QueryFusion AI ✨

The universal Text-to-Query (Text2SQL) interface. Connect to any database (SQL, NoSQL, Graph), use any LLM (OpenAI, Gemini, Claude, Llama), and generate DB queries from natural language.

QueryFusion AI bridges the gap between human language and machine queries. Stop wrestling with complex syntax and start talking to your data. This open-source project provides a sleek interface and a powerful backend to generate SQL, Cypher, MQL, and more, using your favorite local or cloud-based Large Language Models.

Screenshot

---

Key Features

• Universal Database Support: Built with a modular connector architecture.
  • SQL: PostgreSQL, MySQL, SQLite, DuckDB, ClickHouse, TimescaleDB.
  • NoSQL (WIP): MongoDB, Cassandra, HBase.
  • Graph (WIP): Neo4j.
• Universal LLM Integration: Powered by LiteLLM, supporting 100+ models out-of-the-box.
  • Cloud APIs: OpenAI (ChatGPT), Google (Gemini), Anthropic (Claude), Grok, etc.
  • Open Source & Local: Llama 3, Mistral, CodeGemma via Ollama, vLLM, and other local inference servers.
• Modern Tech Stack: Fast, scalable, and easy to contribute to.
  • Backend: FastAPI (Python)
  • Frontend: Vite + React + TypeScript + Tailwind CSS
• Intelligent Schema Detection: Automatically introspects your database to provide the LLM with the necessary context for accurate query generation.
• Developer Friendly: Easy setup, clean code, and designed for extension.

---

Architecture Overview

QueryFusion AI uses a decoupled frontend/backend architecture for maximum flexibility.

[User on Vite Frontend] --> [REST API call to FastAPI Backend] --> [LiteLLM Gateway] --> [Chosen LLM API (e.g., Ollama or OpenAI)]

The backend handles all the heavy lifting: database inspection, secure credential management, and prompt engineering.

---

🏁 Getting Started

Prerequisites

• Python 3.9+ and Node.js 18+
• Access to a database
• API keys for desired LLMs (e.g., OPENAI_API_KEY) or a local Ollama server.

Installation

1. Clone the repository:

   git clone [https://github.com/lokeshsk/queryfusion-ai.git](https://github.com/lokeshsk/queryfusion-ai.git)
   cd queryfusion-ai

2. Setup the Backend:

   cd backend
   python -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt
   cp .env.example .env
   # Edit .env and add your API keys

3. Setup the Frontend:

   cd ../frontend
   npm install

---

🚀 Part 2: Running & Testing the Full Application

Let's test it end-to-end. We'll set up a sample PostgreSQL database, run the backend and frontend, and generate a query.

1. Set Up a Sample Database

If you have PostgreSQL running, connect to it and run this SQL script to create some sample data.

-- Create a 'customers' table
CREATE TABLE customers (
    customer_id INT PRIMARY KEY,
    first_name VARCHAR(50),
    last_name VARCHAR(50),
    email VARCHAR(100) UNIQUE,
    registration_date DATE
);

-- Create an 'orders' table
CREATE TABLE orders (
    order_id INT PRIMARY KEY,
    customer_id INT REFERENCES customers(customer_id),
    order_date DATE,
    total_amount NUMERIC(10, 2)
);

-- Insert some sample data
INSERT INTO customers (customer_id, first_name, last_name, email, registration_date) VALUES
(1, 'Alice', 'Johnson', 'alice.j@email.com', '2023-01-15'),
(2, 'Bob', 'Smith', 'bob.s@email.com', '2023-02-20'),
(3, 'Charlie', 'Brown', 'charlie.b@email.com', '2023-03-10');

INSERT INTO orders (order_id, customer_id, order_date, total_amount) VALUES
(101, 1, '2024-05-10', 150.75),
(102, 2, '2024-05-12', 200.00),
(103, 1, '2024-05-15', 75.50),
(104, 3, '2024-05-18', 300.25),
(105, 1, '2024-05-20', 50.00);

2. Run the Backend

Navigate to the backend directory. Create and activate a virtual environment if you haven't already.

• Install dependencies: pip install -r requirements.txt
• Create your .env file and add your OpenAI API key.
• Run the server: uvicorn app.main:app --reload

Your backend is now live at http://127.0.0.1:8000.

3. Run the Frontend

Open a new terminal and navigate to the frontend directory.

• Install dependencies: npm install
• Run the dev server: npm run dev

Your UI is now live at http://localhost:5173.

4. Test a Sample Query

1. Open http://localhost:5173 in your browser.
2. In the Database Connection section, fill in the details for your PostgreSQL database.
3. In the Ask a Question section, type in a question like:

   What are the names and emails of customers who have spent more than $100?

4. Select your desired LLM model from the dropdown (e.g., "OpenAI GPT-4o").
5. Click the ✨ Generate Query button.

---

Contributing

Contributions are welcome! Whether it's adding a new database connector, improving the UI, or enhancing the prompt engineering, please feel free to open a Pull Request.

---

License

This project is licensed under the MIT License. See the LICENSE file for details.