AI agents are supposed to transform commerce, autonomously researching products, comparing prices and eventually making purchases on our behalf, but the reality of how people actually shop suggests a considerably more complicated future.
In this episode of Retail Media Unboxed, we sit down with Justin Sparks, GM of Retail and Commerce Media at Teads, to separate the practical applications of AI in commerce from the hype. Sparks argues that the immediate opportunity isn’t autonomous shopping at all, but AI-assisted research, particularly for expensive or complicated purchases where consumers genuinely benefit from having information synthesized for them.
We explore why the infrastructure for true agent-to-agent commerce still has a long way to go, how LLMs could reshape product discovery before shoppers ever reach a retailer, and why publishers, reviews, Reddit, YouTube and the open web could become even more important as sources of trusted information. We also discuss what this means for retail media networks whose lucrative sponsored-search businesses depend on shoppers actually searching retailer websites, and why the arrival of conversational commerce could force retailers to rethink both monetization and the customer experience.
Most importantly, Sparks challenges the idea that AI will make brands less important by turning shopping into a purely rational exercise. If AI recommendations are ultimately informed by reputation, preference, loyalty, trusted editorial coverage and the information surrounding a brand across the web, brand building may become more consequential in an agentic world, not less.
Five Key Takeaways
1. Agentic research is arriving much faster than agentic purchasing.
The most compelling current use case for AI isn’t handing over the credit card and allowing an agent to shop autonomously, but helping consumers navigate complicated, high-consideration purchases such as televisions, electronics or travel. For habitual purchases and everyday CPG products, autonomous agents may actually introduce more complexity than they remove.
2. The open web could become more important, not less.
LLMs need reliable information to formulate useful recommendations, which means publishers, editorial reviews, consumer reviews, YouTube, Reddit and other sources can influence what products ultimately enter an AI-generated consideration set. Brands therefore need to think about their visibility across the information ecosystem rather than focusing exclusively on what happens at the retailer.
3. GEO could become a new battleground for challenger brands.
Smaller brands have an opportunity to establish themselves inside AI-driven product discovery by ensuring accurate and consistent information appears across publishers, reviews and user-generated content. If AI systems increasingly mediate the beginning of product research, earning visibility in the sources those systems rely upon could become as strategically important as conventional search visibility.
4. AI may make brand building more important rather than less.
An AI can rationally compare price, availability and specifications, but consumers still bring preferences, loyalty and perceptions of quality into the process. If those preferences become part of the context agents use to make recommendations, strong brand equity becomes a signal that algorithms must account for rather than something algorithms eliminate.
5. Retail media’s sponsored-search model could face a fundamental challenge.
Retail media networks have built significant businesses around monetizing search-results pages, but AI-driven discovery could mean shoppers arrive at retailers already knowing considerably more about what they want. If that reduces searches and page views inside retailer environments, RMNs will need new ways to monetize conversational interfaces and capture demand without compromising the customer experience.
Listen & Subscribe
