Itaú builds ia.i, a conversational AI layer, with OpenAI models

ia.i: one place for banking and advice

Itaú first introduced Inteligência Itaú capabilities for specific financial needs in 2024. ia.i is the next stage: a complete conversational layer, activated by text or voice, that brings interactions related to checking accounts, credit, cards, insurance and other services into one environment. It also connects customers to specialized human support whenever needed.

The first phase reached approximately 300,000 customers. The bank is using this period to learn about the real relationship between customers and the product. The assistant covers situations from a question about an expense in an account statement to investment advice and convenient transactions on WhatsApp.

The solution rests on three principles: interactivity, multimodality (text, voice, image and clicks in the same experience) and individualization. Carlos Eduardo Mazzei, CTO of Itaú, says the aim is "making the technology almost unnoticeable". A study by Itaú with Grupo Consumoteca found that 80% of Brazilians want to better organize their financial lives, and that 65% would like technologies that act as a financial copilot.

Customer DNA and a range of models

At the center of the experience is Customer DNA, a layer of financial context that helps the bank understand the needs, habits and life stages of approximately 70 million customers. Under governance and responsible AI controls, GPT models help maintain and enrich this representation. According to the source, GPT-5.x models are part of Customer DNA.

Mazzei says the breadth of the product guided the choice of models. Simple cases need to be efficient in cost and latency, while more complex ones require reasoning and grounding to produce an appropriate answer. Using a range of models rather than a single one made it possible to deliver the complete solution in one product.

IARA: governance at bank scale

To operate AI at the scale of a bank, Itaú built IARA, a governance and orchestration platform that gives teams controlled access to different models in line with internal policies. OpenAI is among the model providers available in it. According to the source, IARA addresses three problems at once: cost control per token, accuracy of responses and application of responsible AI policies.

The legal department is one example. The platform automates document reading, analysis and interpretation on a repository of more than 90 million documents. It executes more than 2 million tasks per month, with efficiency gains of 20% to 80% in stages of the department's processes.

Mazzei stresses that Responsible AI is "a discipline we take great care with", including reducing any risk of bias that may arise in frontier models. Responsible AI principles guide the lifecycle of solutions, from development to continuous monitoring.

Internal use and what comes next

The bank's AI First strategy also covers staff. The same intelligence layer behind ia.i assists specialists, managers and customer service teams, for example with summaries before meetings, meeting preparation and developing plans and alternatives. Itaú expects three times more time available for customer relationships, service up to twice as fast and conversations up to three times more effective.

Planned capabilities for the next stages of ia.i include product simulations and sign-ups, as well as recurring financial analyses of income, expenses and habits. For other organizations in regulated sectors, the source's lessons point to building the governance layer before scaling, choosing several models rather than one, and monitoring token efficiency from the start so the solution stays economically viable.