Agentic commerce is loosening the grocery aisle’s grip on the shopper

Grocery shopping has become a strange paradox of modern retail, where consumers can order dinner from their phones, receive groceries at their doors, and have previous baskets saved for instant repurchase. Yet the basic chore of deciding what to buy appears to remain stubbornly intact. The convenience revolution may have changed the logistics of grocery shopping; it has not removed the burden of shopping itself.

McKinsey’s latest North American grocery research puts the demand for convenience into sharper focus, finding that 70% of consumers prefer home delivery for online grocery fulfillment. Among those shoppers, 67% point to time savings as a reason. The appeal is obvious. Grocery shopping takes time, and consumers have learned to remove as much friction from the physical trip as possible.

AI is now opening a different question for an industry that has spent decades refining search, merchandising, and recommendation systems: What happens if the technology starts with the person rather than the product?

McKinsey’s 2026 grocery research identifies AI as a force capable of reshaping discovery and decision-making, while grocery retailers expect fully personalized promotions to rise sharply over the next several years. The technological shift holds the potential to make browsing itself less necessary.

Dmytro Lylyk and Vladyslav Mehera, co-founders of Neomi, a grocery shopping AI assistant, have built their argument around precisely that premise. They believe grocery commerce has optimized for making products visible when it should be learning to understand the circumstances behind a purchase. “Our philosophy is around foods for health, not for shelves,” Lylyk explains. “Because currently, retailers and brands optimize their strategies around packaging, placing themselves on shelves to be noticed, to be picked.”

He argues that shoppers typically choose between searching through catalogs and repeating previous orders, with both behaviors reinforcing the same product-led system.

Their answer is an approach they call “foods for health, not for shelves.” Lylyk and Mehera believe retailers have historically invested heavily in making products visible, from their packaging and placement to their position within digital storefronts.

Their view is that the person making the purchase should become the starting point. Lylyk explains, “What we do differently is that we understand the needs, both emotional and physiological needs of a human, including information around health, dietary restrictions, preferences and personal goals as signals that can determine what belongs in a basket.”

Today, shoppers develop increasingly specific expectations around food, as research suggests. Almost half of consumers look for particular functional benefits such as high protein or low sugar, while nearly 55% say retailers best supporting wellness provide personalized nutrition recommendations. Their model responds to this shift by treating the grocery basket as an expression of intent rather than a collection of isolated searches.

AI can make that possible because it can interpret context. A shopper thinking about a romantic dinner, for instance, may know they need a meal without knowing every product required to create it. Lylyk recalls a Neomi user who selected a “Romantic Dinner for Two” experience and unexpectedly found wine included in the basket. The product had not been specifically requested; its relevance came from the context. “We can provide them with a broader, more contextual search which adds products which really help,” he says.

The same logic changes the role of grocery advertising. Lylyk believes promoted products should enter the shopping journey only when they genuinely suit the shopper’s stated intention. Retail media has often relied heavily on visibility, yet an AI system capable of influencing a basket can introduce a more consequential standard, where relevance has to come before promotion. “AI could push products into the cart just because they are promoted, but they wouldn’t fit the cart, or they wouldn’t fit the expectation,” he says. “That’s the borderline which must not be crossed.”

Mehera brings a technical perspective shaped by work in neuroscience and machine learning, alongside previous experience in grocery technology. According to the co-founders, the partnership has a practical balance between Lylyk’s future-facing approach and Mehera’s more grounded view of what can be built.

Their combined perspective has also exposed a less obvious challenge: shoppers themselves have to learn how to communicate with AI.

Based on the user behavior within 120 department store chains, Mehera found that users initially approached the technology like a conventional search engine, asking for milk before moving on to another individual product. Neomi, he notes, began teaching shoppers to communicate their comprehensive needs instead. In his view, a description of what someone eats during a week, a preferred cuisine, or a particular dietary goal can give AI enough context to construct a more coherent basket. Mehera views this as a learned behavior that will take time to change.

The impact can also reach across the grocery ecosystem. Both Lylyk and Mehera envision shoppers eventually communicating their food needs through retailer websites, AI assistants, recipe platforms, and diet applications, with technology translating those intentions into relevant products. Such a model could give retailers a different relationship with the shopper and give brands a different question to answer: Where does a product fit within an actual human need?

Lylyk and Mehera believe retailers should begin testing intent-based shopping against their own shoppers and inventory now, while the model is still taking shape. Grocery commerce, Lylyk notes, has spent decades helping people find products, but the next phase may depend on technology learning to understand the person searching for them.