AI Agents Upend Online Checkout
AI shopping agents are putting existential pressure on Buy Now, Pay Later (BNPL) services, forcing them to aggressively iterate to defend their spot at checkout. In response, infrastructure players like Spreedly are enabling "agentic commerce" for merchants, while Pagos is turning transaction data AI-ready for richer analytics.
AI shopping agents function as autonomous software that can discover, compare, and even purchase products on a user's behalf with minimal human input. This shift from manual online shopping to automated discovery is creating a new "agentic commerce" channel. Projections suggest this market could orchestrate up to $1 trillion in U.S. B2C retail revenue by 2030. For product managers, this means the user experience is shifting from visual website design to structured, machine-readable data. AI agents don't "browse"; they parse APIs and evaluate products based on logic like price, availability, and return policies. As a result, brands that fail to provide clean, accurate product data risk becoming invisible to these algorithmic shoppers. The rise of AI agents directly challenges the visibility of Buy Now, Pay Later (BNPL) services at the point of sale. To counter this, BNPL providers like Klarna and Affirm are integrating with tokenization infrastructure, such as Stripe's Shared Payment Tokens, to ensure their installment options are presented even when an AI completes the purchase. This represents a move to embed their services at a deeper, protocol level. Infrastructure providers are building the rails for this new form of commerce. Spreedly's platform, for example, allows merchants to accept payments from AI agents through their existing payment processors, ensuring they remain the merchant of record. Meanwhile, Pagos is focused on transforming fragmented payment data from various processors into a harmonized, AI-ready format, enabling richer analytics and optimization for these new agentic workflows.
