This post discusses how Pinterest enhanced its Text-to-SQL capabilities by implementing Unified Context-Intent Embeddings, which help analysts translate natural language queries into validated SQL efficiently. This system intelligently retrieves relevant analytical tables and patterns, enabling quick and reliable data analytics across Pinterest's vast data warehouse. By leveraging historical query knowledge and sophisticated embeddings, the Analytics Agent reduces the time analysts spend exploring tables and debugging, thus streamlining data analysis on the platform.