BASED ON THE LATEST AI TECHNOLOGIES
↓
Built from extensive research using custom-made generative solar models, Google MetNet and optimised data pipelining, we have developed a neural net which takes feeds from numerical weather prediction, satellite imagery, real-time solar generation and geometry
Our team draws on a wide array of industry experience from Google Deepmind, NASA, scale.ai, wind forecasting and transmission system operators.
Satellite imagery provides over 40% of the accuracy delivered by earth observation.
-
Data feeds from atmospheric weather models and 5-minutely, geostationary satellite imagery
-
Azimuth and elevation of the sun based on location, day of year and time of day
-
Historical and real-time PV generation data
Accuracy ratios calculated by ablation study
All our forecasts are available through a fully documented API.
Easy to use
Fully documented
Fast and reliable
Easy to use UI
Actuals vs Forecast
Drill into regions
Inspect results on the fly
Open Source
We’re a group of creative coders who have built a business to change the world.
Why open?
Maximise transparency: No black-box models
Collaboration: Rapid development with contributors from across the world
Constant evolution: Avoid legacy technology lock-in
10s of terabytes of weather and satellite data
Multiple models with over 100 contributors
