C1G8_Snow White & 4 Dwarf

Team Member Name

Group member: Cheah Jing Yik Chuah Shin Yee (Leader) Kendrew Lim Yan Zhe Khor Jie Shen Liew Kaiy Bin

Project Info

JIMAT-AI is an intelligent household energy assistant designed to help users better understand and manage their electricity consumption. Using machine learning, the prototype forecasts monthly electricity costs based on household characteristics such as peak hour usage, number of occupants, and energy habits. This allows users to identify patterns of high energy use and make informed decisions to reduce their bills.

Beyond its current predictive engine, JIMAT-AI is built with a future vision: empowering households to transition from passive consumers to active participants in a sustainable energy ecosystem. The roadmap includes features such as appliance-level energy insights and a community-based renewable energy sharing system, enabling neighbors to distribute and trade excess solar energy within local microgrids.

By transforming complex energy data into simple, actionable insights, JIMAT-AI supports SDG 7 by promoting affordable, clean, and efficient energy usage for smarter, greener cities.

SDG Alignment (Goal 7: Affordable and Clean Energy) Target 7.3 (Efficiency): We double the rate of improvement by targeting the primary cost driver—Air Conditioning. Our AI detects AC usage and recommends specific settings (24°C) to achieve a measured 20% bill reduction.

Target 7.1 (Affordability): We ensure access to modern energy services by demystifying complex tariffs. Our backend accurately models the RP4 Tiered Tariff, providing financial transparency.

Target 7.2 (Renewable Energy): We facilitate the green transition by quantifying the financial ROI of solar panels, turning renewable energy into a clear, profitable asset for homeowners.

We engineered Jimat AI using a decoupled full-stack architecture to ensure both speed and scalability. On the frontend, we leveraged Next.js 14 combined with TypeScript to build a robust, type-safe user interface, utilizing Tailwind CSS and Shadcn UI for a responsive design.

For the backend, we chose a lightweight Python Flask server hosted on Render. We utilized Pandas to process data before feeding it into our core innovation: a Hybrid AI Architecture. This system combines a Scikit-learn model for numerical forecasting with Google’s Gemini 2.5 Flash for computer vision and natural language processing. Finally, we integrated physics-based heuristics directly into our logic layer to ensure every financial recommendation is mathematically verified.

Project Idea

The core idea behind Jimat AI was born from the realization that while cities are the world’s largest energy consumers, the average household operates in a complete "data black box." Most residents receive a monthly electricity bill that tells them how much they owe, but provides zero insight into why the cost is so high or how specific daily habits—like leaving an air conditioner on at 20°C—directly impact that final number. We identified that this lack of real-time visibility and financial literacy is the primary barrier preventing households from improving their energy efficiency and transitioning to renewable sources.

We envisioned a solution that acts not just as a tracker, but as a proactive, intelligent energy auditor for every home. Jimat AI was conceptualized to bridge the gap between complex utility data and actionable behavioral change. By combining predictive machine learning to forecast future consumption with generative AI to interpret historical billing data, we aim to demystify the energy grid. The platform is designed to answer the two most critical questions for any homeowner: "Where am I wasting money right now?" and "Is installing solar panels actually a profitable investment for me?"

Ultimately, the project idea is to democratize energy management. We wanted to build a tool that removes the complexity of tiered tariffs and the financial uncertainty of green investments. By transforming a static, confusing PDF bill into a dynamic, interactive dashboard, Jimat AI empowers users to take control of their consumption, turning the abstract goal of "saving energy" into concrete, measurable financial wins that align directly with the global targets of SDG 7.