Clairvoy

Inspiration

Buildings account for nearly one-third of global energy use and emissions, making them a crucial target for climate action. With climate change intensifying, there is an urgent need to optimize energy consumption without sacrificing comfort. Our team was motivated by this challenge: what if we could forecast and prevent waste before it happens? This question drove us to design an AI-powered short-term load forecasting system aimed at enabling smarter building management, renewable integration, and real-time anomaly detection.

What We’re Building

Our project is being developed in phases to balance feasibility with long-term impact.

Phase 1 (Proof of Concept): Due to early deadlines and limited resources, we worked with our own dataset and implemented a baseline forecasting pipeline using a Linear Regression model. To make results accessible, we deployed it with a simple Flask web application and used season- and time-based feature engineering to capture basic temporal patterns. This phase demonstrates the viability of the approach and provides a starting benchmark.

Phase 2 (Scalable System): Building on the proof-of-concept, we will expand to a robust pipeline leveraging boilerplate code and richer datasets provided by authorities. Here, we plan to employ Time Series Foundation Models (TSFMs), which generalize across temporal domains and can deliver higher accuracy without retraining from scratch.

What We Learned

Phase 1 helped us understand the importance of rigorous preprocessing (cleaning, resampling, normalization) and the role of engineered features in improving baseline forecasts. Deploying the model via Flask also revealed the value of accessible interfaces for stakeholders.

Challenges

The key challenges remain handling sparse and noisy datasets, aligning forecasts with renewable availability, and ensuring interpretability for building managers.

Vision

By progressing from a proof-of-concept baseline to advanced TSFM-powered forecasting, we aim to deliver a practical yet scalable system that reduces energy waste and accelerates the transition toward net-zero buildings.