ReleaseBytes
databricks Databricks Release Notes ·

Databricks AI Search Enhances Querying with Struct/Map Column Support and Filter-Only Queries

aidatabricksgaengineer
feature

Databricks AI Search now supports indexing and filtering for struct and map columns, expanding the range of data types that can be directly searched. This update allows users to perform filter-only queries, enabling retrieval of matching rows based solely on filter criteria without requiring vector or keyword retrieval. These new capabilities provide greater flexibility and precision when searching data within Databricks AI Search indexes. Developers and data architects using Databricks can leverage these features to build more robust and targeted search applications.

  • →Index and Filter Struct and Map Columns
  • →Execute Filter-Only Queries
Features (2) ›
  • Index and Filter Struct and Map Columns

    Databricks AI Search now supports indexing and filtering struct and map columns. This enhancement allows users to leverage these complex data types directly within their AI Search indexes, improving the granularity and comprehensiveness of searchable data.

  • Execute Filter-Only Queries

    Users can now run filter-only queries in AI Search, which return rows that match specified filter criteria without requiring vector or keyword retrieval. This provides a more efficient way to retrieve specific data subsets when the search intent is purely based on attribute filtering.

Read the original announcement →

https://docs.databricks.com/aws/en/release-notes/product/2026/september#ai-search-supports-struct-and-map-columns-and-filter-only-queries

Related releases