This runs the same Model-Predictive-Control algorithm shipped in the accelerator (ported to JavaScript). Adjust the price curve and comfort band, then optimize: the controller pre-heats when energy is cheap and coasts when it's expensive — never leaving the comfort band. This is a marketing demo; the product ships the unit-tested NumPy solver, thermal-model training, and savings SQL.
The complete Predictive Heating Optimization Accelerator package includes: