How Banco Bradesco Cuts Coding Time in Half with Databricks Assistant

Banco Bradesco operates at a scale that makes data engineering complexity acute: 99.5 million account holders, 85,000 employees, and a data analytics community of more than 1,000 people. The analytics team maintained a data platform that supported batch and streaming data across a large organization, but the work came with consistent friction. Complex and lengthy code development, documentation overhead, and time-consuming debugging were slowing productivity. More fundamentally, the team wanted to democratize access to data — enabling business users who did not know how to code to contribute and extract insights without routing requests through engineering.

The team adopted Databricks Assistant, a context-aware GenAI companion embedded in the Databricks Data Intelligence Platform, and deployed it to all 500+ current users of the platform — including non-technical business users. The tool operates across the full data engineering workflow: writing new code, translating SAS queries to Python or SQL, generating unit tests, guiding exploratory data analysis, optimizing and refactoring existing code for readability and performance, and producing documentation. Engineers can instruct it in natural language starting from the ground up, and it provides detailed explanations that make the logic traceable rather than just the output.

The most direct productivity signal was a 50% reduction in development and analytical task time. Fixing errors that previously required hours of debugging now happens with guided natural language interaction. Writing unit tests, historically a discipline that took engineers out of their primary workflow, is now handled within the same environment. SAS-to-Python translation, a conversion task that had previously created a distinct workload, became fast and reliable. Platform tools including Spark Declarative Pipelines further reduced error rates by helping both technical and non-technical users avoid common mistakes.

The broader transformation was in who can now use the platform. Non-technical business users who previously depended on data engineering support can now create solutions, run queries, and generate insights independently. 'Even noncoders can transform data like seasoned engineers,' in the words of Lead Data Engineer Pedro Antonio Boareto. With 500+ users across data engineering, SQL, BI analysis, and data science all using the tool, collaboration has spread across the organization and business decision-making has become less dependent on technical gatekeepers.

Banco Bradesco continues to develop the platform, exploring how Databricks's broader AI capabilities can support its mission to make financial opportunity accessible to all Brazilians — maintaining its position as Latin America's third-largest bank while expanding the intelligence embedded in its operations.