AI Agents Will Split Markets Between The Cheapest And The Best Fit

Javier Rodríguez Soler does a convincing impression of the customer that may soon reshape banking. Conjuring an AI agent sizing up a bank on someone’s behalf, he slipped into its voice: “I don’t mind if your brand is famous, if you have beautiful branches, you have very nice people... I’m an agent. I have no sentiment. I never get frustrated. And as soon as you are safe, I trust you.”

BBVA, the Madrid-based bank where Rodríguez Soler heads corporate and investment banking worldwide, is preparing for that customer rather than waiting for it. The bank, whose operations stretch from Mexico to Turkey, built its customer assistant, Blue, on OpenAI’s models, and in July it completed its first purchase initiated by an AI agent, working with Visa.

I’ve been interviewing people in banking, payments, travel, and equipment finance about agentic commerce, and their answers suggest that agents like the one Rodríguez Soler voiced will split many markets in two. They are likely to reward the cheapest offer at one end and the best fit at the other while the comfortable middle erodes, and the winners will be the companies that choose their ground early.

The Comfortable Middle

In 1980, Michael Porter argued in Competitive Strategy that a company should choose between low cost and differentiation, and he warned against getting “stuck in the middle,” yet plenty of companies ignored him and did fine. They were inexpensive enough and good enough, and their customers rarely had the time to find anything clearly better, because information has never flowed like water.

Banking, Rodríguez Soler said, has lived on “margins on friction,” such as deposits that sit earning little because moving them is a bother. Agents may compete those margins away, since software compares every offer every time and beautiful branches don’t sway it.