Context. Today, e-commerce is one of the most active and promising segments of the global economy, characterized by steady growth of volumes and active improvement of tools and technologies. Intellectualization of e-commerce systems today is one of the main trends in their development and effective functioning in a competitive environment. One of the most promising means of solving this task category is the use of intellectual agents (IAs). The modern approach to modeling the decision support process, in particular, in e-commerce, uses the Boyd's loop principle, which involves repeated cycle repetition of four successive interrelated processes (stages): observation -orientation -decision-making -action. According to Boyd's hypothesis, the higher speed of the cycle and the accuracy of the evaluation of the stages of the cycle provide an advantage over the competitor.Objective of the research is to develop tools for effective planning of actions of intellectual agents of e-commerce systems in a competitive environment based on ontologies, and to set the task of planning the actions of intellectual agents to the problem of dynamic programming.Method. The mathematical support and method of action planning of IA based on adaptive ontologies are developed. This method can be used to effectively operate IA in a competitive environment of the SEC, which is modeled by Boyd's loop. Adaptive ontology is used to plan IA actions. To this end, we have added two scalar quantities (the importance of concepts and relationships) that are used to find an effective way to implement the IA plan in a generally accepted three-element tuple that defines ontology (a set of concepts, relationships, and their interpretation). Such an assessment of the ontology elements allows reducing the task of planning IA actions in a competitive environment to the task of dynamic programming.Results. In the article the model of planning of actions of intellectual agents of electronic commerce systems using an ontological approach is constructed. The method of evaluation of actions of intellectual agents on the basis of adaptive ontologies is developed. Such an assessment makes it possible to reduce the task of scheduling actions to the task of dynamic programming.Conclusions. Analysis of the results of IA actions can determine the causes of the formation of the target audience by a set of characteristics of the functioning of the SEC. By regulating the content, its uniqueness, the efficiency of its formation and its adequate management according to the individual needs of a regular user, one can model the boundaries of the target social audience and the number of unique visitors from search engines. In fig. 5-6 the results of the work of the developed systems in the form of charts are presented, from which it follows that in the presence of all types of IA significantly increases the volume of visits and unique users.