E-Commerce
By Jason Kumpf, Strategy Advisor · October 28, 2026 · 5 min read

AI assistants can now research products, call stores and complete a purchase once the shopper says yes. Here is what agentic commerce looks like today and how online sellers can welcome it.
For two decades, online shopping has followed the same script: search, click, compare, add to cart, check out. That script is being rewritten. AI assistants can now research products, track prices, call stores and, with the shopper's approval, complete the purchase.
The industry calls this agentic commerce, and 2025 and 2026 turned it from a demo into a product. For retailers, it opens a new storefront that lives inside conversations. For shoppers, it means less time on tasks they never enjoyed and more time on the choices they do.
An AI agent is software that can act, not just answer. In commerce, that means an assistant that understands a request in plain language ("a waterproof hiking jacket under three pounds that packs small"), gathers options across many stores, narrows them down and carries out the next step for you.
The key shift is the hand-off. Earlier shopping tools sent you to a list of links. Agentic tools stay with the task until it is done, then ask you to confirm. The shopper stays in charge of the decision while the assistant handles the legwork.
Three capabilities make this possible today:
In November 2025, Google introduced a set of agentic shopping features for the holiday season. Shoppers in AI Mode in Search can describe what they want in natural language and get organized responses with visuals, reviews and inventory information. The same shopping features arrived in the Gemini app for users in the United States.
Two additions stood out. "Let Google Call" lets the AI phone nearby stores to ask about stock and promotions, then report back by text or email. Agentic checkout lets a shopper track an item and have Google buy it on the merchant's site once it meets their criteria, but only after the shopper approves the purchase and shipping details. Launch merchants included Wayfair, Chewy, Quince and select Shopify merchants.
The engine underneath is scale. Google says its Shopping Graph holds more than 50 billion product listings, with 2 billion of them refreshed every hour. That freshness is what lets an assistant answer "is it in stock near me today?" with confidence.
In January 2026, at the National Retail Federation's annual event, Google announced the Universal Commerce Protocol (UCP), an open standard that gives agents and retail systems a common way to work together. Google developed it with Shopify, Etsy, Wayfair, Target and Walmart, and ten other companies endorsed it, including Best Buy, Flipkart, Macy's, The Home Depot and Zalando.
Google paired the protocol with Business Agent, a chat tool in Search that acts as a virtual sales associate able to answer product questions in a brand's own voice. Lowe's, Michaels, Poshmark and Reebok were among the first brands to use it.
Google is not alone. In September 2025, OpenAI launched Instant Checkout in ChatGPT, letting users in the United States buy from US-based Etsy sellers without leaving the chat. Over a million Shopify merchants, including brands such as Glossier, Vuori, Spanx and SKIMS, were named as coming next.
The checkout runs on the Agentic Commerce Protocol, an open standard co-developed by OpenAI and Stripe. Because it is open, businesses can adopt it regardless of which provider they use behind the scenes. The practical result for a merchant: products can be discovered and bought where shoppers are already asking questions.
The card networks have built matching foundations. In April 2025, Visa introduced Visa Intelligent Commerce and Mastercard unveiled Agent Pay, both designed so that verified AI agents can complete purchases for consumers using tokenized credentials and rules the shopper sets. Visa named partners including OpenAI, Microsoft, Anthropic, Stripe and Samsung. As Visa's Jack Forestell put it at the launch, "Soon people will have AI agents browse, select, purchase and manage on their behalf."
Agentic commerce works because it removes friction shoppers already feel. Comparing a dozen tabs, checking whether a size is back in stock and waiting for a price to drop are chores. Handing them to an assistant feels natural, the same way people once handed directions to a map app.
Agents also bring patience that people rarely have. An assistant can watch a product for weeks, check several retailers at once and wait for the right size to return, all without the shopper keeping a single tab open.
The approval step matters just as much. Every major launch so far keeps the shopper in the loop: the agent proposes, the person confirms. That design builds the trust needed for people to try it on bigger purchases over time.
It also suits the way people shop across categories:
The good news for online sellers is that the groundwork looks a lot like good ecommerce practice. Agents rely on clear, structured, current information. The brands that show up well in AI conversations are the ones that make their catalogs easy to understand.
Agentic commerce adds a new front door to your store rather than replacing the ones you have. Shoppers will still browse websites, watch videos and visit stores. A growing share will also ask an assistant to do the searching, and those assistants will reward catalogs that are complete, current and clear.
A practical first step this quarter: pick your top 50 products and review them as an AI agent would. Are the attributes complete? Is stock accurate? Would a plain-language request lead to this item? Small improvements there help today's shoppers and prepare your store for the assistants already arriving.