Worldline [Euronext: WLN] has introduced a payment handler that lets retailers take payments started by artificial intelligence assistants, using the open Universal Commerce Protocol as the shared technical language.
The feature sits on Global Collect, the company’s platform for international online checkout, and is aimed at shops that want to sell through conversational and agent-driven surfaces without building a separate payment stack for each one.
Agentic commerce describes a shopping path in which software acting for a person finds products, assembles a basket, and finishes a purchase under rules the shopper has set.
That model only works if catalogs, carts, and money movement can talk to one another in a predictable way. UCP, developed with Google and other industry participants, is meant to be that common layer.
A merchant publishes a profile of what it can do. Compatible AI surfaces read that profile, negotiate checkout, and hand over a secured payment credential rather than raw card numbers.
Worldline’s handler is the piece that turns that credential into a live authorization.
Retailers declare their accepted instruments once—cards, mobile wallets, and methods common in Europe—and can then take those same options on any UCP-connected assistant.
The company presents the handler as a universal adaptor: agent platforms stay on one side, merchant systems and Worldline stay on the other, and sensitive account data does not need to travel through the AI layer.
Stijn Gasthuys, who leads Global Commerce and Global Collect, framed the release as an answer to operational strain.
As assistants take a larger role in discovery and buying, he said, sellers want new demand without extra complexity.
The point of the product, in his telling, is to reuse payment rails merchants already rely on.
Gertjan Dewaele, vice president of product and technology for the same unit, described the handler as a bridge that lets firms get ready now and later expand across more agent platforms without repeating integrations.
The timing fits a broader push.
Worldline has already connected language models to payment actions through Model Context Protocol servers, so agents can create hosted checkout links, check status, or request refunds in natural language.
It has also taken part in live European trials of agent-started payments with banks and card networks.
The UCP handler is a different kind of building block: instead of a merchant’s own chatbot talking to Worldline APIs, third-party AI marketplaces can complete a sale against the merchant’s existing acquiring setup.
For retailers, the practical appeal is one configuration instead of many.
For shoppers, the intent is that an assistant can finish a purchase while the merchant remains the seller of record and Worldline remains the processor. UCP-style checkout on Google’s conversational products is rolling out market by market;
Worldline’s documentation also covers using the Global Collect handler in a merchant’s own UCP implementation so the same pattern can apply beyond a single assistant.
None of this removes the need for consent, authentication, and compliance.
The protocol’s payment model is built around negotiation of methods, acquisition of a tokenized credential, and completion on the merchant backend.
Worldline’s role is to process that last step through infrastructure sellers already know.
As more assistants become places people actually buy, payments firms that sit between agents and merchant systems will decide how fragmented that world becomes. Worldline is betting that an open protocol plus a single handler is enough to keep agent-started commerce from turning into a maze of one-off connections.