Itaú expands ia.i generative AI with IARA governance platform
Itaú has introduced ia.i, a conversational generative AI interface that reached approximately 300,000 customers in its first phase. The Brazilian bank plans to make the service available to its entire customer base by the end of 2026, combining text and voice interactions across checking accounts, credit, cards, insurance and other services.
The initiative is part of Itaú’s AI First strategy for both customers and employees. Alongside ia.i, the bank has built IARA, a governance and orchestration platform that gives teams controlled access to different AI models under internal policies. OpenAI is one of the model providers available through the platform.
A single interface for financial interactions
Itaú began introducing Inteligência Itaú capabilities for specific financial needs in 2024. ia.i is described as the next stage: a conversational layer that can address a question about an account-statement expense, support investment advice and enable transactions through channels including WhatsApp. When needed, it can connect customers with specialised human support.
The bank says its model strategy is designed for different workloads rather than a single AI system. Simpler interactions can prioritise cost and latency, while more complex requests can use reasoning and grounding appropriate to the response. GPT-5.x models are among the components used in Customer DNA, Itaú’s financial-context layer for approximately 70 million customers.
Customer DNA is intended to represent customer needs, habits and life stages. Itaú says it is maintained and enriched under governance and responsible AI controls so interactions can become more relevant over time. The visible experience combines interactivity, multimodality across text, voice, images and clicks, and individualisation.
IARA supplies control for regulated AI deployment
IARA sits behind the customer-facing services as a layer for governance, observability and access to multiple models. Carlos Eduardo Mazzei, Itaú’s CTO, said the platform addresses token-cost control, response accuracy and the application of responsible AI policies in a highly regulated environment.
The platform also supports a legal-department solution that automates document reading, analysis and interpretation. That system operates on a repository of more than 90 million documents and executes more than 2 million tasks each month. Itaú reports efficiency gains of 20% to 80% across stages of the department’s processes.
Responsible AI principles apply from development through continuous monitoring, with the bank citing safety, transparency, ethics and regulatory compliance as objectives. Mazzei also highlighted the need to preserve information within the bank’s context, provide secure model access and monitor token efficiency so that deployments remain economically viable.
Internal teams are part of the AI First plan
The same intelligence layer is also being used by specialists, managers and customer-service teams. Itaú lists meeting summaries, meeting preparation, issue understanding, and the development of plans and alternatives among the types of work the tools can support.
The bank expects the rollout to create three times more time for customer relationships, make service up to twice as fast and make conversations up to three times more effective. Future ia.i capabilities are planned to include product simulations and sign-ups, plus recurring analyses of income, expenses and habits.
For businesses deploying generative AI in regulated operations, Itaú’s approach illustrates that customer interfaces and internal productivity tools can be developed alongside a shared control layer. Model choice, observability, responsible AI policies and token economics need to be designed into the operating model before wider deployment.