This paper presents a novel approach to the optimization of a system using soft computing. In many instances optimum solutions to complex problems can be calculated using analytic or numeric approaches, but with significant barriers for practical use, such as computational complexity, sensitivity to parameter variations, and necessity to use many measured variables. In this study, a traditional optimization method is used to calculate off-line optimum solutions to an indoor environmental control problem at numerous operating points. These solutions are, in turn, used as exemplars to train an intelligent system such as a fuzzy logic system or a neural network, resulting in a control system whose behavior exhibits the desirable features of the family of optimum solutions. It has been shown that this methodology (named Modeled Optimized System - MOS) results in a controller for an indoor environmental system, which improves occupant satisfaction, saves energy, and can be implemented in a practical fashion.