🔍 Read the full analysis: Are AI Agents Customers, Or Are Retailers Still In Charge? 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 United States and Canada set up its Muse personal agent within about six weeks, with shopping its most common use. Walmart is creating a channel for approved agents, while Amazon has blocked Muse and says outside agents must follow its terms. The emerging dispute concerns access, payment, customer data and which company controls the shopping relationship.
Walmart is opening a controlled route for AI shopping agents while Amazon has blocked Meta’s Muse, a divide that has sharpened as Meta reports rapid early adoption of its shopping assistant. The dispute will shape whether people can delegate purchases to software, and how retailers, payment firms and publishers retain control of transactions.
On October 21, Walmart announced Agent Profiles and a Walmart Agent Gateway. The gateway is designed to let approved agents search products and make purchases through a structured connection, rather than acting like a person clicking through a retail website. Agent profiles can set spending limits, according to the source report. Walmart’s approach offers access, but on terms the retailer can define.
Meta’s Muse takes a different route when a retailer has no agent-specific connection: it can browse and interact with websites through a computer running in Meta’s cloud. Amazon blocked Muse on September 20. Amazon said the agent had not identified itself and appeared to capture and store customer credentials; Meta’s launch post said Muse had no visibility into users’ passwords or payment methods. Those are opposing company statements, not independently established findings in the supplied material.
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 its most common use and the average purchase was more than $200. These are Meta’s figures; the source does not give an independent measure of completed agent purchases or a comparison baseline.
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 Checkout
The contest is about more than whether an automated assistant can find a lower price. A retailer’s own agent connection can keep product discovery, checkout and order information within a system the retailer manages. A general-purpose agent may instead become the shopper’s main point of contact, potentially influencing where a purchase happens and which company receives transaction data or fees.
That change could affect retailers, payment providers and publishers. The source report says Meta expects to earn a small fee from transactions, while Amazon generated more than $68 billion in advertising revenue last year, citing GeekWire. The figures point to commercial stakes, but they do not establish how much agent-led shopping will shift advertising or sales. Affiliate publishers also face a possible attribution problem: if an agent reads a review but completes a purchase without using its tracked link, a publisher may not receive a commission under a click-based arrangement.
For shoppers, access and convenience come with questions about spending authority, credentials and returns. A retailer-controlled gateway can impose rules on which agents may transact. A browser-based agent may work across more sites, but retailers can object to automated access, and users need to know what they have authorized and which merchant’s policies apply.
Products Worth Considering
Sonos Era 100 SL Compact Microphone-Free Speaker, Black
Kindle Paperwhite Kids (2024), 16 GB, 7-inch Display, Cyber City Cover
Sonos Play Portable Wi-Fi and Bluetooth Speaker, White
As an affiliate, we earn on qualifying purchases.
Six Weeks of Competing Approaches
Meta introduced Muse on September 8, describing it as a personal agent able to open a browser, fill out forms and act for a user. Its cloud-based setup is called Muse Secure VM; the source says payments use Link by Stripe and a one-time-use card. On September 20, Amazon restricted Muse. Three days later, Meta announced retail connectors with Best Buy, Gap, Sephora, Walmart, Wayfair and Expedia.
The broader industry is testing structured ways for agents to interact with businesses. Shopify has offered browser-agent checkout tools, while a draft Personal Agent Protocol, also called Poppy, was announced by Sierra and Meta with partners including Walmart, Shopify and Stripe. Google and Shopify’s Universal Commerce Protocol and the OpenAI-Stripe Agentic Commerce Protocol are other efforts described in the source. Their scopes and maturity differ; Poppy was at draft 0.1 on October 9, while the source describes UCP as published and ACP as open source.
Protocol development does not by itself guarantee broad adoption or reliable purchases. The source notes that OpenAI stepped back from its initial ChatGPT Instant Checkout approach in March 2026, saying it would let merchants use their own checkout experiences while OpenAI focused on product discovery. That history illustrates the challenge: connecting an agent to a store is not the same as resolving control over checkout and the customer relationship.
“Muse has no visibility into people’s passwords or payment methods.”
— Meta, in its Muse launch post
Adoption and Access Remain Unsettled
Meta’s user and purchase figures are company-reported, and the source provides no independent audit or detailed definition of an “average purchase.” It is not clear how many Muse users completed purchases through agents, how frequently they do so, or what share of retail sales might eventually involve agents.
The source also does not establish how the competing protocols will work together, how many retailers will support them, or what safeguards will be standard across services. Amazon’s concerns about Muse and Meta’s response remain competing claims in the material. Access may also vary by retailer: the source reports that Muse encountered Target’s bot check on one occasion and passed it on another, and that users had difficulty with Delta and Amtrak.
It remains unclear how shoppers will resolve disputes over unauthorized purchases, returns or personal information when an agent works across multiple companies. Those details may depend on the agent’s permissions, the retailer’s rules and the payment arrangement used for a transaction.
Retailers Set the Next Terms
The next test is whether Walmart’s gateway and agent profiles move from an announced system to routine shopping activity, and whether other retailers adopt comparable connections. The source describes Poppy as a draft and lists a wider group of design partners; participation does not confirm that each partner has launched a working service.
For consumers, the practical steps are to check an agent’s permissions before use, set a spending cap or require approval for each purchase, and understand how payment and return policies apply. Retailers and agent providers will need to make clear which party is responsible for a transaction and what data is shared. No common resolution or adoption timeline is confirmed in the supplied material.
Key Questions
What is the main difference between Walmart’s and Amazon’s approaches?
Walmart announced a controlled gateway for approved shopping agents and profiles with spending limits. Amazon blocked Meta’s Muse and said outside agents must comply with its Conditions of Use.
How many people have set up Meta Muse?
Meta said on October 21 that more than 10 million people in the United States and Canada had set it up in about six weeks. This is Meta’s own figure, not an independently verified count in the supplied material.
Why did Amazon block Muse?
Amazon said Muse had not identified itself as an agent and appeared to capture and store customer credentials. Meta said Muse could not see users’ passwords or payment methods. The supplied source does not independently establish which account is correct.
Do AI shopping agents complete most purchases?
The source cites HUMAN Security data from September stating that 81% of agent requests went to product and search pages, while 2.4% went to checkout and payment. The figures describe requests in that dataset, not all shopping agents or all retail purchases.
What should shoppers check before allowing an agent to buy?
Check what the agent can purchase, whether it requires approval, and whether a spending cap is available. Also review the payment method, data access and the retailer’s return policy before relying on automated checkout.
Source: ThorstenMeyerAI.com
