The paper presents both an intelligent approach to better management of pedestrian flows by means of a mobile application and a simulation model to asses this approach. The proposed concept is supposed to be used at heritage sites, which are overloaded by tourists, and which have a number of potential attractions within a walking distance. The proposed approach assumes to use data of a number of visitors at particular attractions to suggest application users the best attractions to visit next. To evaluate the proposed idea, an agent-based simulation model is developed. The model allows evaluating how introduction of such application among a number of visitors allows reducing waiting times in queues to frequently overloaded attractions. The structure of the simulation model and behaviour of agents in the model are described in details. The positive and negative sides of the proposed concept are determined through simulation experiments and discussed.