The aim was to create a mathematical model describing the development of a production (shop-to-shop) routing of mechanical engineering products based on a 3D model and allowing the cost of the final product to be reduced. The developed mathematical model was simulated based on 3D models designed in the Siemens NX system, which were subsequently imported into the *stp format and recognized by a designed module written in the Phyton programming language. The factors of the production environment affecting the formation of the production routing of mechanical engineering products were determined. A diagram of the algorithm for the “constructive element - technological operation - means of technological equipment (equipment-tool)” relationship was developed. Based on the results of testing the developed mathematical model, the use of neural networks as a tool for the implementation and automation of the work was found advantageous as compared to the standard scheme of work of a process engineer when developing a production routing of mechanical engineering products. These advantages include a decrease in the time for the development of a routing and the cost of the final product. The developed model has a practical limitation consisting in a rather complex geometry of some structural elements of a unit, which impedes the development of an algorithm for recognizing their structure. The use of a neural network prototype in automatic mode is advisable for relatively simple parts (including a flange, hole, chamfer and rounding). However, since the number of simple units from the recognition point of view amounts to about 40% among the nomenclature of manufactured units, the reduction in the development time of the technological process in comparison with the conventional approach comprises only 10–25% of the total time of technological preparation.