Labour productivity growth is a necessary condition for social and economic progress, in general, and to overcome the economic crisis facing most of the world, in special. Applying innovative solutions, based on the ITC, is one of the straight ways for achieving that objective, both important and necessary. This paper presents a software solution applicable to industrial production based on numerically controlled machines. It involves a distributed client -server communication system, combined with MLP neural networks for the recognition of the 2D industrial objects, viewed from any angle. The information on prismatic and rotational parts to be processed by numerically controlled machine, are stored on a database server together with the corresponding processing programs. The client applications run on the numerically controlled machines and on the robots serving groups of machines. While the machines are fixed, the robots are mobile and can move from a machine to another. As a novelty of the proposed solution, in some well defined situations, the clients are allowed to change messages among them, in order to avoid the server overload. The neural networks are used to help robots to recognize the parts before and during manipulation.