KPT's engine works by solving a real multivariate optimization problem over your operational variables. It finds either the maximum point of a productivity function or the minimum point of a cost function — over the joint surface defined by your SKU mix, line capacity, changeover settings, energy windows, and dozens more variables.
↓ Minimization examples: operating cost per ton, kWh per unit, broke at grade transitions, unplanned downtime hours.
↑ Maximization examples: production efficiency, on-time delivery, OEE, throughput per shift.
↔ Mixed objectives are supported — e.g. maximize OEE subject to changeover loss < X and energy cost < Y.
Every recommendation comes with the variables, weights, and constraints that drove it. Glass-box AI by design — no "trust us" Fuzzy rules.