The Checkout Moves Into the Chat
Walmart has done something that would have sounded far-fetched two years ago: it has handed a slice of its checkout experience to someone else's AI assistant. Through a new integration with Google, shoppers can now search for products, assemble a cart, and complete a purchase entirely inside the Gemini interface, paying with Google Pay without ever landing on walmart.com. The rollout begins in the United States, with international expansion planned, and it represents one of the first times a foundation-model agent controls the purchase path at America's second-largest retailer at meaningful scale.
The mechanism underneath is Google's Universal Commerce Protocol, an open standard designed to let AI agents recommend products, build carts, and finalize transactions across many merchants. Walmart joined a roster of more than two dozen partners including Shopify, Target, and Etsy. The protocol is the connective tissue: it lets an agent speak a common commercial language to any participating retailer, so the shopper's intent expressed in natural language can be translated into a real, paid order without a bespoke integration for every store.
Open Partnerships as a Deliberate Doctrine
Walmart is not being coy about the philosophy behind this. Hari Vasudev, the company's executive vice president and US chief technology officer, put it plainly: "That approach of open partnerships is something that we strongly believe in." He reached for a familiar analogy to explain why the retailer is willing to build for platforms it does not own, comparing the strategy to mobile development. "Even today, we have iOS and Android platforms, and we build our resources for both those platforms," he said.
That framing is more strategically loaded than it first appears. Vasudev is arguing that agentic commerce will resemble the smartphone era, a world of a few dominant surfaces where the retailer's job is to be present and excellent on each, rather than to force customers onto a single owned destination. Walmart is already working with both OpenAI and Google on AI commerce, which signals a portfolio approach: be wherever the shopper's agent happens to live, and let the customer, not the retailer, choose the front door.
Amazon Takes the Opposite Road
The contrast with Amazon is the most interesting subplot. Amazon was conspicuously absent from the UCP partner list, and its strategy has been to build proprietary agentic capabilities such as its Rufus assistant, along with tools branded Buy For Me and Auto Buy, kept inside its own ecosystem. Where Walmart is optimizing to appear inside other companies' agents, Amazon is optimizing to keep the agent, the catalog, and the payment rails under one roof. Two of the largest retailers in the world have looked at the same future and chosen diametrically opposed architectures.
We think both bets are rational given each company's position. Amazon already owns the demand surface, the logistics, and the customer relationship, so a closed stack protects its most valuable asset. Walmart, competing against that gravity, benefits from an open protocol that lets it ride third-party agents it could never build the audience for on its own. The open approach trades some control for reach, and for the challenger, reach is the scarcer resource.
What Agentic Checkout Changes for Retail Tech
For retail technology leaders, this shift rewires several assumptions at once. Discoverability stops being about search engine optimization and paid placement on a website and starts being about whether an agent can cleanly parse, trust, and act on a product catalog through a protocol. Payment orchestration becomes a negotiation between the retailer's systems and the agent's preferred rails, in this case Google Pay. And the analytics that retailers have relied on, built around sessions and page views, degrade when the customer never visits the site at all.
There is also a harder question lurking beneath the convenience: who owns the customer when an agent stands between the shopper and the store? If Gemini becomes the place where purchase decisions happen, the assistant, not the retailer, accumulates the behavioral signal and the moment of trust. Walmart is betting that being present and frictionless on that surface beats being absent from it, but the long-term power balance between merchants and the agents that broker their sales remains genuinely unsettled.
The Bet Beneath the Bet
Underlying Walmart's move is a wager on standards. Open protocols only pay off if they achieve critical mass, and UCP's value scales with every retailer and every agent that adopts it. By joining early alongside Shopify, Target, and Etsy, Walmart is helping to manufacture the network effect it needs, lending its enormous catalog to make the protocol worth building against. It is the same playbook that made earlier web and mobile standards durable: get the big participants in the tent, and the ecosystem follows.
The risk is that agentic commerce fragments into competing, incompatible protocols, forcing retailers to support several agent standards the way they once maintained separate mobile apps. Walmart's hedge, working with multiple AI providers while backing an open standard, is a reasonable response to that uncertainty. For now, the company has planted a flag: in its view, the future of shopping is agent-mediated, protocol-driven, and emphatically not confined to any single walled garden.
