The News, Explained
OpenAI published a case study of Brazilian bank Itaú’s use of generative AI on October 16, 2026. The bank reports that its conversational banking service, ia.i, reached approximately 300,000 customers in its first phase. Availability to the entire customer base is an expectation for the end of 2026. The date of the case study and the target for wider service availability describe different events. Source
Three figures with different meanings
The source also refers to approximately 70 million customers. That figure describes the scope of Customer DNA, a financial-context layer used to understand customers’ needs, habits and life stages. In plain language, this is background information that helps the bank understand a customer’s situation. The 300,000 figure describes first-phase ia.i reach; 70 million describes context coverage; the entire customer base is the intended future audience. Treating them as one current user count would misread the service’s reach. The figures and expansion plan are the bank’s account in the case study. Source
The customer conversation and internal management
ia.i is a text- or voice-activated layer for interacting with banking services. Here, a layer means a point of contact between customers and multiple services. The source describes bringing checking-account, credit, card, insurance and other interactions together and connecting customers to specialized human support when needed. One example it gives is asking about an expense in an account statement. For readers, that illustrates how a banking entry point can be designed around a question alongside menu navigation. This explains the announcement; it is not a hands-on review. Source
IARA manages bank teams’ access to multiple AI models under internal policies. While ia.i is the point where a customer asks a question, IARA coordinates model use inside the bank. The source names token-cost control, response accuracy and responsible AI policies among its management aims. Tokens are small units of text processed by a model, so their usage is connected to operating costs. A simpler customer interface sits alongside these operational choices. These are the bank’s described controls and aims; OYOPICK has not independently tested their accuracy or safety. Source
OYOPICK’s Take
We find reader value in seeing the entry point to a financial service together with the operations behind it. Starting with an everyday question about an account-statement expense could help people who are less familiar with financial terminology and menus. Turning that possibility into convenience requires understandable answers and a smooth route to human help. That is why we welcome the inclusion of specialized support in the service’s design.
Assessing the year-end expansion will also require keeping first-phase reach, context coverage and actual service availability separate. We hope the bank’s future rollout reports explain which customers can have which interactions. A clearly described scope would let readers compare delivery with the stated goal and judge how close conversational banking has come to their own daily lives.