GadoAI is an AI assistant for managing operations and KPIs in a corporate system. Users can ask questions in natural language about operations, costs, and KPIs. The system uses LangChain + Groq LLM to interpret queries, execute SQL against the database, and return precise, data-driven responses.
Originally focused on livestock management, the project has been adapted to a broader corporate operations and KPIs theme, keeping all AI query and analysis logic intact.
- KPI Queries: Total operations, total cost, average cost, completed, failed, and in-progress operations.
- Natural Language Chat: Ask short or medium-length questions about system data.
- Off-topic Filtering: The AI responds only to questions relevant to operations/KPIs.
- Interactive Frontend: Built with React + TailwindCSS for easy interaction.
- Consistency Verification: Answers match real database data.
backend/
│
├── app/
│ ├── core/ # Configuration and logging
│ ├── db/ # Database scripts and connection
│ ├── services/ # SQL & LLM services
│ ├── api/ # FastAPI endpoints
│ ├── prompts/ # LLM prompts and examples
│ └── utils/ # Helper functions
│
├── scripts/ # Seed and auxiliary scripts
├── tests/ # Unit tests
├── .env # Environment variables
├── requirements.txt # Python dependencies
└── README.md # Project documentation
- Python 3.12+
- FastAPI → backend and API
- PostgreSQL / Supabase → database
- LangChain + Groq → language model for SQL queries
- React / TailwindCSS → frontend chat interface
Clone the repository:
git clone <REPO_URL>
cd backend
Install Python dependencies:
pip install -r requirements.txt
Configure environment variables in .env:
SUPABASE_DB_URI=<Your database URI>
GROQ_API_KEY=<Your Groq API key>
Start the API:
uvicorn main:app --reload --host 0.0.0.0 --port 8000
Open the React frontend and start chatting with the AI assistant.
"What is the total cost of operations?" "How many operations have been completed?" "How many operations are in progress?" "What is the average cost per operation?"
Contributions are welcome!
If you want to test or improve the AI responses, feel free to open issues or submit pull requests.
This project is licensed under the MIT License.