This article proposes a decision support system project to find the optimal milling cutter design. At the preliminary design stage, morphological analysis is used. It allows us to find and systematize all possible milling cutter structures with the necessary functional purpose. To automate the design formation process, an algorithm based on a resolution method is applied, with the algorithm using the logic of first-order predicates. An enumeration of possible states and assembly of a milling cutter structure from ready-made elements are carried out. It is expedient to describe this algorithm in terms of logical operations. It consists in a deductive derivation of the sentence of the form: "There are dimensional parameters x1,...,xn and forces such that a constructive solution satisfying the given properties P (x1...,xn) is deduced from a set of possible connections x1... xn (K1(x1,...,xn)... Kn(x1,...,xn)  В(x1,...,xn)), which denote dimensional, force and other real connections that arise between the parts of real structures." The condition for the transition from the specific parts to the logic of first-order predicates is the capability of a certain part (for example, plate P) to function in real conditions if and only if there is a set of fixing forces f1, f2, ..., fn that are applied at points x1,x2,...,хn. The basic effectiveness parameters of milling cutter design are: reliability, productivity, and energy efficiency that are set as objective functions. They also take into account the static and dynamic design characteristics. The variable parameters are the geometric shape and dimensional parameters of a milling cutter. For each variant of the geometric form, a 3D model of the milling cutter is constructed and its static and dynamic characteristics are calculated. These parameters are then included in the objective functions. Optimization is carried out on the basis of the gradient descent method. The optimal design is chosen with the interaction of intelligent agents. In this case, the milling cutter design provides the best ratio of the objective functions. The architecture of the system is based on the integration of CAD/CAE systems with a multi-agent system (MAS). The search for a solution is carried out automatically as a result of the interaction of independent task-oriented software agents. To build MAS, we use the Java Jade library in the NetBean development environment. The considered approach allows us to reduce the time expenditures in designing or choosing the design of a metal cutting tool.