Walmart Is Building the Access Rules for Agentic Commerce

Walmart is opening its ecommerce infrastructure to outside AI shopping agents, but on terms designed to protect customer permissions, transaction data and control of checkout. The strategy offers multi-location brands an early model for treating agentic commerce as a governed channel rather than uncontrolled bot traffic.

Opening the Storefront Without Surrendering Control

Walmart is preparing for a shopping experience in which customers no longer need to visit Walmart.com, open its app or search digital aisles themselves. An AI agent could find products, compare options and complete a transaction on their behalf.

To enable that shift, the retailer has introduced Agent Profiles and Walmart Agent Gateway. Together, the tools will let customers shop at Walmart and Sam’s Club through approved third-party agents while giving the retailer a structured way to manage access. The capabilities are expected to begin appearing in customer accounts in November.

Agent Profiles will let customers authorize individual agents, monitor purchases and establish spending limits. Shoppers can require approval before a purchase, review an agent’s activity separately and revoke access. Agents will not receive customers’ payment-card information.

Agent Gateway provides the technical connection between outside assistants and Walmart’s commerce systems. Instead of navigating pages designed for people, an approved agent can use structured access to search products and complete permitted tasks. Walmart is effectively establishing an identity, permissions and transaction record for each participating agent.

Walmart and Amazon Choose Different Paths

The strategy contrasts with Amazon, which has blocked Meta’s Muse shopping agent, restricted access from other third-party assistants and sued Perplexity over allegedly unauthorized automated shopping activity.

Amazon has cited concerns involving customer credentials, data security and third-party interference with its shopping experience. Restricting outside agents also helps it protect its advertising business, recommendations and customer data.

Walmart is not offering unrestricted access. Agents must follow its pricing, returns and commerce policies, and noncompliant agents can be blocked. Only the customer information required to fulfill a specific request is supposed to be shared.

Walmart is therefore creating a permissioned channel rather than simply opening its website to bots. The approach could broaden distribution as consumers adopt assistants from Meta, OpenAI and Google, while Amazon is betting that shoppers will remain inside its own AI ecosystem.

Local Inventory Becomes Part of AI Discovery

For multi-location retailers, agentic commerce will depend heavily on accurate location-level information.

An agent buying groceries, household supplies or electronics needs more than a national product catalog. It must know whether an item is available nearby, which store can fulfill the order and whether pickup, delivery or shipping is available.

That puts additional pressure on inventory accuracy. An agent that recommends an unavailable product or routes an order to the wrong location creates a cancellation, substitution or service failure associated with the retailer’s brand.

Local prices, promotions and fulfillment capacity must also remain current. Multi-location companies already synchronize this information across websites, apps, listings and delivery platforms. AI agents add another channel that must receive reliable, machine-readable data.

Customer Ownership and Attribution Get Harder

The model raises questions about who controls the customer relationship. Walmart may process the transaction, but a third-party agent could control discovery, comparison and the final recommendation.

That changes the role of traditional ecommerce merchandising. Product-page placement, search rankings and sponsored listings become less influential when an agent creates the shortlist before the customer enters a retailer’s environment.

Retail-media attribution will also need to evolve. Brands will want to know why an agent selected one product over another, whether paid placements influenced the recommendation and which interaction ultimately produced the sale. Retailers must determine how sponsored products can participate without undermining trust in an assistant expected to represent the customer.

Walmart’s advantage is its combination of first-party data, thousands of stores and extensive fulfillment infrastructure. Agent Gateway could make those assets accessible through more consumer interfaces while keeping purchases inside Walmart’s systems.

A Governance Model for Multi-Location Brands

Most multi-location brands cannot build infrastructure on Walmart’s scale, but they still need policies before automated traffic reaches product catalogs, booking systems, loyalty accounts and checkout pages.

Those policies should define which agents are permitted, what information they can access, how consent is recorded and when purchases require approval. Brands will also need standards for agent identification, fraud prevention, returns and attribution.

Walmart’s approach recognizes that an AI agent is not simply another marketing channel. It is a new participant in the customer journey, one capable of searching, deciding and transacting. The brands that benefit will make their products and local availability accessible to trusted agents while retaining control of permissions and transaction rules