Itaú Unibanco chooses Harness for governed AI-era delivery

By Nathan Brooks

Harness said on November 12, 2026, that Itaú Unibanco selected its platform to change how engineers build and ship software across the bank’s engineering organization. The deployment will begin with an initial adoption phase, followed by broader use in the months ahead.

Harness describes Itaú Unibanco as Latin America’s largest private-sector bank. Its engineering workforce numbers in the thousands, supporting digital banking, payments and platform services for millions of customers. Itaú wants to increase engineering speed and use more generative AI while preserving the governance and resilience required of a global financial institution.

Itaú plans to standardize delivery through templates with built-in governance, automated checks and policy-based controls. Harness says its platform will shorten build times, automate rollbacks and make deployments more consistent across the programming languages and cloud environments the bank uses most.

The bank also plans to put security, audit and compliance requirements inside every pipeline as default controls. The system intended to accelerate releases will therefore apply policies and verify software as it moves toward production.

In Harness’s design, verification and policy enforcement travel with the code toward production. The control set becomes part of the release path, alongside the steps used to build and deploy software.

“The future of banking will be built on software, and the institutions that win will be the ones that can move fast without sacrificing control,” Carlos Eduardo Mazzei, Itaú Unibanco’s CTO, said. Mazzei connected that balance to the bank’s competitive requirements.

What does the Harness rollout change for Itaú’s engineers?

For engineers, the immediate change is a common delivery model with governance, automated checks and policy controls built in. Harness says that model will shorten build times, automate rollbacks and improve consistency across the bank’s most-used programming languages and cloud environments.

Itaú expects the wider platform to give engineering leaders a consistent view of delivery, controls and economics. The bank wants engineers to spend less time and effort getting committed code into production, leaving more capacity for customer-facing experiences that distinguish it.

Itaú frames the adoption as tying faster, AI-supported engineering to a common operating model for delivery. Its stated priority is consistent security, audit and compliance controls across teams, programming languages and cloud environments as usage expands.

Harness’s announcement sets out a sequence that begins with initial use and extends across the complete software-delivery lifecycle. The staged structure gives the bank room to establish the common pipeline model before wider adoption. Itaú expects the resulting foundation to connect code directly to production under one set of operating rules.

Itaú has placed AI inside that roadmap. Its broader strategy applies AI across engineering processes to improve the speed and quality of software construction and delivery. Harness says templates, verification and policy controls will govern how the resulting software reaches production.

André Palma, Itaú’s director of platform engineering, said the bank continually evaluates capabilities that strengthen software construction and delivery as product cycles accelerate. His comments positioned the Harness selection as one step in an ongoing technology strategy.

Arpit Patel, Harness’s chief customer officer, said Itaú’s scale and governance requirements demand a delivery platform built for the AI era. He said Harness would support the bank’s evolving DevOps foundation as adoption expands.

The next test comes as the rollout moves beyond initial adoption. If governed templates deliver shorter build times and reduce the effort of moving code into production without weakening security, audit or compliance controls, Harness could become the common operating layer for AI-supported engineering at Latin America’s largest private-sector bank.

The broader deployment will show whether that model holds across thousands of engineers, multiple programming languages and different cloud environments. Success would establish a governed foundation for faster, AI-supported software delivery without sacrificing compliance.