Databricks has launched a new generation of Online Feature Stores powered by Lakebase, offering scalable, low-latency access to feature data for real-time machine learning and generative AI. This session introduces the new architecture, highlights key differences from previous feature store implementations, and explains when and how to use them effectively. Through live demos, we’ll explore publishing feature tables, enabling streaming updates, and serving features to real-time applications using Unity Catalog and Lakehouse-native workflows.