Russian steel maker Magnitogorsk Iron & Steel Works Rolling shop No 11 is undergoing pilot tests of an automated advisor system for internal certification of coils produced at a continuous hot dip galvanizing unit. The digital solution, which is based on machine learning and predictive analytics, is aimed at improving the quality of metal products and preventing possible claims from consumers. The system was developed on the basis of data obtained from the Parsytec flaw detector, which, using a special scanner and software, analyzes the moving belt, detecting and classifying possible defects along its entire width. The flaw detector in automatic mode is able to recognize more than 40 flaws in metal products, such as holes, rolled particles, film, scale and others, and issue a roll defectogram to the operator.The system, using information from the flaw detector and a mathematical model trained on historical data on the results of roll certification, issues a recommendation to assign a status to each produced roll: “good” or “defective”. In addition, at the stage of shipment of finished products to the consumer, the advisor system will assess the likelihood of a claim from the consumer: in case of a positive conclusion, the products will be sent for additional examination (visual inspection, examination of defectograms, etc.). If it is found that the metal does not meet the requirements of the consumer, the shipment will not be made.MMK is constantly working to improve the satisfaction of its customers, and the advisory system developed by specialists from the Competence Center for Mathematical Modeling and Advanced Analytics MMK-Informservice will significantly reduce the likelihood of deliveries of products that do not meet customer requirements, which will reduce the number of complaints from consumers. During the testing period during the internal certification of coils of the continuous hot dip galvanizing unit in sheet rolling shop No. 11, the new solution showed an efficiency of 91%.