Google's Universal Commerce Protocol Puts AI Agents in Charge of Checkout - With Travel Up Next
Google just introduced the Universal Commerce Protocol (UCP), a transactional layer that lets AI agents handle discovery, checkout, and post-purchase support inside a single conversation. No switching tabs, no bespoke one-off integrations for each stop in the funnel.
UCP will show up in eligible product listings across Google Search's AI mode and Gemini. Purchases can complete in-chat using shipping and payment details stored in Google Wallet.
What UCP Actually Does
UCP removes the need for custom connections between different assistants and services during a buying journey. It standardizes how agents find products, take payment, and trigger fulfillment so teams don't rebuild commerce logic for every new interface.
Retail is the starting point, built with partners like Walmart, Target, and Shopify. But the architecture is meant to scale into more complex transactions.
Why Travel Is Squarely in Scope
Google separates payment instruments from transaction handlers inside UCP. That design makes it easier to extend from simple product sales to multi-step bookings like flights, hotels, and ancillaries.
Crucially, businesses remain the Merchant of Record and keep ownership of customer data, fulfillment, and the post-purchase relationship. That keeps control with suppliers as agents take on more of the buying flow.
Payments Stack: UCP + AP2
More than 20 industry players back the effort, including Visa, Mastercard, Stripe, Adyen, and American Express. PayPal says it will support both UCP and Google's Agent Payments Protocol (AP2), giving merchants a single integration path into multiple AI platforms.
AP2 is an open protocol to securely initiate and complete agent-led payments across platforms, including scenarios like coordinating a flight and hotel under a single budget. For an overview, see Google's write-up on AP2 here.
Interoperability With Travel Infrastructure
UCP is compatible with the Model Context Protocol (MCP), which providers like Sabre and Amadeus have been adopting. MCP supplies the context agents need before a transaction happens - inventory, rules, and constraints that reduce misfires later in checkout.
Google also teased an agent-driven booking tool with partners such as Expedia and Marriott. Expect deeper experiments that move from simple purchases into live inventory, fare rules, and rebooking logic.
The Competitive Push
Microsoft rolled out Copilot Checkout powered by PayPal, enabling in-chat purchases. OpenAI launched Instant Checkout in ChatGPT with Stripe and Shopify and added interactive apps from travel players like Booking.com and Expedia.
The race is clear: whoever owns the transaction layer inside AI conversations influences what products get found, trusted, and bought.
New Merchant Tools From Google
Direct Offers: an ads pilot that lets brands surface context-aware discounts inside conversational search. Business Agents: branded assistants merchants can embed on their sites for customer support.
Google is also launching Gemini Enterprise for CX to help retailers and restaurants manage service and logistics. The direction is steady: move transactions into conversational interfaces, then expand to more complex bookings over time.
What Teams Should Do Now
If you own product, engineering, or operations, treat agent-led checkout like a new sales channel with its own constraints and signals. Keep control of your data, pricing logic, and policy enforcement while reducing friction for the buyer.
- Data readiness: standardize titles, attributes, availability, and pricing. For travel, nail room/seat types, fare rules, and change/cancel policies.
- MoR and policy: confirm you stay Merchant of Record. Define refunds, disputes, and SLA triggers for agent-led orders.
- Payments: enable wallet tokens and multi-processor support. Plan for fraud controls and step-up flows without breaking chat UX.
- Fulfillment and support: expose clear status events, returns/rebooking endpoints, and live issue-resolution paths.
- Risk and safety: rate-limit sensitive actions, log agent decisions, and require explicit user consent for high-impact steps.
- Observability: track discovery-to-purchase metrics inside chat, plus error rates, drop-off points, and human handoff frequency.
- Pilots: start with a narrow catalog or a single itinerary type, then expand as policies, data quality, and conversion stabilize.
Bottom Line
UCP gives AI agents a consistent way to complete transactions while keeping merchants in control of money flow and customer relationships. For travel, the design choices point to bookings, changes, and support happening end-to-end inside a conversation.
If your team is building agent-led commerce, this is the moment to align data standards, payments, and post-purchase ops - before the bots start buying at scale. Looking to upskill your team on practical AI workflows and agent tech? Explore our latest programs here.
Your membership also unlocks:
