Lakebase Search: Build AI Retrieval for Agents with Lakebase Postgres
Databricks LakeBase Search adds native vector and BM25 search to Postgres for AI agents
Databricks introduced LakeBase Search, two new Postgres extensions (LakeBase Vector and LakeBase Text) that bring optimized vector search and BM25 full-text scoring directly into their Lakebase managed Postgres. The product targets agentic workloads where agents simultaneously read and write searchable data at high concurrency, treating search as an OLTP workload rather than a separate analytics concern. The significance is eliminating the need for a separate vector database by integrating search natively with operational data via Databricks' separated storage-compute architecture.