The idea for this project comes from seeing how much energy buildings waste every day, and realizing that most of it is preventable if we could just “see ahead.” A school leaving the AC running all weekend, an office with a faulty chiller eating electricity, or a new apartment block with no energy history but huge demand peaks—all of these problems add up to emissions on a global scale.
At the same time, in other fields like language and images, foundation models have completely changed what’s possible. They don’t need to be retrained for every new problem—they already “know” the patterns. That sparked the thought: what if energy could be treated like language? It has rhythms, it has seasonality, it even has its own “grammar” of peaks and troughs. If we had a foundation model for energy, every building could have an instant forecasting brain.
And that’s where the “Energy Twin” comes in. Inspired by the concept of digital twins in engineering, we imagined each building having its own intelligent twin: always watching, always learning, predicting tomorrow’s demand and spotting when something feels off. Instead of treating buildings as silent consumers, they become active, self-aware players in the energy system.
The inspiration is simple: use the same kind of AI that can write a poem or predict the next word to instead help buildings cut waste, save money, and lower emissions. It’s a way to scale decarbonization fast, without waiting years for custom models.
Log in or sign up for Devpost to join the conversation.