← All updates Databricks Update One-Pager Navy version
databricks Platform · Updated Jun 11, 2026

ai_classify

Now generally available for text classification.

GA Open docs ↗release note ↗
What it does

The `ai_classify` function enables users to classify text content into predefined categories based on custom labels, supporting features like multi-label classification and label descriptions. It is designed to work with unstructured data stored in Unity Catalog, allowing for efficient document classification. Users can create classification agents through a no-code UI to streamline the classification process and improve accuracy over time.

Why it matters

This release enhances productivity by providing a robust tool for automating text classification, which can lead to improved decision-making and operational efficiency. Businesses can leverage this capability to better understand customer interactions, categorize documents, and analyze sentiment, ultimately driving better customer experiences and insights.

Key capabilities
Multi-Label Classification
Classify documents with multiple labels to capture complex categorization needs.
Custom Label Descriptions
Define and refine classification labels with optional descriptions for clarity.
No-Code UI for Agents
Utilize a user-friendly interface to create and manage classification agents without coding.
Integration with SQL and Spark
Run classification at scale using SQL queries or create ETL pipelines for automated processing.
Prerequisites
Serverless compute enabled in the workspace.
Unity Catalog must be enabled.
Access to a serverless usage policy with a nonzero budget.
Unstructured data must be in a Unity Catalog volume or table.
At least one unlabeled document or row in the data source.
Recommended use cases
Classifying customer call transcripts by intent.Categorizing documents by content type.Analyzing product reviews for sentiment.Organizing unstructured data for better retrieval.
Limitations · remember
!Function availability is limited to specific regions.
!Requires compliance with regional support for Enhanced Security and Compliance add-on.
Architecture
Data Ingestion → Unity Catalog → ai_classify Function → Classification Agent → SQL Queries or Spark Pipelines
Get started
1
Select Source Data
Choose a volume or table containing the documents to classify.
2
Configure Labels
Add and refine classification labels manually or import from a table.
3
Review Classification Responses
Provide feedback on classification accuracy to improve the agent.
4
Use the Classification Agent
Deploy the agent for large-scale classification using SQL or Spark pipelines.
5
Iterate and Optimize
Compare versions of the agent to enhance performance based on feedback.
Key takeaway
The `ai_classify` function empowers organizations to automate and enhance text classification, driving better insights and operational efficiency.
2026-06-11-ai-classify-is-now-generally-available Generated by the Databricks Updates Agent · browse all updates