Learn to build scalable entity resolution systems that overcome common implementation challenges through this 37-minute conference talk. Explore the limitations of traditional vendor software and open-source DIY approaches, including scalability issues, single model architecture constraints, lack of metrics and explainability, and static implementations that fail to improve over time. Discover how transformer-based approaches combined with fast vector search lookups and Databricks components like Databricks Apps and Agent Eval create the foundation for composable entity resolution systems that continuously improve with experience. Watch a practical demonstration showing how to integrate these components to build an adaptive ER system optimized for your specific data requirements, presented by Databricks data science experts Ninad Sohoni and Yinxi Zhang.