Itaú generative AI is being rolled out as a customer-facing and internal platform designed to simplify everyday financial tasks for millions of users while remaining compliant with strict governance requirements. The São Paulo-based bank has introduced two complementary elements — ia.i, a conversational, multimodal assistant for customers, and IARA, an internal governance and orchestration platform — to scale generative AI across retail banking and internal operations without compromising cost controls or Responsible AI safeguards.
At the customer level, ia.i functions as a single conversational layer that accepts text, voice, images and clicks to handle queries and actions related to checking accounts, credit, cards, insurance and investments. Itaú positioned ia.i as an evolution of its prior Inteligência Itaú capabilities and deliberately designed the assistant to be interactive and individualized. Central to that personalization is Customer DNA, a contextual layer representing the needs, habits and life stages of roughly 70 million customers; GPT models are used to maintain and enrich that profile so responses become more relevant over time.
The bank began a measured rollout of ia.i to about 300,000 customers in an initial phase, using that deployment to study real usage patterns. Itaú said the assistant manages a broad spectrum of interactions: from simple account-statement questions to investment guidance and transactions initiated via WhatsApp. The company cited consumer research from Grupo Consumoteca showing demand for such features — 80% of Brazilians want better organization of their finances and 65% want technologies that act as a financial copilot — and positioned ia.i to meet that need.
Behind the consumer experience sits IARA, an internal platform focused on governance, orchestration and observability. IARA gives teams controlled access to multiple model providers, including OpenAI, and was built to confront three practical constraints at scale: token-level cost control, response accuracy and enforcement of Responsible AI policies. The platform enables teams to choose models appropriate to specific tasks, balancing cost, latency and reasoning requirements. As CTO Carlos Eduardo Mazzei noted, “OpenAI’s models were very important in delivering this experience, because they were able to operate across this range of solutions.”
IARA also supplies the policy controls and operational visibility necessary for a regulated financial institution. The platform already supports a legal-department solution that processes more than 90 million documents and executes over 2 million tasks per month, delivering efficiency improvements that vary from 20% to 80% across different process stages. These figures illustrate how governance and orchestration can make large-scale model use economically viable while preserving sensitive information.
Itaú has applied the same intelligence layer internally to help specialists, managers and customer service teams. Internal use cases include automating routine tasks, generating pre-meeting summaries, preparing staff for customer conversations and producing plans and alternatives that previously required substantial manual effort. The bank projects measurable efficiency gains: up to three times more time available for relationship-building, service speeds up to twice as fast, and customer conversations that can be up to three times more effective. Mazzei said these tools free relationship managers to “focus on what matters most: creating value for our customers.” He also described the bank’s technology aim as making capabilities “almost unnoticeable,” emphasizing simplicity and direct benefit for users.
Itaú has emphasized several operational lessons from its program: prioritize simplicity over unnecessary sophistication, use multiple models to match diverse tasks, build governance before scaling, treat cultural alignment and bias mitigation as product requirements, and deploy AI internally to elevate staff toward higher-value work. The bank has cited GPT‑5.x models as part of the Customer DNA architecture and said that controlling token use is vital to keeping the overall solution economically sustainable.
Looking ahead, Itaú plans to make ia.i available to its entire customer base by the end of 2026. Future capabilities under development include product simulations, on-boarding or sign-ups, and recurring financial analyses of income, expenses and habits. For Mazzei and the bank, the initiative is part of a broader transformation: responding faster to customer needs and helping people engage with financial products in simpler, more direct ways.
Itaú’s approach demonstrates how large financial institutions can combine conversational assistants and strict governance to scale generative AI across millions of retail customers and complex internal operations, balancing innovation with compliance and cost discipline.
Source: Read the original source
Leave a Reply