Itaú’s ia.i combines conversation and specialized models

On October 16, 2026, OpenAI published a customer story about ia.i, Itaú’s conversational interface for inquiries about accounts, credit, cards, and insurance by text or voice. The initiative builds on the capabilities of Inteligência Itaú introduced in 2024 and connects customers with specialized human support when they need it. (OpenAI)

Conversation and financial context

The experience combines text, voice, images, and clicks. Examples described include asking about a charge on a statement, receiving investment guidance, and making transactions through WhatsApp. Carlos Eduardo Mazzei, Itaú’s CTO, explains that the bank combines models: simple inquiries require cost and latency efficiency—the response time—while others need more reasoning and grounding, meaning answers tied to relevant context. (OpenAI)

Customer DNA provides this financial context about the needs, habits, and life stages of about 70 million customers. According to the case study, GPT‑5.x is among the components that maintain and enrich this information. ia.i’s initial reach was approximately 300,000 customers; the bank’s goal is to extend the experience to 100% of its customer base by the end of 2026. (OpenAI) (openai.com)

Model access and internal use

IARA is the governance and orchestration platform that gives teams controlled access to different models in accordance with internal policies; OpenAI is among its providers. It allows the bank to monitor how solutions operate, control cost per token—the text units processed by models—and apply the bank’s responsible AI policies. (OpenAI)

The legal department’s solution works with a repository of more than 90 million documents and performs more than two million tasks per month. The case study reports efficiency gains of between 20% and 80% at stages of the department’s processes, where document reading, analysis, and interpretation are automated. (OpenAI) (openai.com)

The same intelligence layer also helps specialists, managers, and customer service teams prepare for meetings, summarize information, and develop options for customers. For this internal use, the bank expects to have three times more time to engage with customers, serve them up to twice as fast, and make conversations up to three times more effective. Simulations, product sign-ups, and recurring financial analyses are planned for future phases of ia.i. (OpenAI)