This article aims to explore how to apply Transformer networks to the design of user recommendation systems for WeChat mini programs, in order to solve the problems of personalized recommendations, long-distance dependencies, and data sparsity in traditional recommendation systems. This article introduces the basic principles and characteristics of Transformer networks, as well as their successful applications in the field of natural language processing. On this basis, the Transformer network coding and interaction mechanism can be used to better capture the interests and behavior of users, so as to achieve personalized recommendations. Through experiments, the performance of the two methods in recommendation precision and user satisfaction is compared. Finally, this article presents a new recommendation method based on Transformer network, which can implement the recommendation of WeChat applets, and its coverage can reach 91%. The research results of this project can provide a new research idea and method for improving the performance of recommendation system and improving the user experience.