🔍 Read the full analysis: Amazon, Walmart And The Question Of Who Owns An AI 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
Walmart announced an Agent Gateway and spending-controlled profiles for AI shopping agents on Oct. 21, while Amazon has blocked Meta’s Muse agent from accessing its site. The dispute concerns who controls the customer relationship, how purchases are attributed and what protections apply when software shops for a person.
Walmart is opening a route for approved AI shopping agents to search its products and make purchases, while Amazon has blocked Meta’s Muse from accessing its site. The contrasting moves, reported in October 2026, put retailers and technology companies in a dispute over who controls shoppers’ transactions, data and relationship with sellers.
Walmart announced its Agent Gateway and Agent Profiles on Oct. 21. The gateway is based on the Model Context Protocol, or MCP, and is intended to let approved agents search inventory and buy through a structured connection rather than interacting with retail pages as a person would. Profiles can set spending limits. The available source does not detail every control, participating agent or rollout date.
Meta’s Muse uses a different approach: it operates a browser in a dedicated virtual machine in Meta’s cloud. Amazon blocked Muse on Sept. 20. Amazon said the agent had not identified itself, that Meta had not told the company Muse would shop on its site, and that Muse appeared to capture and store customer credentials. Meta’s launch post said Muse had no visibility into users’ passwords or payment methods. Those are competing accounts; the supplied material does not independently resolve them.
Meta said on its Oct. 21 earnings call that more than 10 million people in the United States and Canada had set up Muse in roughly six weeks, that shopping was its most common use and that average purchases exceeded $200. These are company-reported figures, not independently verified usage data. Meta CEO Mark Zuckerberg said the company expected to earn a small fee from transactions, a model that could put an agent provider between shoppers and retailers.
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 Purchase
The dispute is about more than whether a bot can browse a product page. An agent that selects items, compares prices and completes an order can become the shopper’s main point of contact with a retailer. Retailers may lose control over how customers discover products, what offers they see and which service receives the purchase. Agent providers, by contrast, may gain leverage if users rely on their tools to choose where to buy.
That shift also affects advertising and referral income. The source reports that Amazon earned more than $68 billion from advertising last year, citing GeekWire. If agents make decisions without showing shoppers ads or clicking tracked links, existing ways to attribute sales may not apply. Amazon Associates commissions, for example, rely on a customer following a tracked link. The scale of any resulting revenue change is not established by the material.
For consumers, convenience does not settle questions about purchase approval, spending limits, payment security and returns. A purchase made through an agent may involve the agent provider, payment service and retailer, while responsibility for resolving a mistaken order or return may depend on the terms governing that transaction.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Jackery HomePower 1000 Plus v2 Portable Power Station
DJI Osmo Pocket 3 Filter Bundle – Vlogging Camera with 1 CMOS Sensor and 4K/120fps
As an affiliate, we earn on qualifying purchases.
Three Routes to Agent Checkout
The current dispute reflects three approaches to AI-led shopping. A browser agent acts through pages designed for people, potentially using a shopper’s logged-in session. A retailer can instead provide a dedicated interface, such as Walmart’s MCP-based gateway, so an approved agent can use structured shopping tools. A third approach is to develop shared protocols that describe how agents, merchants and payment services exchange information.
The timeline has moved quickly. Meta launched Muse on Sept. 8, describing it as able to open a browser and fill out forms. After Amazon blocked it, Meta announced retail connectors at its Connect event on Sept. 23, including Walmart, Best Buy, Gap, Sephora and Wayfair. Shopify also introduced structured tools for browser-agent checkout in September. On Oct. 6 and 9, Meta and Sierra announced the Personal Agent Protocol, also called Poppy; its draft was published with additional design partners.
Other commerce standards are also in play. The source describes Google’s Universal Commerce Protocol as published and the OpenAI- and Stripe-backed Agentic Commerce Protocol as open source. Yet protocols do not guarantee broad access or adoption. OpenAI stepped back from its initial ChatGPT Instant Checkout approach in March 2026, saying the initial version did not offer the flexibility it wanted and that it would focus on product discovery while merchants used their own checkout experiences.
“Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
— Amazon, in the message shown to Muse users
Access, Liability and Adoption
It is not yet clear how widely Walmart’s gateway will be available, which agents will qualify for access or how spending limits and purchase approvals will work in practice. The source also does not establish whether Amazon will permit any outside shopping agents through a separate interface, or whether the two companies will agree on common access rules.
Claims about Muse’s handling of credentials remain disputed in the supplied material: Amazon raised the concern, and Meta denied that Muse could see passwords or payment methods. Independent technical findings are not provided. Nor is there evidence here that the announced protocols have achieved broad adoption or that agent shopping is already shifting a substantial share of retail sales. Meta’s usage and purchase figures are its own reports, and the source does not specify how average purchase value was calculated.
Payment disputes, returns, mistaken orders and commission attribution also remain open questions. The source notes that agent requests in HUMAN Security’s September data were mostly directed to product and search pages, with 81% going to those pages and 2.4% to checkout and payment. It does not provide enough detail here to generalize that snapshot to all agents or retailers.
The Next Test Is Checkout
The next test will be whether Walmart’s announced tools move from a retail access plan to widely used agent checkouts, and whether other merchants provide comparable routes. Watch for details on which agents can connect, what customers must approve and how transactions are handled when something goes wrong.
More evidence is also needed on how shoppers use these services beyond setup figures: how often agents complete purchases, whether customers trust their recommendations and how sales are credited to retailers, publishers and agent companies. Until those details emerge, the clearest confirmed development is a split in strategy: Walmart is building an access path, while Amazon has barred Muse and disputes Meta’s account of how the agent operates.
Key Questions
What did Walmart announce?
On Oct. 21, Walmart announced an Agent Gateway based on MCP and Agent Profiles that can set spending limits. The source says the gateway is intended to let approved agents search inventory and buy through a structured connection.
Why did Amazon block Meta’s Muse?
Amazon said Muse had not identified itself, that Meta had not disclosed the shopping use and that the agent appeared to capture and store customer credentials. Meta said Muse could not see users’ passwords or payment methods. The supplied material does not independently resolve the disagreement.
How does an AI shopping agent buy something?
It can use a browser like a person, connect through a retailer’s agent-specific interface, or rely on shared protocols that coordinate agent, merchant and payment services. Each route raises different questions about access, oversight and responsibility.
How many people have set up Meta’s Muse?
Meta said on its Oct. 21 earnings call that more than 10 million people in the United States and Canada had set it up in about six weeks. The figure is company-reported and does not, by itself, show how many users completed purchases.
What should shoppers watch for?
Check whether purchases require approval, whether the agent has a spending cap, how the payment is protected and which retailer’s return policy applies. The source says practical details about protections and responsibility are still developing.
Source: ThorstenMeyerAI.com
