Four months ago, industry veterans were debating whether AI shopping agents could disrupt Amazon and Walmart’s dominance. Today, those same retailers are racing to build the infrastructure that makes those agents possible. That’s how quickly this is moving.

For the past 25 years, the retail website has been sacred territory. Brands controlled the narrative, captured data, and converted browsers into buyers. AI shopping agents are about to do to websites what e-commerce did to storefronts: not eliminate them, but fundamentally transform their purpose.

Analyzing over 20 million AI shopping agent conversations revealed an unmistakable pattern. Consumers aren’t just experimenting with conversational commerce; they prefer it for the purchases that matter most. In electronics alone, buyers ask 50% more questions before purchasing than in any other category. When retailers deploy AI shopping agents for these complex products, they’re seeing 25% engagement rates and conversion lifts 10 times higher than traditional website experiences.

High-consideration purchases—the ones that drive the most margin and require the most customer service resources—are already moving to conversational interfaces. Your laptop buyers, your furniture shoppers, your customers researching the perfect gift aren’t browsing product grids anymore. They’re asking questions, comparing features in natural language, and expecting instant, intelligent guidance.

I recently watched this shift play out in real time. At a summit of 50+ retail executives from brands like Wayfair, Lenovo, and Foot Locker, we held a live debate: Will AI agents replace websites within 10 years? Initially, 82% of the room defended the website as irreplaceable. After hearing the evidence on engagement data, conversion metrics, and the changing consumer behavior, sentiment shifted enough that the pro-agent argument won. These retail veterans recognized that AI is fundamentally separating two functions their websites have always combined: discovery and transaction.

Your website will still exist a decade from now. But it won’t be the only transaction engine. It will be your brand showcase, your content hub, your trust-building destination. Meanwhile, the actual shopping, including the research, comparison, and purchase consumers make, will increasingly happen in AI agent environments, whether that’s ChatGPT, Claude, Gemini, or brand-owned conversational experiences.

Google’s Gemini, Anthropic’s Claude, and OpenAI’s ChatGPT are all rolling out or experimenting with features that enable AI agents to handle ecommerce transactions without users ever leaving the platform. When a consumer can research products, compare options, read synthesized reviews, and complete purchases all within a single conversational thread, navigating to multiple websites, opening countless tabs, and hunting through product filters becomes unnecessary friction.

Forward-thinking retailers who understand this shift are already building what I call “agentic infrastructure.” These are the systems, data architecture, and conversational capabilities that let AI agents access their catalog, understand their brand voice, and transact on behalf of customers. They’re treating AI shopping agents the way forward-thinking retailers treated ecommerce 25 years ago.

That comparison matters. In 2000, when Ralph Lauren invested $200 million in ecommerce, a headline in the (print) Daily Mail declared the internet was “just a passing fad.” The move seemed risky. A decade later, that bet continued to pay off as the brand was “recession proof”  and its digital dominance continues through today. The current AI moment demands similar courage, and the stakes are just as high.

There are no experts in AI commerce yet. We’re all builders in these early days, testing hypotheses and learning what works. But some retailers are testing, iterating, and gathering insights from every conversation their AI agents have, while others are waiting for certainty that will never come. The gap between those two groups will determine who owns the next decade of retail.

The data coming from early AI shopping agent deployments tells a clear story. When purchases require careful consideration, consumers increasingly prefer conversational commerce over self-service browsing. They want to ask follow-up questions and have comparisons explained in plain language. They want the kind of guided shopping experience that mimics a knowledgeable store associate, not a product database. And they want it on demand, not just during business hours. Your website can’t deliver that experience. An AI agent can.

Consumer behavior and the technology giants investing billions in AI commerce have already written the script. Twenty-five years ago, retailers who dismissed ecommerce as a fad learned an expensive lesson. Today, the retailers who dismiss conversational commerce as hype will learn the same one. The difference: the window to experiment, learn, and build is shorter. AI is moving faster than the web ever did.

Your website isn’t disappearing. But if you think the version of it that has worked for the past 10 years will work in the next five years, you’re betting against how your customers want to shop.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.