hylosense

hylosense — Per-asset AI weather models for energy: one site-calibrated model per generation asset, trained on that site's terrain, elevation, land cover, local microclimate and its own SCADA observations, delivering day-ahead and intra-day asset-level forecasts by API. Latent byproduct assets include the paired site-weather-to-SCADA-generation training corpora and the realized forecast-vs-actual error records per asset.. Very early but technically differentiated: 7-person Seed-stage firm (~$470K raised, Skopje) with named utility pilots and quantified accuracy claims (up to 60% better day-ah…

hylosense offers (Alternative, Geospatial, Price, Fundamental) — est. per-asset panels in the millions of rows: each asset yields hourly day-ahead plus intra-day forecast vectors across many meteorological variables (say 8,760 day-ahead cycles x 10-30 variables + finer intra-day refresh), so 20-100 assets in pilots implies 5-50M forecast rows to date; the training corpora are larger still since they include multi-year on-site observation and SCADA series per site..

Day-ahead and intra-day renewable generation forecasting; imbalance-cost reduction for utilities and trading desks; price and intraday position forecasting; grid capacity/congestion and losses forecasting; storage charge/dispatch optimisation; renewable site selection and yield assessment using climate projections under GHG scenarios; training data for weather-to-generation models and for reinforcement-learning bidding agents; benchmark corpora of per-asset forecast error by technology and region.

The data is with 3 years of history.

Coverage spans US, APAC, Europe, Japan; Application Software, Electric Utilities; alternative, geospatial, price, fundamental; commodities, derivatives, equities.