Databricks: 100,000 Agents Now Built on Its Platform

Databricks reported that more than 100,000 AI agents have now been built on its platform, alongside taking Document Intelligence to general availability — with ai_parse_document, ai_extract, and ai_classify now available as plain SQL functions, not specialised ML pipelines.

Why 'It's Just SQL Now' Is the Real Story

Document extraction and classification used to require a dedicated ML pipeline, a data science team, and meaningful lead time to stand up. Making it a callable SQL function means any analyst who can already write a query can now extract structured data from unstructured documents directly inside their existing workflow — a genuine democratisation of a capability that used to require specialist skills.

100,000 Agents Is a Scale Signal, Not Just a Vanity Number

• That volume indicates agent-building has moved well past early-adopter experimentation into routine, repeated practice across Databricks' customer base
• It also means Databricks now has meaningful production experience data on what agent patterns actually work at scale — a real competitive asset going forward
• For anyone evaluating a data platform, this is a genuine signal about ecosystem maturity, not just a headline metric

What to Do With SQL-Native Document Intelligence

If your organisation processes unstructured documents (invoices, contracts, support tickets, compliance filings) and already has SQL-literate analysts, this closes a real gap between 'we have the data' and 'we can query it' without requiring a new specialist hire or a separate tooling investment — worth a pilot on your highest-volume unstructured document type.