Walmart opens Agent Gateway for AI shopping agents

Walmart is moving toward agentic commerce with a new Agent Gateway for AI shopping agents, according to TechCrunch. While Amazon has started blocking AI agents from buying for shoppers, Walmart says approved agents will be able to connect, search inventory and make purchases with customer controls in place.

For creators, marketers and small businesses watching AI commerce, the important point is not just that a large retailer is experimenting with agents. It is that Walmart is trying to define a permissioned model for AI-assisted buying, with account-level profiles, spending limits and rules for what agents can access.

Walmart’s new gateway for agentic shopping

Walmart announced the Agent Gateway on Wednesday, TechCrunch reported, describing it as a new MCP server for approved agents. The system is meant to let agents connect to Walmart using the model context protocol, a protocol originally created by Anthropic for linking AI agents with outside data sources, tools and workflows.

In Walmart’s case, the company says the gateway will let an AI agent search across the retailer’s online inventory and complete purchases for the person using that agent. The technology will support shopping at both Walmart and Sam’s Club, according to remarks from Walmart president and CEO John Furner at the Masters of Scale Summit in San Francisco.

Walmart had already moved in this direction with a partnership involving Meta’s Muse and integrations with popular AI apps including ChatGPT and Google Gemini, according to TechCrunch. The new gateway appears to formalize how outside agents are allowed to interact with Walmart’s commerce systems.

Profiles, approvals and spending limits

The most practical part of Walmart’s announcement is Agent Profiles, which the company says will roll out next month. These profiles will let customers authorize each AI agent they use inside their Walmart account, so agent-initiated purchases can be separated from purchases made directly by the shopper.

Walmart says customers will be able to decide how much control an agent has. They can require approval for every transaction or allow purchases up to a spending limit they set. The company compares the idea to adding an authorized user to a credit card, where that person’s transactions can be identified and capped.

The setup process, according to Walmart, begins the first time a shopper uses a particular agent through the Agent Gateway. The shopper authorizes and links an Agent Profile to a Walmart account, then chooses the relevant controls or spending limits. Walmart says profiles can also be deleted at any time, and a Walmart+ membership is not required.

For everyday users, this matters because it creates a middle ground between full manual shopping and handing broad authority to an AI tool. For teams experimenting with AI assistants for procurement, restocking or household buying, the approval layer may be the difference between a useful workflow and an unacceptable risk.

Why retailers are split on AI agents

Walmart’s approach contrasts with other online businesses that have locked out AI agents, including Amazon and some travel and booking sites, according to TechCrunch. The tension is becoming familiar: agents can make digital tasks easier for users, but websites still need to control identity, payments, fraud risk and customer intent.

That divide is part of a broader shift we have been tracking, including reports on websites pushing back on AI agents, Shopify opening its catalog to more AI shopping agents, and Amex introducing an intent framework for AI shopping agents. Walmart’s move adds a major retail example of a more controlled, permission-based path.

Walmart also says its gateway can help customers shop for their usual brands or items without sharing personal information or purchase history directly with the agent. That distinction could be important for shoppers who want personalization but do not want every assistant they try to receive a broad view of their past buying behavior.

What marketers and small businesses should watch

Walmart is also introducing terms for AI agents, setting rules for how they operate and what they can do on the site. The company says agents cannot act beyond what the shopper intends, cannot access customer payment information, and can only access the information needed for the task. Agents that do not follow the terms will be blocked.

One customer-facing detail stands out: Walmart says its returns process applies, including when an agent does not behave as intended. That could reduce some hesitation around early agentic shopping, because shoppers may be more willing to test agents if unwanted outcomes still fall under the retailer’s normal return process.

For creators and marketers, this points to a future where product discovery may happen through assistants as well as search boxes, ads and storefront pages. The source does not say how Walmart will rank or present products to agents, so it is too early to draw tactical conclusions. But teams should watch how agent access, product data and customer permissions develop, because AI shopping systems could become another route between buyers and products.

For small businesses, the immediate takeaway is more cautious: agentic commerce is not just an AI feature, it is also a trust and operations problem. Walmart’s bet is that profiles, approvals, limits and enforceable agent terms can make automated buying acceptable to mainstream shoppers.