||The broader context of this presentation is meant to be a preparation of a possible project at SCCH. It is suggested a concept of an automated decision making process with an inter-face to a back-office system. First of all we introduce the architecture of such a tool. Secondly we describe the mathematical methods of the single components. In principal, this paper is concerned with uncertain reasoning for the knowledge representation, and a brief introduction into decision making. Also we present a new method of machine learning, such that the described tool is able to adjust to its environment.