OCI Generative AI Agents platform released March 26, 2025

Ready-to-use SQL Tool

We’re excited to introduce the pre-built SQL tool as part of the Agents platform experience. This tool revolutionizes how users interact with structured enterprise data by enabling data retrieval and analysis through natural language, unlocking valuable insights with minimal effort.

Key Features:

• SQL Generation: Automatically generate SELECT statements based on natural language input.
• SQL Explanation: Provide intuitive explanations for SQL statements to help users understand syntax and logic.
• Self-correction: Automatically detect and correct syntax errors in generated SQL queries.
• SQL Execution: Execute SELECT statements and return key results and insights.
• Schema Linking: Easily provide your database schema to the agent.
• In-Context Learning Examples: Provide examples while configuring the SQL tool to help improve the accuracy of the responses.
• Custom Instructions: Adjust the behavior and tone of the SQL tool based on business needs.
• Table/Column Descriptions: Add descriptions to enrich the schema context and help improve response quality.
• Model Selection: Select between smaller or larger models to optimize for latency and accuracy.
• Multi-dialect Support: Fine-tuned in Oracle SQL and compatible with several SQL dialects including SQLite.

Ready-to-use RAG (Retrieval-Augmented Generation) Tool

Today, we're releasing key enhancements to the RAG tool with improved functionality based on customer feedback.

Key Features:

• Enhanced Response Quality: Improve accuracy by refining contextual understanding and relevance of generated answers.
• Hybrid Search: Combine keyword and vector search for highly precise, efficient retrieval.
• Improved Multi-Modal Parsing: Extract insights from rich content formats such as images and charts.
• Custom Instructions: Define preferences to fine-tune behavior and response styles of the RAG tool.
• Multi-Lingual Support: Seamlessly interact across several languages, including French, Spanish, Portuguese, Arabic, German, Italian, and Japanese.
• Multiple Knowledge Bases: To help improve coverage and depth of knowledge, set up the agents to retrieve information from several knowledge bases.
• Metadata Ingestion & Filtering: Use metadata to refine searches and categorize content more effectively.
• Cross-Region Access: Enable vector data retrieval across regions using Oracle Vector Search (Database 23ai).
• Paragraph-Level Citations: Improve transparency with clear source attributions for each response.

Custom Function Calling Tool

We now offer a custom user-defined function calling tool to expand the actions that an agent covers. Define the function and then in natural language, ask the agent to perform an action. The agent can run the related function and respond based on what the function does.

Other Enhancements

• Seamless Agent Setup and Deployment: Fully managed, cloud native platform for efficient and scalable agent setup and deployment. Supports modern workflows with flexibility for customization.
• Tools Orchestration: Orchestrate several tools and services to build complex workflows and automate conversations.
• Multi-Turn Chat Experience: Supports dynamic, multi-turn dialogues for more human-like interactions.
• Context Retention: Agents remember conversation context across turns for personalized, consistent experiences.
• Guardrails: Have the agent help identify and apply content moderation, and help identify and protect against prompt injection (PI) and personally identifiable information (PII) at the agent endpoints.
• Human-in-the-Loop: Optional feature for real-time monitoring and human intervention.
• Scalability and Security: Get enterprise-grade infrastructure built on OCI with strong security, reliability, and compliance.

For information about the service, see the Generative AI Agents documentation.