GreenAI: Smart AI-Based Electricity Bill Analyzer and Energy Saver

electricity usage prediction; SGDRegressor; FastAPI; React.js; Green Score; energy analytics; carbon footprint; machine learning; incremental learning

 Electricity users in India, households and businesses alike regularly receive bills higher than expected, mainly because they have no way to track their usage in real time. This paper presents GreenAI, a publicly deployed web app that takes electricity meter readings from a CSV file, runs them through a machine learning model, and returns four useful outputs: a predicted next reading, an estimated monthly bill in INR, a CO₂ emissions figure, and a Green Score that communicates energy efficiency in a way anyone can understand. The backend is built in Python using FastAPI to handle requests and SGDRegressor from scikit-learn to learn from new data over time, deployed on Render. The frontend is a React.js web app built with Vite, styled with Tailwind CSS, and deployed on Vercel. Validation against two real-world datasets - a household dataset and a large industrial or factory-scale dataset, shows how the system handles very different types of energy consumers: a household consumer achieved a Green Score of 91 (Excellent Efficiency), whereas the industrial dataset yielded a Green Score of 51 (Needs Improvement), an average daily consumption of 216.77 kWh, and an AI-forecast next reading of 318.21 kWh. The results confirm that ML-assisted energy monitoring can be made accessible across consumer categories without any data science background.