Writing.
Technical notes on how Amanoki is built, the calibration experiments that have shipped and the ones that were scrapped, and the operating choices behind a self-serve wholesale-power API. No product announcements — those live in the changelog (RSS).
-
Posts will appear here as they're written. The changelog records what shipped; this page is for the longer-form reasoning behind the choices — why ESN v2 was scrapped, what
confidence_caveatis measuring, how weather conditioning moves Brier scores, how JEPX scarcity events are structured. If the methodology page is the compressed reference, this is the running notebook.