Walmart's AI Agent Gateway: How to Make Your Online Store Agent-Ready

On October 21, 2026, Walmart CEO John Furner used the Masters of Scale summit in San Francisco to announce a set of tools that let outside AI agents shop on behalf of customers. Furner called agent-led commerce "the next great evolution in retail." The announcement has three parts: the Walmart Agent Gateway, Agent Profiles, and a new set of terms that every AI agent must follow. It covers both Walmart and Sam's Club, and shoppers do not need a Walmart+ membership to use it.

The Agent Gateway is a server built on the Model Context Protocol, or MCP, the open standard that Anthropic originally created for connecting AI assistants to outside tools and data. Through the gateway, approved agents such as ChatGPT and Gemini can search Walmart's inventory and complete purchases for a user. Walmart describes it as a faster and more reliable path for agents than clicking through a website built for humans. It is starting as a beta for companies and developers, so access is limited for now.

Agent Profiles are due to arrive in customer accounts within about a month. Each AI agent a shopper authorises gets its own profile inside their Walmart account, so the agent's purchases are kept separate from the shopper's own. The customer can require approval for every purchase, or let the agent buy on its own up to a spending limit they choose. A profile can be removed at any time. Walmart has compared the idea to adding an authorised user to a credit card.

The new terms set the rules of the road. Agents must not go beyond what the shopper asked for, must not access payment card details, and may only collect the data they need for the task. Walmart says agents see the same prices as human shoppers, its normal returns process still applies when an agent gets an order wrong, and agents that break the rules can be blocked. Coverage has contrasted this with Amazon, which has moved to block third-party AI shopping agents from its site while promoting its own AI tools.

Why does this matter if you do not sell on Walmart? Because it is one of the clearest signals yet that a share of online shopping will be done by software, not people. When the world's largest retailer builds a dedicated front door for AI agents, it tells every brand and store owner that a new type of visitor is on the way. That visitor does not look at your banner images or your homepage design. It reads your product data, checks your stock, compares prices and policies, and then decides.

For a small or mid-size business, the practical question is whether an AI agent could understand your store if it arrived today. Many shops have product titles that are vague, sizes and specifications buried in images, stock levels that are out of date, and return policies that are hard to find. A human shopper might work around that. An agent is more likely to skip to a competitor whose information is clean and complete. Getting your product data in order is useful even if agent shopping grows slowly, because the same work also helps search engines and your own website chatbot.

Consider a direct-to-consumer skincare brand in India selling through its own Shopify store. Customers are already asking AI assistants which sunscreen suits oily skin or which products are free of a certain ingredient. If the brand's catalogue lists ingredients, skin types, sizes and prices as clear text, an assistant has a far better chance of recommending it accurately. The brand can also put a chatbot on its own site, trained on the same catalogue, so that human visitors get the same precise answers.

A second example is a grocery or pharmacy delivery business in Dubai or London with a high volume of repeat orders. Walmart's own design hints at where agent shopping is likely to start: routine, repeat purchases where the shopper knows what they want. A local business could offer its regular customers a simple reordering assistant on WhatsApp or the phone, with the same kind of guardrails Walmart is using. The customer confirms each order, or sets a spending cap, and the system checks live stock before placing it.

A third example is a B2B distributor or wholesaler in the USA or India that takes orders from shops and clinics. Buyers increasingly want to place orders without waiting for a sales rep. An internal ordering agent connected to the distributor's inventory and price lists can take a request in plain language, confirm availability and send a draft order for approval. The ideas behind Walmart's terms, such as staying within the buyer's instructions and never touching payment details, are a sensible blueprint for this kind of system.

There are honest limits to keep in mind. Walmart's gateway is a beta, Agent Profiles have not yet rolled out, and nobody can yet say how many shoppers will hand their buying over to an agent. Retailers are also taking different views, as Amazon's approach shows, so the rules may differ from one platform to the next. Agents can misunderstand requests, which is why Walmart has built in approvals, spending limits and a returns process. Any business building its own version needs the same safeguards, plus clear records of what the agent did and why. Data protection rules, such as India's DPDP Act and the UK and EU GDPR, also apply to any customer data an agent handles.

How Legacy Services can help. Most businesses do not need to build their own MCP gateway tomorrow. The sensible first step is to make your store agent-ready. Legacy Services can review your product catalogue, clean up titles, attributes and stock data, and connect your Shopify or custom Next.js store to your inventory system so that what an AI sees is accurate. We use n8n workflows to keep product data, prices and stock in sync across your store, marketplaces and spreadsheets, so there is one reliable source of truth.

From there, we can build customer-facing tools that use the same data. That might be an AI chatbot on your website or WhatsApp that answers product questions and helps customers reorder, or an AI calling agent that takes repeat orders over the phone. We design these with the kind of guardrails Walmart has announced: confirmation before any purchase, optional spending limits, no access to card details, a clear handoff to a human when something is unclear, and a full log of every action. For businesses that want to go further, we can expose parts of your catalogue through an MCP server so that approved AI assistants can query it directly.

Agent-led shopping is still early, but the groundwork you lay now, clean data, connected systems and safe automation, pays off whether customers arrive through a search engine, an AI assistant or your own website. If you want to understand what this shift means for your store and where to start, book a free 30-minute strategy call with Legacy Services at legacyservices.in/contact. We will look at your current setup and suggest practical next steps, with no obligation.