The Next Era of Retail Won’t Be Browsed, it Will Be Prompted

Your next most valuable customer isn't a person. It's an artificial intelligence agent, and it's already shopping. Whether through personal assistants or retailer-embedded tools, AI is shifting retail from search-driven discovery to intent-driven commerce, where the sale is won before a consumer starts browsing.

Think that's just wishful thinking from people who work in tech? Earlier this year, Walmart announced that shoppers who used Sparky, Walmart’s AI shopping assistant, had basket sizes that were on average 35 percent bigger than non-AI baskets.

The knee-jerk reaction from other retailers will be to rush out their own AI shopping assistant. That's the wrong move. A shiny chatbot is a feature. Features get copied, commoditized, and forgotten. The retailers that win the next decade won't be the ones with the flashiest AI interface. They'll be the ones that re-engineer their foundations for a world where AI agents are the primary shoppers.

That world arrived faster than most expected. Over the past 90 days, Google launched its Universal Commerce Protocol, OpenAI shipped Instant Checkout, and Shopify activated Agentic Storefronts. AI agents can now browse, compare, and complete purchases across brands, meeting consumers wherever they already are. The infrastructure isn't coming. It's live.

So what separates the retailers that capture this shift from the ones that get left behind?

Three foundational decisions, all of which can start today.

First, your product data must be machine-readable. These large language models (LLMs) are scanning hundreds of websites in real time and compiling recommendations for consumers. If they cannot read your product data, how can they recommend your products? If AI agents literally cannot find you, you’re invisible to an entire emerging channel.

Most product catalogs were built for search filters and human browsing, not AI conversations. That means stripping marketing fluff from product descriptions and replacing it with structured attributes, real-world use cases, and contextual detail (who uses this, when, and why) that an AI agent can parse, compare, and confidently recommend.

Along the same line, your checkout process needs to be agent-friendly. As the shift moves more toward fully agentic commerce, retailers need to be prepared to have transactions completed by AI agents. If you’ve done the hard work of cleaning your product data to be machine-friendly and have been recommended by a LLM, but fall short at the checkout process, all that effort will have been for nothing.

Google's Universal Commerce Protocol and OpenAI's Agentic Commerce Protocol are both live and early data shows that retailers supporting both protocols see significantly more agent-driven traffic than those on just one. The retailers that move first on multiprotocol checkout won't just capture a new channel, they'll lock in compounding advantage as agent adoption scales.

Many retailers are already doing this, but they need to double-down on their generative engine optimization (GEO) efforts. Without actively trying to appear on these LLMs, you run the risk of them never finding your brand.

GEO is where search engine optimization was 15 years ago. The early movers will define the category. AI chatbot referral traffic to retail sites grew by over 500 percent last year, and the brands showing up in those recommendations aren't the ones with the biggest ad budgets. They're the ones whose content is structured, cited by third parties, and rich enough that a generative engine can confidently stake a recommendation on it.

None of these decisions are glamorous. None of them will generate a press release. However, in 12 months the retailers that get their product data clean, their checkout agent-ready, and their content optimized for generative discovery will be the ones AI agents recommend first. They'll be the ones building compound growth while competitors are still debating which chatbot to build.

David Blumer is senior director of AI products and agents at Apply Digital, where he leads the company's Lab, builds AI-native products for internal teams and clients, and embeds directly with enterprise partners across retail, CPG and sports to solve real implementation challenges.