Metallurgical plants are complex systems that are difficult to manage. In metallurgical production, the efficiency of the plant is determined not only by the volume of products produced per unit of time, but also by a whole set of factors: the average duration of processing a unit of production in the shop, depending on the type of route of the unit of production, the average load of the units of the plant, and so on. In this study, we took the example of managing a metallurgical plant and used the 'Twin' question-and-answer system for its automation. The 'Twin' system is a multi-channel communication platform for creating voice and chat bots capable of receiving data in one language and transmitting it in another. We focused on the interaction of the chatbot and the user to solve the problem of optimizing the joint work of two workshops of hot and cold rolling of a metallurgical enterprise and optimizing the processes of the enterprise of an automated system for the production of metallurgical products.