Shoppers are asking AI what to buy - will It recommend your products?

Recently, I turned to an AI shopping assistant to help me find a backpack for my son.

The request was straightforward: something durable, comfortable, and appropriate for everyday use – not a hiking pack or fashion statement. The assistant surfaced several options including images and purchase links, and I checked out directly through the agent.

When the backpack arrived, it was toddler-sized. This was not a subtle miss. My son is six-foot-six.

The issue was not just an AI hallucinated product. It pulled from real listings with unclear sizing context, inconsistent descriptions, and incomplete metadata scattered across the web. The assistant had no reliable way to reconcile those gaps or resolve the issue.  

Despite that experience, I continued using AI to shop. That is the paradox retailers need to understand. AI shopping is imperfect, but adoption is accelerating anyway.

AI is already reshaping how shoppers discover products

Retail leaders often ask whether AI shopping assistants will replace ecommerce sites or marketplaces. That is the wrong question. The real shift is happening before a shopper ever lands on a product page.

During peak shopping periods, generative AI driven traffic to retail sites has surged year over year, driven largely by top of funnel behaviors like product research, comparison, and recommendation queries.

That distinction matters. Discovery has always been one of retail’s most competitive battlegrounds. What is changing now is who controls it.

Why AI excels at the moments that matter most

A clear signal of AI’s role in shopping came when ChatGPT announced early retail partnerships with platforms like Etsy and Shopify. That pairing was not accidental.

Large language models excel at handling long tail, highly specific queries that traditional keyword based systems struggle with. Searching specifically for a "durable everyday backpack for a very tall adult" isn’t just an edge case, it reflects how people actually shop when they are overwhelmed by choice.

AI also performs particularly well in higher consideration categories such as laptops, appliances, and car seats. In these categories, shoppers want help comparing features, synthesizing reviews, and weighing tradeoffs. Instead of bouncing between tabs, consumers can ask follow up questions, refine criteria, and receive distilled recommendations in a single conversation.

For shoppers, this improves their experience. For retailers and brands, it raises the stakes – because when AI becomes the interface, data becomes the differentiator.

The hidden risk is invisibility, not inaccuracy

Most conversations about AI in retail focus on hallucinations, errors, or misinformation. Those are real concerns. But there is a quieter and potentially more damaging risk emerging for brands – invisibility.

When an AI shopping agent generates recommendations, it can only work with the information it can reliably access. If your product data is inconsistent across channels, missing key attributes like size, fit, or intended user, or buried behind outdated listings, the model will not guess. It will simply move on.

This is especially true for details that matter late in the decision process but early in AI evaluation. Shoppers increasingly expect AI to answer these questions instantly. If it cannot, confidence erodes.

The uncomfortable reality is that many retailers still treat this information as secondary. In an AI driven discovery environment, it is table stakes.

What it takes to show up and stay relevant in AI discovery

Retailers do not need to predict exactly how AI commerce will evolve to start preparing for it. They do, however, need to get serious about the fundamentals that AI systems depend on today and will rely on even more heavily tomorrow.

1. Centralize and standardize product data

When AI pulls from the open web, inconsistencies compound quickly. A centralized product feed provides a single source of truth across channels.

This is not just an efficiency play. It is a visibility strategy. Brands that maintain accurate and real time data at scale are far better positioned to be surfaced and trusted by AI systems.

2. Treat policies and specifications as first-class data

Returns, shipping, warranties, delivery promises, and critical sizing and fit are no longer details to bury below the fold. They are often the deciding factor in AI mediated recommendations.

If these attributes are not embedded directly in product pages and structured in a way machines can understand, AI assistants cannot answer the most common shopper questions with confidence. That is when recommendations break down or disappear altogether.

3. Optimize for long tail and contextual discovery

AI shines when shoppers are unsure what they need, creating a powerful opportunity for brands that enrich their metadata with use cases and context.

This kind of contextual optimization helps AI connect products to real world needs rather than just specs and dramatically increases the odds of appearing in high intent discovery moments.

4. Monitor how AI represents your brand

Forward looking teams are already auditing how their products appear in ChatGPT, Perplexity, and other AI shopping environments. Are key attributes like size and fit accurate. Are flagship products being mentioned? What questions does the AI struggle to answer?

These insights will inform not just optimization efforts but future attribution models as AI and commerce become more tightly integrated.

5. Readiness today determines relevance tomorrow

AI shopping assistants are not a passing experiment. They represent a structural shift in how consumers discover and evaluate products, and that shift is unfolding faster than many retail organizations are prepared for.

The brands that succeed will not wait for the technology to mature. They will strengthen their data foundations now, optimize for AI driven discovery, and adapt as new interfaces emerge.

That toddler sized backpack was not just a bad recommendation, it was a clear signal. Shoppers are already asking AI what to buy. The only remaining question is whether your products are ready to be part of those answers.

Suzin Wold is chief marketing officer of Rithum.