The Unfinished Promise of AI-Powered Shopping

We’ve all been there: You pull up a retail website to hunt for one specific item—say, a new pair of running shoes—and what starts as a single quick search quickly spirals into overwhelm. Before you know it, you’re sifting through hundreds of irrelevant results: styles you’d never wear, children’s sneakers even though you’re an adult shopping for yourself, and dozens of options that land way outside your budget. When you’re drowning in low-quality junk, more choices don’t feel helpful—they feel like a waste of time.

Unlike traditional search tools that just dump every possible result in your lap, AI has the potential to deliver curated, guided shopping experiences that feel much like working with a knowledgeable, helpful in-store associate. But for the most part, that potential hasn’t been realized yet.

AI Is Already a Daily Default

Even as AI lags in shopping, consumer expectations are climbing fast. For a growing share of people, AI is already the go-to interface for everyday tasks. We use generative AI daily to answer random questions, plan trips, troubleshoot tech problems, and make both small and big decisions.

Data from Constructor and Shopify shows nearly two-thirds of people now use tools like ChatGPT in their daily lives, a huge jump from just 29% in 2023. That number climbs even higher for Gen Z, with 78% of the cohort reporting regular generative AI use.

It’s only natural that this comfort and habit with AI carries over into shopping. Today, consumers aren’t asking whether AI belongs in the shopping experience—they’re asking why it doesn’t work better than it currently does.

We’re Still in the Early Innings

The reality is, we’re only just getting started with AI in retail. When it comes to using AI to help people find products, the core challenge is not a language problem—it’s a decision problem.

Today’s AI systems can already understand and respond to complex, natural-language queries that would have been unthinkable in a traditional search bar just a few years ago. Ask “I’m planning a tailgate—what do I need?” or “Help me find new running shoes for long runs” and you’ll get coherent, sensible recommendations in response. That’s a massive leap forward.

The larger, more pressing problem is whether those recommendations actually make sense for you, the individual shopper. That’s where the decision gap lies: figuring out what to show each specific shopper is incredibly difficult. It requires detailed detective work, since purchase decisions are almost always rooted in a person’s past actions, unique preferences, browsing behavior, and more. While modern large language models excel at generating confident, polished answers, they often struggle to connect those answers to real-world context and outcomes—like, which pair of running shoes is this specific shopper actually most likely to buy?

Why the Context Gap Exists

To truly help shoppers, AI first needs to understand what makes each person tick. But general-purpose AI tools like ChatGPT and Claude have no access to the critical clues that paint this picture: what you’ve bought before, what you added to your cart but didn’t purchase, what you’ve returned in the past, and more.

This personal data is fragmented across individual retailer systems and is almost always proprietary. But it’s essential to building a full picture of a shopper’s needs—and without that full picture, AI can’t narrow down options to fit what you specifically want.

Take the running shoe example again: A serious competitive runner will care most about stability, toe box width, and whether the shoe is built for trails or pavement. They may favor a specific brand or have loved the last version of a model so much they want an updated pair. A casual runner, meanwhile, may just want something comfortable for occasional weekend jogs. A generic “what are good running shoes” prompt can’t connect these dots on its own. And if a shopper has to explain every single preference and use case manually, AI isn’t actually simplifying their experience at all. To do its job well, AI needs the right data and context at the right moment to guide a shopper’s choice.

Early Traction for Context-First AI

A context-driven approach to AI shopping is already showing promising results. Some leading retailers have built their own in-house AI agents that combine full product and inventory data with individual shopper information, including real-time on-site behavior, past purchases, and loyalty status. When a shopper asks for guidance, this AI can move far beyond generic recommendations to surface items that person is actually likely to want.

Not all shoppers are ready to use these tools yet, and adoption is still early. But even among the relatively small group of consumers using these AI tools right now, the impact is already meaningful:

• Amazon reports that shoppers who use its AI shopping assistant are more than 60% more likely to complete a purchase during their browsing session. Usage is also growing rapidly, with engagement up nearly 400% year over year.

• Walmart has seen similar positive trends: Customers who use its Sparky AI assistant have an average order value 35% higher than shoppers who don’t use the tool.

• Our own data shows that during last year’s Black Friday to Cyber Monday shopping period, more than 10% of total revenue on sites with AI shopping agents came from shoppers who used the tools.

That said, most retailers haven’t mastered the context piece yet. I recently tested the AI agent on a major national department store’s website: I’d added four pairs of shoes to my cart during a previous visit, and when I returned the next day I asked the AI to recommend similar styles to what I’d been browsing. Its response? “To help me narrow this down, were you looking for men’s or women’s shoes?”

It’s important to remember this space is still new, and retailers are still experimenting heavily. Brands are still working to figure out where conversational AI agents add the most value when paired with the right context. So far, high-intent use cases like retail search bars and customer chat have delivered the best results. Key decision moments, like when a shopper is already on a product page, are another natural fit—at that stage, shoppers often just need answers to specific lingering questions like “Do these run true to size?” or “Will these work for wide feet?”

What’s Next for AI in Shopping

As retailers get better at integrating personal context into AI tools, we can expect AI to become a far more useful and widespread shopping companion. We’ll also likely see more adaptive, user-centric interfaces that don’t just infer your preferences—they ask targeted clarifying questions as they learn and adapt to what you want.

We’ll also see a shift from AI that only answers questions to AI that acts for you, with agents guiding your choices more directly and helping you complete every step of the purchase process.

Ultimately, the future of AI in shopping will be defined by how well these tools understand individual context and help people move forward with confidence. Once that gap is closed, AI won’t just give you a list of options—it will leave you walking away sure you bought the right product for your needs.