Generative AI in the Court of Accounts, São Paulo

with Fabio Correa Xavier

Fabio Correa Xavier is the CIO of the Court of Accounts in the State of São Paulo, Brazil which oversees government spending for the state of 44 million people.

The role of the Court of Accounts in the State of São Paulo

Fabio Correa Xavier is the CIO of the Court of Accounts in the State of São Paulo, Brazil (known as TCESP). Xavier started by explaining the role of the organization “The Court of Accounts of the State of São Paulo plays a key role in overseeing public funds. We basically check on everyone who handles public money from government officials to contractors. We make sure that they are using the money properly, following the law, of course, and not wasting public resources.”

“Our work is very impactful, especially since São Paulo is the largest and the richest state in Brazil. We oversee a massive budget, well over $100 billion across thousands of jurisdictions.”

“From the biggest infrastructure projects to education to health care, we ensure that the money is spent, achieves what it’s supposed to do. By making sure these funds are used efficiently and transparently, we not only improve the service that millions rely on, but also build trust between the public and the government.”

Xavier shared the role of his team “As CIO, I lead about 100 tech staff, to make sure we are using the latest technologies to enhance our auditing process. So my team and I work on integrating things like AI to make our audits faster, more accurate and predictive. So it’s all about using tech, not just to keep up with the times, but to actively improve how we oversee and manage the public funds.”

Getting starting with generative AI

Xavier explained how he and his team have been using generative AI to develop a new set of capabilities for the organization, in a platform called ANIA [Assistente Natural com Inteligência Artificial]. “ANIA is designed to help us analyze documents, pull out data, and even provide insights that help us to make better decisions.” In addition Xavier identified that “It’s also allowed us to be more proactive in how we advise the entities we audit, prevent issues before they become bigger problems, to act before the problem.”

Having started by developing ANIA.pdf, Xavier outlined some of the additional components of ANIA now in use, including “ANIA.legis for dealing with legal documents, ANIA.dados for data analysis and structured data analysis, ANIA chat for secure communications. There is also ANIA.sei which integrates our electronic process systems to streamline administrative tasks. All these tools are about making our data, smarter, faster, and more insightful.”

Starting small to show value

Xavier characterized adoption of generative AI as “a significant turning point for our organization”, and emphasized that “achieving widespread support required careful planning and demonstration. Initially, support for Gen AI was not widespread in our organization, of course, so we strategically implement small pilot projects.”

“These early projects not only demonstrated the substantial improvements in our process, but also highlighted the practical benefits and efficiencies Generative AI could bring to our daily operations. The visible success of these pilots projects played a pivotal role in generating enthusiasm and support for all levels of the organization, from top management down to our auditors.”

Having built support and successfully delivered the initial components of the new platform, Xavier and his team plan to keep building capabilities in new areas. “We have another four apps planned to be delivered this year. One of them is to make use of the generative AI to read and understand the decisions of our Court of Accounts and analyze these decisions to help our employees.”

Developing AI skills within the organization

Xavier explained that “We didn’t start with a bunch of AI experts. We didn’t have anyone with these abilities. So we really focused on training all our current team should do these tasks. So we set up a training program that covered the basics of AI, and then got everyone hands-on experience with our Gen AI tools.”

“We also built a small but mighty team of developers and analysts who were really excited about AI. They were crucial in getting our AI projects off the ground and helping spread that knowledge and enthusiasm throughout the organization. So these three or four people were very important to create this culture here in our IT department.”

Xavier emphasized the importance of the excitement of the team in using new tools to successfully deliver results “I always say it’s this success is because our team is very engaged and very excited about innovation and about AI.”

“This work made us proud of our team. I love my job, and when we do things like ANIA, I love my job more. It’s the opportunity to make a difference for our world.”

The moment is now

Xavier shared his recommendation for others wanting to explore how to use generative AI in their organizations “I would say, don’t wait for the perfect moment or the perfect thing. Start with clear, achievable goals, and small projects that can prove what AI can do.”

“Make sure there is plenty of support and training for your staff and you’re really listening to the feedback they give you. Small wins are great for building momentum and show everyone the real-world benefits of bringing AI into government work. So start, that’s my recommendation, start, because the moment is now.”

Listen to this episode on any of these podcast platforms:

Spotify

Apple Podcasts

Deezer

Libsyn