EnergyCanvas forecasts electricity demand a day ahead for every grid in the United States — and measures itself against the people who run those grids.
What it is
A forecasting platform covering all seven US grid operators system-wide, plus dozens of their regional zones. Every forecast is hourly and day-ahead, with confidence bands and live accuracy tracking.
What sets it apart is the yardstick: each forecast is scored against the day-ahead forecast the grid operator publishes for itself — a real, public benchmark rather than a number I picked. There's also a build-your-own mode: point it at any region with a history of demand and it produces a tailored forecast, even somewhere no official forecast exists.
Why I built it
Grid operators run serious forecasting operations — billions of dollars of power dispatch and trading ride on getting tomorrow's demand right. Building something that competes with them, market by market, was a problem worth taking on. I wanted to own the whole stack end to end: the data, the models, the evaluation, and a clean interface worth using every day. It's built for the people who live by the load number — trading desks and utilities alike.
What's next
A deeper write-up is on the way. On the roadmap: broader team access, a public API and an AI/MCP connector to pull forecasts straight into your own tools, and longer multi-day horizons.
Visit the live product → · or reach out to learn more — happy to walk you through it.