GTMR deals with the development of an application based on advanced deep learning mechanisms, capable of generating models predicting the behavior of users of any mobile or desktop application. Based on the information provided about the use of the application, models with neural networks are able to predict which user is likely to end the subscription and take preventive steps in advance. As part of the project, a number of models were developed and tested to create a deep learner platform, and based on it, a prototype of an IT tool was implemented. This application provides automatic research on the behavior of computer application users, in particular mobile applications, in the context of their retention.