Select AI: Oracle Adds the “Summarize” Action in Autonomous Database

 

Introduction

Oracle continues to strengthen its Select AI capabilities in Autonomous Database (ADB).
The September 2025 update introduces a brand-new AI action — summarize — that enables you to generate concise summaries directly from your data or query results using plain English prompts.

Instead of running SQL, exporting data, and using an external LLM or BI tool to interpret it, you can now ask Oracle’s Select AI to summarize data inside the database — powered by Oracle’s integration with Generative AI providers like OCI AI, OpenAI, and Azure.

What’s New in This Release

According to Oracle’s release notes:

“The new summarize action in Select AI enables users to create concise summaries of query results or text-based data using natural language prompts.”

This makes Select AI even more powerful, expanding beyond existing actions like runsql, showsql, explainsql, and narrate.

Example 1: Using SELECT AI to Summarize Sales Data

Let’s say you have a SALES table with the following columns:

ORDER_ID	REGION	PRODUCT_CATEGORY	SALES_AMOUNT	QUARTER
101	East	Electronics	25000	Q2
102	West	Apparel	14000	Q2
103	East	Furniture	18000	Q2
104	North	Electronics	32000	Q2
105	West	Apparel	13000	Q2

Now you want a quick summary of sales trends in Q2 2025, instead of writing SQL and manually analyzing it.

Here’s how you can do it using Select AI:

SELECT * FROM AI (
  action => 'summarize',
  prompt => 'Summarize the sales trends for Q2 2025 by product category and region. Highlight top-performing categories.'
);

AI Output:

“In Q2 2025, Electronics dominated sales with strong performance in the North and East regions. Apparel showed moderate results in the West, while Furniture had steady growth in the East. Overall, the quarter reflected a 10–12% increase compared to Q1, driven primarily by Electronics demand.”

No complex SQL required — just one natural-language prompt.

Example 2: Summarizing Textual Data (Customer Feedback)

Suppose you store customer feedback in a CUSTOMER_FEEDBACK table:

CUSTOMER_ID	FEEDBACK_TEXT
1001	"Love the new product features but wish delivery was faster."
1002	"Customer support was excellent, but website checkout was confusing."
1003	"Fast delivery and helpful service, great overall!"
1004	"Product quality good, but packaging needs improvement."

You can summarize overall feedback using:

SELECT * FROM AI (
  action => 'summarize',
  prompt => 'Summarize overall customer sentiment based on feedback records.'
);

AI Output:

“Customers generally appreciate product quality and service responsiveness. Key areas of improvement include delivery speed, packaging, and the checkout process.”

Example 3: Using PL/SQL Procedure DBMS_CLOUD_AI.SUMMARIZE

If you prefer procedural execution (e.g., inside a job or scheduled process), you can use:

DECLARE
  v_summary CLOB;
BEGIN
  v_summary := DBMS_CLOUD_AI.SUMMARIZE(
    profile_name => 'MY_AI_PROFILE',
    prompt => 'Summarize Q2 2025 sales by region, highlight top growth areas.'
  );

  DBMS_OUTPUT.PUT_LINE(v_summary);
END;
/

AI Output Example:

“Sales grew most significantly in the East region (15% YoY), driven by Electronics and Furniture categories. Apparel sales remained stable, while North region saw a small decline.”

This approach is ideal for embedding summarization directly in reports, alerts, or dashboards.

Why This Matters

The summarize action helps business users, analysts, and DBAs quickly understand what’s happening in their data — without exporting it to other tools.

Key Benefits:

Simplifies reporting — get summaries without writing SQL joins or aggregations
Reduces time-to-insight — natural language replaces manual analysis
Enhances collaboration — share summaries directly from the database
Fits into automation — integrate summarization in daily or weekly jobs

Best Practices

Tip	Description
Be specific	Instead of “Summarize sales,” say “Summarize top 3 products by sales growth this quarter.”
Use filters	Narrow the dataset before summarizing (e.g., WHERE clause).
Validate facts	Cross-check AI summaries if numeric accuracy is critical.
Automate	Combine with DBMS_SCHEDULER to auto-generate daily summaries.
Conversational refinement	Use chat action to refine summaries in context.

Conclusion

The new Select AI Summarize feature represents a major step in Oracle’s vision of AI-driven data management.
By integrating summarization directly into the Autonomous Database, Oracle empowers users to move from data retrieval to insight generation — in seconds.

Whether you’re generating executive reports, summarizing customer feedback, or monitoring system metrics, this feature makes it simpler, faster, and smarter.

I’ll be recording a short demo on Select AI soon and will share it at the earliest opportunity.

Author: Narasimharao Karanam