How Itaú is building generative AI for its entire customer base

OpenAI’s case study says Itaú is applying an AI First strategy through ia.i, a conversational interface (text or voice) spanning checking, credit, cards, insurance and more, plus the IARA governance and orchestration platform for employees. Roughly 300,000 customers are in the first phase; the goal is 100% of the customer base by the end of 2026. CTO Carlos Eduardo Mazzei emphasizes simplicity and making the technology almost unnoticeable.

IARA gives teams controlled access to multiple models under bank policies. A legal solution on the platform works against a repository of more than 90 million documents and executes more than 2 million tasks per month, with reported efficiency gains of 20% to 80% in process stages. Customer DNA — financial context for about 70 million customers — uses GPT models among other components under responsible-AI controls. An Itaú/Consumoteca study cited says 80% of Brazilians want better financial organization and 65% want a financial copilot.

This is the grown-up version of a bank chatbot: multimodel routing, governance before scale, and a clear end-of-year coverage target. Regulated finance is exactly where “ship the demo” fails and IARA-style platforms matter.

We’re bullish on useful AI that disappears into advice and transactions — Mazzei’s standard. The risks are familiar: bias, hallucinated advice, and data residency. Itaú’s framing treats those as product requirements, which is the right posture if monitoring stays real.

I’m watching whether the 100% customer-base goal holds by year-end, and how much of the legal-department efficiency claim survives outside OpenAI’s case-study framing.

Planned next stages include product simulations and sign-ups plus recurring analyses of income, expenses, and habits. Internal goals cited include more time for customer relationships and faster, more effective service conversations.