Personal recommendations from friends and store staff are losing ground as generative AI models take over product selection. Four competing AI-commerce architectures are emerging, but financial and trust risks slow adoption.
Retailers and digital platforms are in the middle of a shift: generative AI agents are starting to take over the work once done by human advisors, friends, influencers, and in-store staff. The change isn't sudden, but the direction is clear. Within five years, AI-driven systems are expected to reshape how people choose and buy products.
On September 2, 2026, Anthropic introduced ready-made blueprints for two types of AI agents in retail: shopper agents for customer scenarios and merchant agents for internal seller tasks, clarifying that these agents can recommend products and add items to the cart but do not complete purchases on behalf of clients.
Four main approaches are now competing to shape AI-commerce. The first is ecosystem AI-commerce, where assistants built into major tech platforms help users make buying decisions. The second is retail AI-commerce, with proprietary assistants inside retailer platforms. GenAI-commerce uses open, universal agents that work across brands and stores. Marketplace AI-commerce brings agent interfaces directly into large e-commerce marketplaces.
In January 2026, Google and Shopify announced the Universal Commerce Protocol, which by September 2026 was described in industry materials as an open standard for agent-driven commerce, connecting discovery, cart building, checkout, and post-purchase support.
For publishers and creators, the impact is immediate. As generative models take over the comparison and recommendation layer, traditional strategies built around product reviews, affiliate links, and expert guides are losing ground. The fight for visibility is moving from SEO and social reach to integration with AI agents that shape the user's path to purchase. This matches recent analysis of how AI-driven interfaces are already changing traffic and conversion patterns.
Retailers who don't adapt risk being left out of the new agent-driven commerce stack. Those who invest early in AI integration-through proprietary assistants or partnerships with open agent platforms-could capture a larger share of future transactions. The next five years are likely to bring more than small changes to e-commerce UX; they will shift trust and decision-making from humans to machines.
This is not a theoretical debate but a practical race for control over the last mile of digital commerce. The companies that succeed will be those who see that relying on human persuasion and static content is no longer enough. Instead, the future will belong to those who can build and align AI agents that act as real proxies for user intent, while managing the risks of bias, liability, and technical failure. The window for securing a place in the agent-driven economy is closing quickly.
According to a Reuters industry report, Anthropic's agent templates are designed for retail, travel, and ticketing, aiming to meet the demand for conversational search and to prepare for the holiday shopping season.
Meanwhile, regulatory approaches differ by region. At the G20 meeting in North Carolina on September 1, 2026, U.S. officials supported a "hands-off" approach to AI regulation, arguing against new rules for commercial AI agents, as reported by Reuters. In contrast, India is preparing a framework for agentic payments within its UPI system, which could let AI agents make small digital payments without separate confirmation for each transaction, with spending limits, audit trails, and identity checks built in.
