This paper presents a systems theoretic approach to develop a solver component for a management information system (MIS). In this approach, a model is described in the set theoretic terms based on systems concepts and is implemented in the extProlog, an extension of the Prolog. Yasuhiko Takahara is a Professor Emeritus of the Tokyo Institute of Technology and a Professor in the Department of Management and Information Sciences, Chiba Institute of Technology, Chiba, Japan. He is pursuing research in applications of the mathematical general systems theory to the organization theory and to MIS. In particular, an intelligent DSS, end user development and intelligent data mining are his current topics in the MIS area. In 2003, he has had two monographs, “Organizational Structure: Cybernetic Foundation” and “Advances of DSS Development-General Systems Theory Approach”, published by Kluwer Academic Publishers, U.S.A., and Economics and Sciences Publisher, China, respectively. Although the approach adopts the set theory for representation and uses a logic programming language for this implementation, it is completely different from the conventional constraint programming or the rule-based optimization, which starts from a given problem and tries to find a solution algorithm (Cantone, Omodeo and Policriti, 2001; Hooker, 2002). Our approach starts from the investigation of a general problem-solving system. The solving system is the main subject. The extProlog is not used as a declarative language but as an easy implementation tool for a set theoretic description. We propose a design and implementation procedure on an abstract level using the goal-seeking model of the general systems theory. When it is applied to a real problem, it must be made concrete to meet the properties of the specific problem. A problem classification scheme is introduced to meet the requirement. According to the classification there are eight classes. The development procedure is formalized for each specific class. This paper investigates a class called E-C-C, which is the most structured, to show the formulation scheme. According to the formulation, a solver is divided into two independent components: the user model and the goal seeker. This paper adopts a modified dynamic programming (DP) method for designing the goal seeker and investigates its formal properties. Implementation is done based on the investigation. The traveling salesman problem is used to illustrate the entire procedure.