An important aspect of the process of constructing artificial neural networks is the ability to visualize the process of their construction and operation. It allows us to gather much more information about the structure and processes of interaction of neurons, analyze the efficiency of each component of the network. Such opportunities are provided by the neural networks constructor with a graphical user interface. Existing applications, that support the construction of the artificial neural networks, provide the tools to construct networks of strictly defined, standard types, which heavily limits their usability and makes it impossible to use them as an experimental platform for the rapid testing of new models and algorithms of artificial neural networks. To develop a neural networks constructor, which will solve these problems and limitations of the existing applications of such type, a number of scientific activities were accomplished, among those: the identification of functioning features of artificial neurons and artificial neural networks of different topologies; the development of a user-friendly interface with support of building a graphical scheme of artificial neural networks from a set of main structural elements, choosing the activation functions of the neural neurons, setting up the optimization algorithm and loss function; the development of means of internal representation and interpretation of the graphical scheme of artificial neural networks in the corresponding algorithm of functioning; the software implementation of the functioning of artificial neural networks based on the interpreted algorithm of internal representation. The neural network constructor developed in the course of the research is a software system that allows the user to build models of neural networks of arbitrary configuration with the help of a graphical interface, as well as to perform calculations according to an algorithm adequately formed by an internal interpreter and investigate the influence of various functional parameters on the results of training and operation of the artificial neural networks. The software development tools provided maximum flexibility of implementation of the interaction with the user through a graphical interface. The software product can be used both by beginners in the field of artificial intelligence and by scientists as an experimental platform for the rapid testing of new models and algorithms of artificial neural networks. Keywords: artificial neural networks, constructor, activation function, graphical scheme of neural networks, graphical interface.