Walmart Announces AI Agent Shopping Controls, Complete with Spending Limits

Amazon, meanwhile, continues to block third-party agents.

To power this initiative, Walmart launched three major updates:

• Agent Profiles, allowing agents to buy on a user's behalf subject to set spending limits;
• Agent Gateway, a dedicated, direct connection built specifically for agents; and
• A set of Purchasing Rules tailored exclusively for AI agents.

According to Walmart, Agent Profiles will roll out within the next month, while the Agent Gateway is currently in a limited beta for select companies and developers—with no firm launch date or confirmed participant list announced yet.

This stands in stark contrast to Amazon's stance. The e-commerce giant blocked Muse, Meta's shopping agent, in September—less than two weeks after Muse launched.

Behind the divergence: Why Walmart and Amazon are taking opposite approaches to AI agents

The two retail giants make money in different ways, which shapes how they view agents.

As an analysis on X pointed out: "If your margin lives in the interface, agents are a threat."

Amazon's multi-billion-dollar ad business relies heavily on the search results page, where shoppers scroll past sponsored listings.

An AI agent doesn't scroll—it reads data, compares options, and executes the buy. Every time an agent buys, that search results page is completely bypassed.

That said, Walmart runs its own ad business, Walmart Connect, and is currently pitching brands directly ahead of the holiday season. How Walmart plans to balance AI agents with its advertising revenue remains to be seen.

The real driver behind Walmart's move likely comes down to scale:

Muse reached over 10 million downloads in the US and Canada in just 1.5 months; shopping was the most frequent action users took; and the average order value exceeded $200.

SSeijin Jung@SeijinJung

15 muse takeaways from Meta's q3 call today:

1/ 10m+ muse setups in US/CA in 1.5 month

2/ shopping is the #1 use case, 25%+ of weekly actives

3/ average purchase size is $200+

4/ week 1 volumes were so high they had to raise credit limits with stripe

5/ plaid said they had not seen connector activity like this on any other service

6/ zuck expects muse ARPU to beat family of apps over time because of commerce,…

Oct 21, 2026 · View on X

This is no longer a niche group of early tech adopters—it is becoming a sales channel with higher average order values than many traditional product categories.

Walmart also cited a purely technical reason for building a dedicated gateway. Currently, agents must shop via web browsers, navigating UI steps designed for human eyes, which makes the process slow and prone to breaking mid-checkout.

By building a dedicated gateway, Walmart bypasses those steps while establishing clear guardrails to block non-compliant agents.

For Shopify stores, the gateway for AI agents opened on September 21

A merchant in Japan tested Shop MCP and shared that agents like Claude or ChatGPT were able to discover products on their Shopify store and complete purchases using Shop Pay. Orders flow directly into the store admin labeled as AI agent orders. Merchants don't need to install anything extra, though customers must pay using a card already saved in Shop Pay.

However, on Shopify, being accessible to AI agents doesn't guarantee your products get picked

This is where many sellers might face a reality check. Having agentic storefronts enabled simply means your catalog is indexable by agents—it does not mean your products will automatically be recommended to shoppers.

An analytics firm ran 50 test queries and found that Muse's internal system prompts systematically favored well-known merchants when ranking top-5 products, according to a post on r/ecommerce.

While this sample size is small and Meta hasn't commented, it raises a critical question for store owners: how do you optimize your store so AI agents actually recommend your products?

Walmart takes a more optimistic view. According to Manish Joneja, SVP of Walmart Global Marketplace, agents open doors for lesser-known brands, but they also raise the bar for product data, pricing, inventory accuracy, and delivery promises.

Both perspectives converge on one key point: what matters most isn't how famous your brand is—something you can't drastically change in a single quarter—but the quality and completeness of the data you provide.

Four things Shopify store owners should audit right now

Start with the low-hanging fruit inside your admin panel:

• Check your settings for agentic storefronts and strike-through pricing display options. If your store frequently inflates list prices to display steep discounts, keep in mind that agents read those raw numbers—and compare them directly against competitor pricing.
• Review the core data fields highlighted by Joneja: price, stock levels, technical specs, and estimated delivery times. If a product lacks technical specs, agents have nothing to evaluate and will likely skip your listing before a shopper ever sees it.
• Run an end-to-end test using an AI agent. Have it search, compare, and actually buy a product from your store, then note where it encounters friction. This is the easiest way to see your store through the eyes of an AI agent and spot product descriptions that make sense to humans but fail with machines.
• Track agent-driven orders separately in your reporting. Shopify already labels these orders, so treat them as a distinct sales channel with its own average order value and return rates.