there. Amazon says Muse did not identify itself as an agent and that it appeared to capture and store customer credentials. Meta’s launch post says Muse has no visibility into users’ passwords or payment methods. The accounts differ, and the supplied material does not independently resolve the dispute.
Other developments point to a broader effort to set rules for agent purchases. Shopify has introduced structured tools that let browser agents complete checkout, while Meta and Sierra have published a draft Personal Agent Protocol, known as Poppy. A separate commerce protocol from Google and Shopify, and OpenAI and Stripe’s Agentic Commerce Protocol, are among the projects seeking to connect agents and merchants. Their existence does not mean retailers have adopted a common system.
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
Control of the Agent Purchase
The dispute affects who gets to mediate a sale. If an assistant finds products, chooses a retailer and initiates checkout, it may become the shopper’s main interface—even when the retailer supplies the product, fulfilment and customer service. Meta chief executive Mark Zuckerberg said the company expects to earn a small fee from transactions, while the source report says Amazon received more than $68 billion from advertising last year, citing GeekWire. Those figures reflect different business models, but both point to the commercial value of influencing what shoppers see and buy.
There are also implications for publishers and affiliate businesses. The source report says Amazon Associates commissions depend on tracked customer clicks. If an agent reads a review but purchases through a different route, the referral may not be recorded. The extent of that effect is not established by the material provided, but it highlights a practical question: how will merchants credit recommendations and pay for referrals when an agent, rather than a person, completes the journey?
For shoppers, easier purchasing comes with choices about permission, payment and returns. A spending cap or purchase-by-purchase approval can limit what an agent is allowed to do. Users may also need to know which seller’s return rules apply and how to dispute a transaction if the agent selects the wrong item.
Popular right now
TCL 85-Inch QM6K Series QLED Mini LED 4K HDR Smart Google TV
KitchenAid Artisan 5-Qt Tilt-Head Stand Mixer, Cardamom
iRobot Roomba 105 Vac Robot Vacuum
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
Agent shopping can happen in several ways. An agent may operate a browser like a person, using pages designed for human shoppers. That approach works across many websites, but can make it harder for a retailer to distinguish an agent from its customer. Meta says Muse runs through a dedicated virtual machine in its cloud and uses Link by Stripe for payment with a one-time-use card.
A second model gives agents a retailer-approved connection. Walmart’s Agent Gateway uses MCP, or Model Context Protocol, to let approved agents search products and make purchases through a structured interface rather than clicking through ordinary web pages. Shopify has also provided checkout tools for agents. Shared protocols aim to define how an agent identifies itself, what the customer has authorized, and how a transaction moves between services.
The timeline has moved quickly: Meta launched Muse on September 8; Amazon blocked it on September 20; Meta announced retail connectors with several companies three days later; and Walmart announced its agent tools on October 21. The protocol push has also met limits. OpenAI and Stripe introduced Instant Checkout in ChatGPT using their protocol in September 2025, but OpenAI said in March 2026 that it was stepping back from that initial checkout approach to focus on product discovery and let merchants use their own checkout experiences.
“Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
— Amazon, as quoted in the supplied report
Adoption, Attribution and User Protections
It remains unclear how many Muse users have completed purchases through the agent, how Meta calculated the reported average purchase value, and how representative its user figures are. The supplied account also does not establish whether Walmart’s gateway will be widely used by independent agents or how merchants will set terms for access.
There is no settled industry-wide method in the source material for crediting referrals or assigning responsibility when an agent makes a purchase. Amazon and Meta also disagree about Muse’s access and treatment of credentials. The material gives no independent technical finding on those claims, and it does not specify how disputes, refunds or mistaken purchases will be handled across different agent and retailer systems.
Retailers Set Access Rules
The next test is whether retailers and agent providers adopt the new gateways and protocols in real shopping journeys, rather than only announcing compatibility. Watch for details on which agents Walmart approves, what permissions Agent Profiles let customers set, and whether merchants define clear rules for agent identification and checkout.
Meta’s reported user and transaction figures may also provide a measure of how often people move from product research to completed purchases. Until companies publish comparable data and clearer terms for payments, referrals and returns, who gets the customer—and who gets paid will remain unsettled.
Key Questions
What is the main news about AI shopping agents?
Meta reports rapid early adoption of its Muse agent, while Walmart is building tools for approved agents and Amazon has blocked Muse. The different policies show that access to retailers is still controlled by individual companies.
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 says Muse cannot see users’ passwords or payment methods. The supplied material does not independently settle the disagreement.
How does Walmart’s Agent Gateway work?
Walmart says its gateway uses MCP to let approved agents connect to shopping services through structured tools, rather than navigating the retailer’s ordinary pages. Walmart also announced Agent Profiles with controls such as spending limits.
Who gets paid when an AI agent buys something?
That depends on the retailer, agent and transaction route. Meta has said it expects to take a small fee from transactions. The supplied report says affiliate commissions can depend on tracked clicks, but there is no common attribution approach described for agent purchases.
What should shoppers check before letting an agent buy?
Review what the agent can purchase, set a spending cap or require approval, and check the payment method and retailer’s return policy. These safeguards matter because the supplied material does not describe a consistent process for handling mistakes or disputes across services.
Source: ThorstenMeyerAI.com
