🔍 Read the full analysis: AI Shopping Is Reshaping The Retailer-Customer Relationship on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- 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 shopping agent in about six weeks, though those figures come from the company. Walmart has announced tools to let approved agents search and buy, while Amazon has blocked Muse. The competing approaches raise unresolved questions about customer data, payments, retailer access and how publishers receive referral credit.
AI shopping agents are drawing a sharper divide between retailers that are building ways to serve them and those restricting access. Meta says more than 10 million people in the U.S. and Canada set up its Muse agent in roughly six weeks, while Walmart has announced a gateway for approved agents and Amazon has blocked Muse from its site. The contest could determine who controls the shopping relationship when software, rather than a person, searches and buys.
Meta shared its Muse figures during its October 21 earnings call, saying shopping was the agent’s most common use and that the average purchase was worth more than $200. These are company-reported figures; the supplied material does not include independent verification or a breakdown of completed purchases. Muse can browse and fill out forms from a dedicated virtual machine in Meta’s cloud, with payments made through Link by Stripe using a one-time card, according to the source report.
Walmart announced Agent Profiles and a Walmart Agent Gateway on October 21. The gateway is based on the Model Context Protocol and is intended to let approved agents search inventory and make purchases through a structured connection. Agent Profiles add controls such as spending limits. Amazon, by contrast, has blocked Muse and says outside agents must identify themselves under its Conditions of Use. The competing policies leave shoppers with different levels of agent access depending on the retailer.
Agent activity does not yet mean that most users are letting software complete a purchase. The source cites HUMAN Security data from September showing that 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, not all shopping agents or shoppers. The gap suggests that product research is more common than automated checkout, at least in the measured traffic.
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 Agent-Led Sale
The dispute concerns more than whether an automated browser can reach a product page. A shopping agent can shape which products a shopper sees, compare prices across sellers and potentially complete a purchase. If the agent becomes the main interface, retailers may lose some control over discovery and customer contact, while agent providers may gain influence over the route to a sale.
That shift also has consequences for advertising and referral businesses. The source reports that Amazon earned more than $68 billion from advertising last year, citing GeekWire, and says Meta has discussed taking a small fee from transactions. Affiliate programs present another pressure point: when an agent reads a review but buys without following a tracked link, the publisher that influenced the purchase may not receive credit. How retailers and agents will measure or share that value remains unsettled.
For shoppers, automated convenience comes with practical questions about authorization and responsibility. A spending cap or approval requirement can limit unexpected purchases, but shoppers also need to know which account is being used, what data the agent can access and which company handles returns or disputes. The source recommends purchase limits and one-time payment cards; their availability and terms depend on the service.
Popular right now
Shark AI Ultra Robot Vacuum with Self-Empty Base
Bose QuietComfort Ultra Earbuds (2nd Gen), Deep Plum
Ninja Foodi DZ201 8-Quart DualZone 6-in-1 Air Fryer
As an affiliate, we earn on qualifying purchases.
From Browser Bots to Retail Gateways
The current dispute developed quickly. Meta launched Muse on September 8, describing it as an agent able to open a browser, fill forms and act for a user. Amazon blocked it on September 20. Meta later announced retail connectors with companies including Best Buy, Gap, Sephora, Walmart and Wayfair. Walmart’s October gateway announcement followed a series of efforts to make agent-to-retailer interactions more structured.
There are three broad approaches in the material. A browser-based agent imitates a person using a website; a retailer gateway gives approved agents a structured route to inventory and checkout; and shared protocols seek common rules for communicating an agent’s identity and the shopper’s permission. Meta, Sierra, Walmart, Shopify and Stripe have backed a draft Personal Agent Protocol. Google and Shopify are associated with the Universal Commerce Protocol, while OpenAI and Stripe have published the Agentic Commerce Protocol.
These systems are not interchangeable, and a published protocol does not mean broad adoption. OpenAI, for example, stepped back in March 2026 from its initial ChatGPT Instant Checkout approach, saying it would let merchants use their own checkout experiences while it focused on product discovery. That change illustrates a challenge for the sector: connecting discovery to a reliable, merchant-approved purchase involves more than building a technical standard.
Access, Purchases and Accountability
Meta’s reported setup and purchase figures do not establish how many users completed a transaction through Muse, how frequently they use it or whether the activity is sustainable. The supplied material gives no independent audit of those figures. The September agent-request data also covers a specific dataset and should not be treated as a market-wide measure of shopping behavior.
It is also unclear how many retailers will adopt Walmart’s gateway or the emerging protocols, and what access each will grant. Amazon’s stated concerns about agent identification and credentials are its position; Meta has disputed the password and payment-access claim. The source does not establish how the parties resolved those differences, or whether customers will be able to use agents consistently across major retailers.
Other open questions include who bears responsibility for an incorrect order, how returns and refunds work when an agent purchases on a shopper’s behalf, and how commissions or advertising value will be attributed. The supplied material does not give a settled industry standard for consent, payment authorization or referral credit.
Retailer Rules and Protocol Adoption
The next indicators will be whether retailers put agent access into regular use, which protocols they support and what controls shoppers receive. Walmart’s gateway and Agent Profiles will be a practical test of whether structured access can support purchases while preserving retailer rules. The Personal Agent Protocol remains a draft in the source material, while other commerce protocols have different backers and scopes.
For consumers, the immediate step is to check an agent’s permissions and payment settings before authorizing purchases. For retailers and publishers, the unresolved work is to set clear rules for access, attribution and post-purchase service. The source material does not provide a date for a common standard or a resolution to the Amazon-Muse dispute, so the balance between open access and retailer control is still developing.
Key Questions
What is an AI shopping agent?
It is software that can help search for products and, depending on its permissions and the retailer’s systems, may take steps toward checkout or purchase. Muse is Meta’s example in this report.
Why did Amazon block Meta’s Muse?
Amazon said Muse was an unauthorized agent that did not identify itself and raised concerns about capturing and storing customer credentials. Meta said Muse cannot see users’ passwords or payment methods. The supplied material does not describe a resolution.
What is Walmart’s Agent Gateway?
Walmart announced a structured connection based on the Model Context Protocol that is intended to let approved agents search products and buy them. Walmart also announced Agent Profiles with spending controls.
Are AI agents already completing most online purchases?
The cited HUMAN Security data does not show that. In its September dataset, 2.4% of agent requests went to checkout and payment, compared with 81% to product and search pages. Those figures apply to that dataset, not the entire market.
What should shoppers check before using an agent?
Check purchase approvals, spending limits, payment methods, account access and the retailer’s return rules. The available controls differ by agent and merchant, and shoppers should confirm what the service is authorized to do.
Source: ThorstenMeyerAI.com
