The complexity and cost of design, as well as the quality of its results, are determined by the volume and depth of engineering knowledge embedded in a computer. The intellectual technology of computerization of engineering activity enables specialists who does not possess deep knowledge of computer science to create specialized workstations for themselves and their colleagues. At the same time, the engineering activity undergoes qualitative changes: the specialist enters technical specifications into a computer and oversees the project generation process, making fundamental creative decisions by selecting one of the options offered by the computer. Such systems with good reason can be attributed to a fundamentally new category of semi-automatic design systems. This paper examines the technology of creating semi-automatic systems for designing cylindrical gear reducers with an option of searching for the best solutions using genetic algorithms. The relevance of this topic is related to the fact that creation of intelligent systems for product design is an important direction in improving the engineering development of machine-building production. Such systems can increase the productivity and quality of designers’ work due to the semi-automatic generation of 3D models of products in typical variant design.