Memorandum
- From
- Derek Weston via Fast Company
- Date
- Filed
- Business·5 min to read
- Re
Google's Universal Commerce Protocol Sets Standard for AI Agent Shopping
ReGoogle's Universal Commerce Protocol Sets Standard for AI Agent Shopping
Google's UCP and AP2 protocols, developed with major retailers, aim to solve the payment glitch that derails AI-driven purchases by enabling pre-authorized spending and card-linked installments.
Google has introduced a new open standard designed to let AI agents shop on behalf of consumers without getting stuck at checkout, a problem that threatens to hold back the emerging market for agentic commerce. The Universal Commerce Protocol, or UCP, launched in January 2026, gives AI agents a common language for the entire shopping journey, from product discovery to checkout and after-purchase support. It was developed in collaboration with Shopify, Target, Walmart, Etsy, and Wayfair, and has already drawn backing from dozens of companies across payments and retail.
The protocol addresses a specific failure point that industry observers call the payment glitch. That happens when an AI agent tries to buy something for a consumer but a failed credit application stops the transaction. Traditional buy now, pay later services require shoppers to apply for new credit at the point of sale. When a human is shopping, a declined application does not have to kill the sale because the person can choose another payment method. But when an AI agent is acting, a failed authorization often ends the purchase entirely. The agent simply moves on, and the merchant never knows it had a ready buyer.
Over time, those failures can train agents to stop recommending certain merchants altogether, making the problem more serious for retailers. The stakes are high. Bain predicts that agentic commerce could account for roughly 15% to 25% of total U.S. e-commerce sales by 2030, a range of $300 billion to $500 billion. Merchants that cannot complete agent-driven purchases risk being left out of that growth.
UCP is paired with a companion standard called the agent payments protocol, or AP2, which creates a verifiable audit trail for every transaction. Together, the two protocols are built around pre-authorized spending limits and payment methods that the shopper establishes upfront. When an agent finds a product that matches the shopper's criteria and completes the purchase, it executes against those pre-set rules rather than triggering a fresh credit decision at checkout. That approach provides what today's buy now, pay later systems lack: payment certainty.
The new infrastructure does not mean buy now, pay later has to disappear. Under the protocols, installments can be linked to an existing credit card, eliminating the need for a new application, a credit check, or a separate account. The full purchase amount is authorized against the card's limit at the time of sale. The shopper pays it down over time, while the merchant gets paid in full upfront. Banks also have a stake in this shift. Every traditional buy now, pay later purchase sends a cardholder to a new lender, costing the card issuer the transaction and the customer relationship. With card-linked installments, the purchase stays on the consumer's existing card, so the issuer keeps interchange revenue and engagement without underwriting a new loan or taking on added risk.
Consumer attitudes suggest a willingness to adopt the technology, but with conditions. A recent PYMNTS Intelligence survey conducted with Splitit found that 61% of U.S. adults would consider letting an AI shopping assistant recommend a pay-later option for at least one common purchase category. Among Gen Z consumers, that figure rose to 80%. The same survey found that consumers want control. They want to review, modify, approve, or reject an agent's recommendation before finalizing anything. That distinction matters because authorization must operate within limits the shopper sets in advance, not through unchecked machine judgment.
As agents take over more purchase decisions, they are expected to favor payment methods that authorize instantly against pre-approved credit, which could keep card issuers in the loop and shift momentum in banks' favor after years of buy now, pay later chipping away at card usage. For merchants, the next step is to join the UCP waitlist, with priority given to those that have accurate, complete, and machine-readable product data. Building payment infrastructure that can execute reliably without triggering a new credit decision at checkout is emerging as a practical requirement for participating in agentic commerce.
