What is the AI-Powered SQL Assistant?

It's a system enabling natural language querying of Oracle Autonomous Databases (ADBS) by generating SQL and performing native vector searches, all powered by a local LLM.

What role does the LLM play in this assistant?

A fully local LLM (e.g., LLaMA 3.2 via Ollama) interprets user queries, generates the appropriate SQL (traditional or vector-based), and can optionally interpret SQL results into natural language answers.

Does it require an external vector database?

No. This tool leverages Oracle ADBS's native vector embedding and querying capabilities, eliminating the need for any external vector database.

How does it handle different types of queries?

The system intelligently switches between generating structured SQL and performing semantic vector searches based on your natural language query, ensuring optimal data retrieval.

Is the system designed for extensibility?

Yes, its modular architecture facilitates easy expansion and integration, allowing for future additions like UI development, confidence scoring, or multi-modal inputs.

AI-Powered SQL Assistant

Name: AI-Powered SQL Assistant
Author: sadanandu

0•

Enables natural language querying of Oracle Autonomous Databases by generating SQL and performing vector searches.

The AI-Powered SQL Assistant is a proof-of-concept system designed to bridge the gap between natural language and database queries, specifically for Oracle Autonomous Database (23ai). It allows users to ask questions in plain English, which are then intelligently translated into SQL or semantic vector searches. The system leverages a locally running Large Language Model (LLaMA 3.2 via Ollama) for query interpretation and SQL generation, a FastAPI-based backend (MCP Server) for execution on Oracle ADBS, and native Oracle Vector Search capabilities, eliminating the need for external vector databases. Its modular architecture ensures flexible query routing and an optional LLM-driven interpretation of results for more human-readable answers.

주요 기능

01Optional LLM-based interpretation of SQL results for natural language answers.

02Native Oracle ADBS vector embedding and querying, no external vector database required.

03Fully local LLM for intelligent SQL generation via Ollama.

040 GitHub stars

05Modular architecture facilitating easy expansion and integration.

06Smart switching between structured SQL and semantic vector search.

사용 사례

01Natural language querying of Oracle 23ai databases.

02Performing semantic searches on unstructured data stored within Oracle Autonomous Database.

03Generating SQL statements from plain English descriptions.