Skip to main content

Banking on better decision-making with AI

Banco Bradesco employs Databricks Assistant for more productive data analysis


Time saved writing code

CLOUD: Azure

“Databricks Assistant plays a pivotal role in boosting productivity. Through pair coding and leveraging Databricks Assistant’s expertise, teams are about 50% faster in their development and analytical tasks.”

— Fabiano Kenith Seki, Lead Data Engineer, Banco Bradesco

Banco Bradesco, the third-largest banking institution in Latin America, faced productivity hurdles due to lengthy code and debugging processes. Their data analytics team sought to streamline operations and democratize data access for all users. Implementing Databricks Assistant, a GenAI-based companion, revolutionized their workflow by simplifying code writing, troubleshooting and documentation. With natural language capabilities, Databricks Assistant empowered nontechnical users to contribute effectively, leading to a 50% reduction in coding time and significant cost savings. Now all users across the organization can easily extract valuable insights from data without overreliance on engineering for support, fostering collaboration and enhancing decision-making capabilities at scale.

Lengthy code and debugging create bottlenecks to productivity

Banco Bradesco has 99.5 million account holders and a workforce of over 85,000 employees to support them. More than 1,000 employees work in the analytics community alone. Leveraging this scale, Banco Bradesco has a mission to create opportunities for any individual to realize their potential and contribute to the sustainable development of business and society.

Pedro Antonio Boareto, Lead Data Engineer at Banco Bradesco, is a technology strategist for their large data analytics team. To keep up with the banking institution’s vision, his team sought to improve their work. “We were encountering challenges with the development of complex and lengthy code and were spending significant amounts of time on documentation and debugging processes,” explained Boareto.

The data analytics team wanted to improve productivity and accuracy while accelerating insights using batch and streaming data. They also sought to democratize access to the Databricks Data Intelligence Platform so that business users who didn’t know how to code could create solutions themselves. To achieve this, they leaned on Databricks Assistant, a context-aware generative AI–based companion pair programmer. 

Databricks Assistant helps all data users quickly transform code into insights

All 500+ current users of the Databricks Data Intelligence Platform at Banco Bradesco, including nontechnical business users, can freely use Databricks Assistant, which has facilitated success across the organization.

Fixing errors and saving time on troubleshooting is a major productivity win for Banco Bradesco. Fabiano Kenith Seki, Lead Data Engineer for the large data analytics team at Banco Bradesco, credits Databricks Assistant with helping the organization quickly identify and resolve coding errors. “It assists in writing code by providing detailed explanations and instructions using natural language to help understand and implement the desired functionality. We can even instruct Databricks Assistant with natural language to write the code starting from the ground up,” explained Seki.

Platform tools such as Delta Live Tables (DLT) and other data engineering processing capabilities help users avoid errors and accelerate development for both batch and streaming data scenarios. Writing unit tests is a big part of the DevOps process, and Databricks Assistant can guide any user through the creation of tests to ensure their code is correct and reliable. The same can be said for writing SQL queries, where Databricks Assistant translates SAS queries to Python or SQL code, enabling all data analytics teams to quickly manipulate data. Databricks Assistant also provides guidance to Banco Bradesco’s data engineers and analysts on how to document their code effectively, enabling them to best prepare it for final documentation and deliverables. 

“Databricks Assistant provides guidance during experimentation, exploratory data analysis (EDA) and model development for discrete data science tasks,” added Thiago Ruiz Aniceto, Data Engineer at Banco Bradesco. “It helps in coding SQL queries using natural language explanations and debugging code. This support empowers business users to create solutions without requiring extensive coding knowledge.” Added Boareto, “Databricks Assistant optimizes and improves existing code by restructuring it to enhance readability, performance and maintainability. All these capabilities mean that even noncoders can transform data like seasoned engineers.”

Cutting coding time in half with Databricks Assistant

Alessandro Silva, Chapter Leader and Data Engineer at Banco Bradesco, believes that his team has become “much more productive” since using Databricks tools. “With the help of Databricks Assistant, we have been able to enhance productivity across data engineering, SQL, BI analysis and data science teams. We’ve not only addressed challenges around coding and debugging in general but also improved code syntax with unfamiliar libraries,” said Silva. Added Seki, “Through pair coding and leveraging the expertise of Databricks Assistant, teams are about 50% faster in their development and analytical tasks.”

When asked what he likes most about its capabilities, Aniceto answered, “I love that Databricks Assistant empowers users to leverage data for decision-making without requiring extensive technical expertise. The natural language capabilities of Databricks Assistant bridge the gap between technical and nontechnical teams.” Now Banco Bradesco benefits from collaboration across a broader user base, enhancing business decisions throughout the organization — all underpinned by the power of generative AI.