Itaú Unibanco, Brazil’s largest private bank, is building generative AI for its whole customer base, not for a small pilot group. According to OpenAI, the bank’s effort rests on two pieces: ia.i, a conversational interface, and IARA, a governance platform that oversees how the AI is used. Both fall under what Itaú calls its “AI First” strategy, which covers customers and employees.
The details OpenAI shares are thin. Still, the direction is clear, and it says a lot about where AI in banking is going.
What Itaú Is Building
As detailed by OpenAI, the plan has two parts:
- ia.i: a conversational interface that lets people interact with the bank in plain language instead of clicking through menus and forms.
- IARA: a governance platform that manages how AI is deployed, monitored and controlled across the company.
- AI First strategy: the umbrella approach that applies both tools to customer-facing services and to internal work for employees.
What stands out is the pairing. Itaú isn’t only shipping a chatbot. It’s shipping the control system that sits behind it, and that tells you how seriously the bank takes the risks.
5 Reasons This Launch Matters
- The scale is unusual. Most banks start with generative AI behind the scenes: summarizing documents, helping call center staff, drafting code. Itaú is aiming at its entire customer base. That’s a far bigger bet, because every mistake a customer sees costs trust, and banks run on trust.
- Governance gets top billing. OpenAI puts IARA right next to ia.i in the headline, so governance isn’t an afterthought here. In a regulated industry, the hard part isn’t building a model that answers questions. It’s proving to regulators, auditors and customers that the answers are accurate, fair and traceable. A dedicated governance platform is how a bank turns AI experiments into something it can actually defend.
- Brazil is a tough place to try this. OpenAI calls banking one of Brazil’s most regulated sectors. Financial institutions answer to the Central Bank of Brazil and have to follow the country’s LGPD data protection law. If Itaú can roll out conversational AI at scale under those rules, it’s a strong sign the approach can work in other tightly regulated markets.
- Employees count as much as customers. The AI First strategy covers both sides of the counter. That matters because internal use is usually where companies learn fastest. Employees can spot errors, flag edge cases and push the tools in ways customers never would, and those lessons feed back into the customer-facing product.
- Conversation is becoming the new banking interface. Apps replaced branches for routine tasks. Now chat-style interfaces are starting to compete with apps. ia.i fits a wider industry shift where you just ask for what you need (a balance, a transfer, an explanation of a fee) instead of hunting for the right screen.
How It Fits the Bigger Picture
Big financial institutions around the world are racing to put generative AI to work, and most are moving carefully. The usual pattern is internal tools first, customer tools much later. Itaú is shrinking that gap.
The fact that OpenAI is telling this story also matters. OpenAI has been pushing hard into enterprise deals, and a showcase customer in heavily regulated banking is exactly the kind of proof it wants to show other industries with strict compliance needs.
What We Don’t Know Yet
The source leaves several key questions open:
- Rollout timing: it’s not clear how many customers can use ia.i today versus later.
- Specific capabilities: OpenAI doesn’t list what tasks ia.i can handle, such as whether it can make transactions or only answer questions.
- How IARA works: the governance platform’s actual controls, like human review, audit trails or model monitoring, aren’t described.
- Results: there are no figures on adoption, accuracy or cost savings.
Those gaps matter. Announcing a governance platform is one thing. Showing it catches problems before customers see them is another.
What Comes Next
Keep an eye on whether other Latin American banks follow Itaú’s lead, and whether regulators start treating governance platforms like IARA as a baseline expectation instead of a nice extra. If this works, “AI First” could stop being a slogan and become the default way large banks operate. You can find more details in OpenAI’s original piece.