Databricks AI Search adds Streaming Tables, Materialized Views, Managed Iceberg

Until now, Databricks customers who wanted to search over Streaming Tables, Materialized Views, or Managed Iceberg tables had to create a separate Delta representation and build pipelines to keep it synchronized. That meant an extra data layer, custom update logic, and ongoing operational overhead. With native support, AI Search connects directly to these sources: you create an index and sync it when needed, and the platform propagates changes to the index while the data stays connected as it evolves. Each table type serves distinct search use cases. Streaming Tables suit continuously arriving events like fraud detection or real-time IoT analytics; Materialized Views support personalized, session-aware search by maintaining aggregated or joined state; and Managed Iceberg tables enable cross-engine agents and unified knowledge search over open-format data without migration. The workflow itself is unchanged: the same Delta Sync index creation via UI or API now extends to these sources, so teams do not need to learn a new process.