Itaú is applying an “AI First” strategy to both customers and employees, with a conversational layer called ia.i that spans checking accounts, credit, cards, insurance, and other services. The bank’s CTO, Carlos Eduardo Mazzei, frames the goal as making the technology almost unnoticeable: direct, simple, and useful rather than model-centric. The project grew out of earlier Inteligência Itaú capabilities introduced in 2024.
The first phase of ia.i has reached about 300,000 customers, and Itaú says its goal is full coverage by the end of 2026. At the core is “Customer DNA,” a financial-context layer that helps the bank tailor interactions for roughly 70 million customers. OpenAI’s GPT models are used across a spectrum of tasks, from simpler ones needing low cost and low latency to more complex reasoning-heavy conversations.
Behind the experience sits IARA, Itaú’s governance and orchestration platform. It gives teams controlled access to models while addressing token costs, response accuracy, and responsible AI policies in a highly regulated sector. In one example, the legal department uses the platform to read and interpret documents from a repository of more than 90 million files, executing over 2 million tasks per month with efficiency gains of 20% to 80% in some process stages.
The report is a single source from OpenAI, so there are no independent accounts to compare. Still, it describes a deliberate architecture: a consumer-facing assistant, a personalization layer, and a governance platform designed to scale AI across one of Brazil’s largest banks while keeping human support available when needed.