A customer looking for a particular game console can ask Google to call nearby stores and check whether it’s in stock. For supported searches in the U.S., Google calls the stores and texts or emails a summary. The store can make the shortlist without the customer ever opening its website.
Now imagine what happens when the customer says, “Buy it if I can pick it up today for less than $800, including tax.” The AI needs to know the exact model, availability at a specific store, the all-in price, and a way to place a pickup order. It needs to know whether it has permission to make the purchase or if it must ask again. The retailer has to accept the order and confirm that the item will be there. Finding an item and getting it into a customer’s hands are different jobs, with different systems responsible for each.
B2A, or business-to-agent, is a catch-all for this emerging process. On one side is a personal agent acting on behalf of the customer. On the other are the websites, apps, and systems sellers use to do business. Between them sit the services that find and compare offers, verify identities and permissions, take payments, and keep track of the orders. Which steps can be connected still depends on the assistant, the seller, and the terms of the integration.
The race to sit on both sides
Google and OpenAI are building ways to attract both shoppers and merchants. Google’s Universal Cart began rolling out in the U.S. across Search and the Gemini app on October 15, 2026. It collects items from different retailers and tracks prices and availability. At participating merchants, shoppers can check out with Google Pay or move the items to the merchant’s own site. In Google’s current native checkout flow for its Universal Commerce Protocol (UCP), an assistant can help build the cart, then the shopper submits the order. The retailer remains the seller of record.
OpenAI and Stripe introduced Agentic Commerce Protocol (ACP) with Instant Checkout in September 2025. In March 2026, OpenAI expanded ACP to product discovery and said it would let merchants use their own checkout while it concentrated on helping shoppers find products. A shopper can find a Shopify product through ChatGPT and check out on the merchant’s site in an in-app browser. In mid-November 2026, OpenAI began rolling out ChatGPT Wallet in dots for eligible U.S. Pro subscribers. The customer approves each purchase, and the payment network supplies a token the agent can use at the merchant’s checkout.
The protocols overlap, and the platforms have different rules for who can participate, where checkout happens, and what the seller can see. An open specification helps systems exchange information. It does not put a specific seller in every assistant’s results or make every cart portable. The more of the buying process a platform handles, the more influence it can have over which offers shoppers see and which path they take to buy.
Walmart is building its own way for customers’ assistants to reach it. Its Agent Gateway is an invitation-only beta. A registered agent can submit a signed record of the customer’s instructions and consent. Walmart may let the agent complete checkout without a separate confirmation if the purchase fits the customer’s limits and the program’s rules. Outside the gateway, Walmart’s terms require the customer to confirm each checkout. That policy is Walmart’s to set, even when the customer sends someone else’s agent.
What are your intentions
Sellers have to distinguish an authorized purchase from a visit. Cloudflare’s Web Bot Auth can verify who sent a signed web request. That signature by itself says nothing about what the customer agreed to buy. And an order record tells the seller what sold, but generally doesn’t show what the customer’s assistant considered elsewhere. OpenAI’s Business Hub, for example, reports aggregate mentions and website clicks and says it doesn’t share individual conversations with businesses.
Make the website agent-ready now
You don’t have to pick a winning protocol to get started. Pick a common, well-measured conversion. Give an assistant the assignment and see how far it gets on the website you already have. Can it match the exact product to the right location? Can it tell what’s in stock and what the customer will actually pay? Can it find the return or cancellation terms before the customer commits? Then follow the transaction through confirmation, and try changing or canceling it.
This exercise will uncover work worth doing regardless of which platform a customer uses. Put prices, variants, availability, and terms where they can be read and kept current. Make the checkout or booking steps predictable, and make sure a failed order looks like a failure to the assistant. Check where your site blocks legitimate, identifiable agent traffic while keeping fraud and access controls in place. Use a feed or a direct integration where it improves a specific step. OpenAI’s merchant guidance, for example, describes both a browser route through the existing website and a direct integration for structured actions.
The store that answers Google’s call has already cleared the first hurdle. The next test is whether the assistant can get a reliable pickup promise, a final price, and a confirmed order without making the customer start over. Send an agent to your own site this week. Wherever it gets stuck because the information is missing or the next step doesn’t work, you have a place to start.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.
About Shelly Palmer
Shelly Palmer is the Professor of Advanced Media in Residence at Syracuse University’s S.I. Newhouse School of Public Communications and CEO of The Palmer Group, a consulting practice that helps Fortune 500 companies with AI strategy, implementation and governance, as well as technology, media and marketing. Named one of LinkedIn’s Top Voices in Technology, he is a bestselling author, covers tech and business for Fox 5’s Good Day New York, is a regular commentator on CNN, and writes the popular daily business blog Think About This. Follow @shellypalmer or visit shellypalmer.com.