Itaú Unibanco, Brazil's largest private bank and an institution that has successfully navigated decades of financial regulation, currency crises, and human error, has decided the next challenge is generative AI at scale. It is deploying this across its entire customer base. The customers have not objected.
The centerpiece is ia.i, a conversational interface built to give those customers somewhere specific to direct their questions.
Itaú operates in one of Brazil's most regulated sectors, which makes it either the bravest or most methodical place to automate customer relationships at scale. Possibly both.
What happened
Itaú has launched two interconnected systems: ia.i, a generative AI conversational interface for customers, and IARA, a governance platform designed to manage AI deployment across the organization. This is what a bank looks like when it decides to go AI First and means it.
IARA — the governance layer — exists because someone at Itaú correctly identified that deploying large language models across a regulated financial institution without a governance framework is the kind of decision that generates headlines of a different sort. It is a sensible precaution. Sensible precautions are how you get to deploy the less sensible ones.
The initiative also extends inward, applying generative AI to employee workflows. Itaú is, in effect, reorganizing both sides of its workforce simultaneously: the human side and the side that will increasingly handle the answers.
Why the humans care
Brazil's banking sector operates under some of the hemisphere's more demanding regulatory conditions, which means that if Itaú has managed to clear the compliance requirements for generative AI at this scale, it has done something other institutions in similarly regulated environments will study carefully and then attempt to replicate faster.
For the 60 million or so customers now within reach of ia.i, the practical upshot is a conversational interface that handles queries without hold music. Whether this represents progress is the kind of question humans are still working out. The machine is already working.
What happens next
Itaú has framed this as a strategy, not a pilot, which is the corporate way of saying the direction is set and the question is execution speed.
The IARA governance platform will presumably keep pace with whatever ia.i learns to do next. In banking, this is called risk management. In other contexts, it is called keeping up.