Copyright adweek
Generative AI chatbots, like ChatGPT, are reshaping shopping, but brands and retailers aren’t fully ready to embrace this new storefront. As generative AI chatbots evolve from search engines into full-fledged shopping platforms, retailers risk giving away control over how products are displayed as well as transaction data. “The role of a website, app or a digital storefront has the potential for really big shifts in that those experiences become nested within these agentic experiences,” said Katie Kelly, EVP of commerce experience at Horizon Media. Every major AI company has built its own protocol for feeding product information into large language models. OpenAI and Anthropic use the Agentic Commerce Protocol (ACP), while Google and Amazon rely on separate, incompatible systems. For brands, that means it is difficult to translate product catalogs into multiple formats. Plus, it’s not clear if AI chatbots will surface a brand’s product, Kelly added. This leads to a fragmented space that makes it hard for brands to plug into multiple AI tools. “There is no universal layer, there’s no common language,” Kelly said. Kelly said that some of Horizon Media’s clients are not rushing to embrace AI storefronts. Feeding detailed brand information such as SKU-level data, inventory, pricing logic and promotions—into large models gives these AI platforms significant control over how products appear to consumers. “Once I start sending all of my granular information to the AI, plus all that consumer transactional information—that starts to sound like a riskier proposition for merchants,” she said. Retailers also worry that if brands directly integrate with AI agents, it could sideline deep relationships that retailers typically have with brands, potentially creating a model favoring brands instead of retailers, according to Kelly. Retailers and e-commerce players are dipping their toes into agentic commerce, but few are diving in. Amazon, Lowe’s, and others have launched AI-powered shopping assistants in recent years, while Walmart, Shopify, and Etsy have struck deals with OpenAI to support checkout within ChatGPT. Still, many retailers are prioritizing their core e-commerce experience this holiday season runs, outweighing near-term AI experimentation, two agency sources previously told ADWEEK. Much will depend on how the AI platforms themselves develop, and whether they wade into the business of powering ecommerce marketplaces or stick to a Google Shopping-style biddable ads layer, noted Nich Weinheimer, chief strategy officer at Skai. “GPT platforms face a strategic fork,” Weinheimer said. If the platforms begin aggregating order integrations, through ACP or otherwise, that would “[introduce] complexity in customer ownership and fulfillment—capabilities Amazon and Walmart have spent decades perfecting,” Weinheimer continued. “Given these challenges, GPTs are more likely to prioritize the ads layer, though ACP integration suggests marketplaces remain a possibility.” That anxiety isn’t new. When Amazon’s marketplace rose, sellers handed over detailed product data that Amazon later used to launch competing private-label brands, Kelly noted. helped Amazon launch competing private-label brands, Kelly noted. Marketers have similar concerns that giving AI tools their data will result in AI tools building walled gardens of commerce. “There’s a lot of scary things that can happen with AI, and I think merchants are sensitive to that,” Kelly said. Specifically, data-sharing is a concern. To enable one-click shopping inside AI agents, merchants must provide AI tools granular product information and transaction data beyond what currently lives on their websites. That risk, Horizon Media’s Kelly said, has pushed some of the agency’s clients to stay in exploration mode rather than fully deploying AI shopping experiences. “For a lot of our clients, it’s unproven, and I think there needs to be some additional protections in place before they’re going to start feeling comfortable jumping into truly agentic commerce,” she said. Some tech companies are working to ease those concerns. Commerce platform Commercetools recently launched AI Hub, which streams product data directly from its brands into large language models. The feature gives brands control over how product information appears in AI results and removes the need to rely on web scraping, said Jen Jones, CMO at Commercetools. Despite brand interest, adoption remains uneven, Jones said. Not all of Commercetools’ partners are enabling shopping through ChatGPT—a sign that the market is still cautious. The AI shopping landscape is moving fast. Even as adoption remains uneven, protocol standards are emerging almost daily. “It’s a matter of figuring out which ones to adopt and quickly, because there’s absolutely a first-mover advantage here,” Jones added.
