Education has undergone several evolutions, made possible by Information and Communication Technologies (ICTs). As an example, we can mention the distance education modality, radios and televisions have been replaced by Virtual Learning Environments (VLEs), which allow a greater effectiveness of the teaching and learning process, because, in addition to integrating students, teachers and tutors in one place, allow better communication between them. On the other hand, the use of VLE has generated a concern with the quality of the teaching and learning process, which is linked to the effective interaction of the actors in this process. In this sense, it is known that the tutor is the actor responsible for promoting interaction in these environments and that their performance directly impacts the effective participation of students. Therefore, this dissertation presents the development of a software agent capable of recommending pedagogical strategies to tutors in VLEs, aiming to improve performance to promote an effective interaction of students in VLEs. The developed agent makes the recommendations through a knowledge model, represented by an ontology, which maps the pedagogical strategies in order to promote the interaction in the VLEs. The pedagogical strategies used by the agent were raised through a systematic literature review and the application of a questionnaire with professionals in the field. For validation, the agent was applied to historical data through a developed system, in which it was subjected to use with 10 professionals working with distance education. The results show that the agent has the potential to improve the performance of tutors and assist in promoting effective interaction within VLEs