Why Agentic Shopping Should Enhance Discovery, Not Replace It

The term agentic commerce is often equated with fully autonomous shopping experiences: systems that search, compare, negotiate, and complete purchases on our behalf. That future may prove valuable for highly utilitarian tasks—reordering household essentials, replacing a broken appliance, or finding the cheapest flight. But much of what we consider shopping is not a task to be outsourced; it is an experience people actively enjoy.

This distinction matters because even today, the majority of retail spending still happens in physical stores—environments designed around browsing, discovery, and inspiration as much as efficiency. Digital commerce has historically optimized for convenience and conversion. The next generation of AI-powered shopping presents an opportunity to bring more of the richness of offline discovery into online experiences.

In physical retail environments, discovery and search naturally coexist. A shopper may enter a store with only a vague objective, browse different styles and possibilities, narrow preferences through comparison, ask for specific items, and continue exploring before ultimately making a decision. Exploration and retrieval occur as part of the same journey.

Digital commerce evolved differently. Historically, browse and search emerged as distinct experiences, driven in part by technical limitations. Search systems optimized for explicit intent expressed through queries, while browse experiences relied on categories, merchandising, and recommendation systems to facilitate exploration. Many retail experiences continue to reflect this separation through distinct home feeds, category based browse, and search surfaces.

Over the past decade, online search systems evolved from lexical retrieval toward semantic understanding of expressed intent. Through embeddings and advances in representation learning, we became dramatically better at interpreting user intent, improving precision and recall beyond literal keyword matching. Yet even as interactions expanded beyond text to include images, voice, and other modalities, these advances remained constrained by a fundamental assumption: that users could clearly express what they wanted.

At the same time, personalization and user journey modeling enabled discovery experiences to learn from long-term preferences and behavioral patterns, improving the relevance of the products and content presented to users. Search became better at understanding explicit intent, while discovery systems became better at anticipating preferences. Yet these capabilities largely evolved in parallel, without addressing the fundamentally interconnected nature of shopping, where exploration shapes intent and intent continually redirects exploration.

As in the offline world, many online shopping journeys are not about retrieving a known answer. Users often discover what they want through exploration. Preferences emerge gradually, and taste becomes clearer through exposure, comparison, and elimination. Saving an item, zooming in on a visual style, refining a recommendation, or dismissing an option all reveal evolving preferences. Similarly, the actions users take within search results can initiate further discovery, prompting comparison, exploration of adjacent possibilities, and the refinement of taste.

As systems gain access to richer session context, longer-term preference signals, and conversational memory, shopping experiences can no longer be designed as a linear progression from discovery to search to purchase. Instead, they must support continuous interaction loops in which explicit queries, implicit behaviors, and multimodal inputs jointly inform an evolving understanding of user intent.

The Query Is Not the Mission

Consider a user searching for a coffee table. A traditional commerce system treats this primarily as a retrieval problem: rank the best coffee tables based on relevance, popularity, price, and conversion likelihood. But the query provides only a partial specification of the underlying mission—furnishing a first apartment, redesigning a minimalist living room, or creating a child-friendly family space. The query is simply the starting point for a richer journey.

As the user repeatedly engages with walnut finishes, dismisses glass styles, gravitates toward rounded edges, and refines toward modern organic aesthetics, these interactions become signals of evolving intent. The opportunity is not merely to retrieve better products in response to the next query, but to help users navigate the broader mission itself—deepening an emerging aesthetic, surfacing complementary pieces, and adapting recommendations as preferences and constraints evolve over time.

The Opportunity

LLMs and multimodal AI capabilities make richer interaction loops possible at scale. But the opportunity is not simply to add conversational interfaces to shopping experiences. It is to fundamentally rethink both how commerce systems model intent and how user experiences evolve beyond static browse and search paradigms.

The goal is not to eliminate the journey altogether, but to make it more informed, adaptive, and enjoyable while helping users build confidence in their decisions. The future of AI-powered shopping will not be defined by building better chatbots, but by reimagining how people discover, express, refine, and ultimately realize what they want.