AI agents have already started buying on behalf of customers. Yet most merchants still lack the infrastructure to serve them. That disconnect sits at the center of PayPal’s first U.S. Agentic Commerce Pulse Survey, based on responses from 498 decision-makers across small businesses, mid-market firms, and large enterprises. Nearly 95% of merchants report that they can already track or observe traffic originating from AI agents, including web crawling from systems like ChatGPT and Google Gemini. But only about one in five have structured their product catalogs in machine-readable formats that those same agents can actually interpret and act on in real time. Many also lack the foundational infrastructure required to participate fully in agentic commerce, including interoperable APIs and agent-compatible checkout systems. The result is a market where demand is evolving faster than the systems built to support it. “We’re seeing the interface layer move first,” Mike Edmonds, PayPal’s VP of agentic commerce and commercial growth, tells Fast Company , noting that where brands once competed for visibility through Google Ads and SEO, agentic commerce is shifting that pressure toward structured product catalogs and how goods appear across LLMs and digital marketplaces. As search becomes more “intent-driven rather than keyword-driven,” Edmonds says, consumers are moving away from basic queries like “running shoes” and instead asking AI systems for highly specific recommendations tailored to their needs. According to the study, that shift is happening faster than most merchants anticipated. Across segments, 86% to 94% of businesses expect agentic commerce to have a positive impact over the next 12 to 24 months. Many report that it already has. “LLMs don’t inherently privilege the largest catalog; they privilege the most structured, most trustworthy data signal,” says PayPal CTO Srini Venkatesan, emphasizing that smaller merchants with strong machine-readable data and credible signals can, in theory, compete alongside much larger players in agent-driven discovery. But while the technology could level the playing field, Venkatesan notes that many small businesses still lack the resources or operational flexibility to navigate complex integrations, making broader ecosystem support essential. Welcome to the Invisible Storefront Economy Merchants say access to new customers is the single biggest benefit, followed by gains in personalization, repeat purchases, and sales growth. More and more, AI agents are handling discovery, comparison, and even checkout. That creates what PayPal executives describe as the “invisible storefront”: a system in which transactions occur without a human ever visiting a merchant’s site. For businesses still optimizing around SEO, paid acquisition, and front-end UX, the shift risks misaligned investment. Venkatesan says LLMs are increasingly capable of refining messy product catalogs while applying broader contextual understanding to connect consumer needs with merchant inventory. By pairing world knowledge with structured commerce data, these systems can better interpret nuanced requests and translate them into viable purchase options. “The agent reasons through the first layer, then queries the second,” Venkatesan says. Despite gaps in operational readiness, trust in AI representation is higher than expected. Some 71% of small businesses and nearly 90% of large enterprises say they “mostly” or “completely” trust AI systems to represent their products accurately. Companies are also reallocating resources by hiring AI talent, upskilling teams, and leaning on platform providers to accelerate deployment. That confidence, however, comes with caveats. Data security ranks as the top concern across segments, particularly among large enterprises, where 28% cite it as a primary barrier to investment. A Growing Divide or a Market Reset? The report also reveals a split in how businesses interpret the opportunity. Large enterprises tend to believe their scale and resources will enable them to capture a disproportionate share of agent-driven sales, with 87% expecting bigger players to pull further ahead. Small businesses, on the other hand, see the opposite potential. Roughly 78% believe a rise in agentic AI could help them compete more effectively with larger rivals by improving discovery and customer access. Edmonds, PayPal’s agentic commerce VP, says consumers are already using LLMs as a core discovery layer to evaluate products and make purchasing decisions—regardless of whether merchants are properly positioned within those systems—creating growing competitive pressure to adapt. But he cautions against businesses embracing agentic commerce as a branding exercise rather than a measurable strategy that involves testing for tangible results and working with experienced partners. For PayPal, the opportunity sits at the center of that transition. As agents increasingly act as a bridge between consumers and merchants, the company is positioning itself beyond payments and into identity, trust orchestration, and dispute resolution. Merchants appear receptive to that role. Businesses are split on expectations: Some want payment providers to lead the transition, while others see them as foundational players responsible for delivering infrastructure and standards. “Visa and Mastercard have been influential in the design of the rails for decades,” Edmonds says, positioning PayPal’s ambitions in agentic commerce as similar to those legacy payment networks: building foundational infrastructure, standards, and trust without dictating market winners directly. He argues that long-term success will depend less on centralized control and more on interoperable systems, asserting, “Open protocols are how you get an ecosystem that’s trusted by merchants, consumers, agents, and regulators at the same time.”