Artificial intelligence shopping agents are creating a new kind of customer at the merchant’s front door: software that may know what a consumer wants, yet arrives without many of the signals merchants have learned to trust.

Pavan HS, senior director of product, agentic commerce at Visa, told PYMNTS that merchants are approaching AI agents as another route to a sale, with the same requirements surrounding payments, fraud, loyalty, returns and customer service that apply elsewhere.

“At the end of the day, for merchants, agentic commerce is just another sales channel,” he told PYMNTS. “The two things that they care about is being trustworthy and making it convenient for their consumers.”

The channel is one that takes shape well before checkout. A consumer planning a winter trip to Norway, for example, might ask Gemini, Google’s AI assistant, to help find a jacket. The merchant then has to supply enough structured product information for the agent to connect “Norway in winter” with the attributes of a suitable coat. Different agents can retrieve that information through different protocols or through web crawling, leaving merchants to decide where and how they want their catalogs exposed.

“Some merchants might choose to sell on other AI platforms, and some merchants might choose to build their own chatbot,” the Visa executive said. “There are pros and cons associated with each of these approaches.”

Providers must understand those nuances. Merchants aren’t likely to replace their payment infrastructure to accommodate agents. The task is to make agent-initiated transactions work with systems already handling authentication, fraud, tokenization, disputes and omnichannel commerce.

Source link