Managing SQL Tools in Generative AI Agents

A SQL tool in Generative AI Agents converts natural language queries into Oracle SQL and SQLite statements and can run the SQL commands to generate responses against a connected database.

Only SQL SELECT queries including join, multiple condition, and aggregation are generated, and optionally executed.

To let a human reviewer monitor and edit a generated query before sending it to the user, select the optional feature Enable human in the loop in the agent endpoint.

Key Features

• SQL Generation: Generate SELECT statements based on your natural language input.
• SQL Explanation: Provide intuitive explanations for SQL statements to help you understand the syntax and logic.
• SQL Self Correction: Enable automatic detection and correction of syntax errors in the generated SQL queries.
• SQL Execution: Run the SELECT statements and return key results and insights.
• Schema Linking: Import your database schema to the agent.
• In-Context Learning Examples: Provide examples of questions and responses to improve the accuracy of responses.
• Custom Instructions: Adjust the behavior and tone of the responses based on your business needs.
• Table and Column Descriptions: Add descriptions to enrich the schema context and improve the quality of responses.
• Model Selection: Select between smaller or larger models to balance performance and accuracy.
• Multiple Dialect Support: Compatible with various SQL dialects, including SQLite and Oracle SQL.

Before You Begin

Before you create a SQL tool, ensure that you review the SQL Tool Guidelines for Generative AI Agents and complete any tasks that are required.

Manage SQL Tools

If you have manage permissions to genai-agent-family, you can manage the SQL tools. See the following links for managing each tool:

• Adding a SQL Tool to an Agent
• Listing the SQL Tools
• Getting a SQL Tool's Details
• Updating a SQL Tool
• Deleting a SQL Tool