Google published the Universal Commerce Protocol on January 11, proposing a common way for AI shopping interfaces to communicate with retailers and payment providers. UCP covers the commercial workflow around a purchase, including discovering capabilities, building a checkout and managing an order.
The technical introduction names Shopify, Etsy, Wayfair, Target and Walmart among the collaborators. Google says the protocol is open source and designed to work with existing retail infrastructure.
A store describes what it can do
UCP organizes commerce functions into capabilities, with extensions for specialized behavior such as discounts. A business publishes a profile that lets an agent discover supported features and payment options.
That addresses a recurring integration problem: a retailer otherwise needs to build custom connections for each new shopping interface. A shared description of available functions can reduce that repeated work, provided the services implement the same capabilities consistently.
The protocol supports several ways for systems to communicate, including APIs, Agent2Agent and the Model Context Protocol. Google also describes an embedded checkout option for retailers that want to retain a customized purchasing experience.
The retailer remains the merchant of record and keeps its business logic. An AI interface can help assemble the transaction without becoming the seller responsible for the underlying goods.
Payment instruments and processors are separate
UCP distinguishes the instrument a customer uses to pay from the handler that processes the payment. That makes room for different providers without defining one mandatory payment rail.
It is also compatible with AP2, Google’s Agent Payments Protocol. The two have different jobs: UCP describes commerce functions, while AP2 provides a framework for evidence of a user’s purchase authorization.
The announcement gives developers a specification and sample implementations. It does not mean every participating retailer immediately supports every feature in every AI application.
For merchants, the useful test is concrete: can an agent read current product information, apply the correct commercial conditions and reach a checkout that the retailer can fulfill?
