Insights from our retail roundtable, part 1
Retail has adapted to intermediaries before. Search engines reshaped discovery, marketplaces redefined assortment, and social platforms collapsed inspiration and transaction into a scroll.
This year, a different shift is emerging. Something quieter, more structural, and potentially even more disruptive than anything before. AI agents are beginning to act as proxies, not just guides. They search, compare, summarize, and even transact. The interface between consumer and retailer is no longer guaranteed to be a website, an app, or even a marketplace. It may be a conversational system operating on the customer's behalf.
At our recent private roundtable, leaders across categories – from grocery to apparel to home and specialty retail – gathered to explore what this means in practice. Not as a thought experiment, but more as a business question. In this article series, we share the pragmatic learnings from the gathering.
The moment AI stops assisting and starts acting
The ongoing revolution was summarized well by one executive at our dinner: "Soon, AI won't just help the customer. It will become the customer.” This shift will cause ripple effects across the entire retail model.
When AI agents search for products, compare prices, evaluate reviews, select items, and complete checkout, traditional assumptions about traffic, conversion, and loyalty begin to loosen.
Retailers have spent decades optimizing for human attention:
- SEO strategies
- performance marketing
- retail media monetization
- UX refinement
- personalization engines
But when the browsing layer shifts from human interface to algorithmic interface, the economics move with it. One participant put it directly: "Retail drives revenue. Advertising drives profit."
If checkout increasingly happens within AI-driven environments, retailers risk losing high-margin advertising income tied to owned channels, not just traffic visibility. The bigger worry is margins eroding quietly over time, well before retailers lose their place in the value chain. When AI agents abstract comparison and compress choice into ranked outputs, everything starts to look the same.
Retailers now face two tasks: structure product data, content, and reviews so AI systems can interpret and prioritize them, and protect direct customer relationships and first-party data wherever possible. It feels like a new visibility war – less about ranking on search engines, more about being legible to large language models.
Starting small in a big shift
Despite the scale of the question, the most actionable insights came from narrow experiments. Before AI becomes the only customer, a lot can be done with agentic systems for your customers. One retailer began with a simple, universal problem: helping customers answer the daily question, "What should I do tonight?"
Rather than launching a sweeping AI transformation, they built a conversational interface focused purely on providing practical suggestions. The assistant connected to an existing database of ideas and helped translate vague intent into actionable options. The scope was deliberately constrained: one clear use case, one specific customer question, one domain of expertise.
From there, the tool expanded gradually into a broader search. Behind the conversational layer, traditional search engines still power results. The AI was an interface enhancement, not a backend overhaul.
How they rolled it out proved just as important as the technology itself. Instead of promoting the assistant as a headline feature, the team embedded it contextually: when customers failed to find results, when scrolling behavior suggested hesitation, when search intent seemed unclear.
The assistant surfaced as a support mechanism rather than a replacement journey. Adoption didn't explode overnight, but it grew steadily, particularly in moments of friction. "You learn more from shipping than from thinking," one leader reflected.
Across the table, a recurring theme emerged: different customer intent requires different journeys. Some users want to be fast and efficient; others want to explore, be inspired, and enjoy the discovery. Retailers described designing for both modes, sometimes deliberately leaving "good friction" – small, meaningful pauses that allow exploration, inspiration, and emotional connection, rather than eliminating every moment of delay.
The broader lesson was repeated: move quickly, but stay focused. Conversational interfaces are lighter than full-scale replatforming efforts. They can be deployed, tested, and refined without restructuring the entire organization.
