Databricks says AI Search can now index streaming tables, materialized views and managed Iceberg v3+ tables directly

Databricks said in a blog post dated Oct. 26, 2026 that its AI Search product can now index new table types. They are streaming tables, materialized views and managed Iceberg v3+ tables. Delta tables were already supported, and the new types sit alongside them. The product documentation lists the same source types. For most teams, the practical question is which of their tables now qualify.

Key facts, per Databricks and its documentation

• New sources: Streaming Tables, Materialized Views and managed Iceberg tables using Iceberg v3+.
• Delta tables: Still supported, alongside the new types.
• Old workflow: A separate Delta copy plus a pipeline to keep it synchronized with the source.
• New workflow: AI Search connects directly and propagates changes to the index.
• Materialized views: External metadata, also called Smart Clone, must be enabled first.

What changed for teams with a Delta copy

Databricks says that before native support, customers had to create a separate Delta copy of the new table types. They also had to build pipelines to keep that copy synchronized with the source. Now AI Search connects directly to these sources. It propagates changes directly to the index, with no intermediate Delta table. The custom synchronization pipeline is eliminated as well.

How each new source reaches the index

Databricks says new records can flow into an append-only streaming table. AI Search indexes them as they arrive, which keeps the search index up to date. Its listed uses are fraud detection, security monitoring and real-time Internet of Things (IoT) analytics.

Materialized views update through the view

Databricks says that when the data under a materialized view changes, the view is incrementally maintained. AI Search then propagates those changes to the index. In the new setup, those changes reach the index without the intermediate Delta table that the old workflow used.

Iceberg tables at v3 or above

Databricks says managed Iceberg tables at v3 or above can serve as AI Search sources directly. That needs no conversion or migration to another table format. Teams with such tables can skip that conversion step. The sources name v3 or above as the condition, so older Iceberg versions fall outside these claims.

Building the index

Databricks says building an index for the new sources follows the same workflow as for Delta table sources. The AI Search documentation says an index is created from a source table of a supported type. For a materialized view, the docs require external metadata to be enabled first. External metadata, also called Smart Clone, must be enabled on the view or on the pipeline that defines it. A decision check below shows which setup step applies to a table type you choose.

Analysis: what to check first

A team that keeps a Delta copy should first check whether its data is one of the source types above. If it is, the copy and the sync job are the parts Databricks says the new setup removes. A copy may also serve other readers, so a team should check those before removing it. As a hypothetical, a team whose source is a streaming table would not need the copy step at all. A team planning a materialized view source should check the Smart Clone setting before starting the build.

What the sources do not show

Every capability claim here comes from Databricks, and none is independently verified. The blog post gives no benchmarks, pricing, or regional availability. The documentation carries no date, so it does not show when the support began.

Where to read next

For setup steps, the AI Search documentation is the next place to read. It sets out which table types qualify and how an index is created from a source table. Anyone planning a move should match their own table type against that list before changing a pipeline.

Sources: Building live intelligence: Databricks AI Search supports Streaming Tables… (databricks.com), accessed Oct 26, 2026 · Databricks AI Search | Databricks on AWS (docs.databricks.com), accessed Oct 26, 2026

ZIPMEX promotes trading on trade.zipmex.com. Promotions are labelled and kept separate from news coverage, which is selected and checked without regard to them. Perpetual futures are leveraged derivatives. Prices can move fast and you can lose all of your margin. Not available in every jurisdiction. Not investment advice.