Chatbot for customer support of a company which operates charging stations for electric cars

A scenario for automated, around-the-clock interaction with users of electric vehicle charging stations, based on artificial intelligence, has been proposed. The relevance of an intelligent virtual assistant for supporting customers by the service departments of companies managing charging station networks is discussed. The necessity of employing market-leading technologies, such as OpenAI's generative artificial intelligence models, for the creation of a chatbot is substantiated. The importance and timeliness of integrating the developed chatbot into WhatsApp, the messenger with the largest audience reach, are described. To implement the solution, the potential for applying the most advanced OpenAI service technologies was analyzed and demonstrated. Specifically, the GPT-4o-mini model was selected for generating text-based responses and customized using system context. The model was subsequently configured in the OpenAIAssistant class, where a method specifying how the model should function according to the provided algorithm or scenario was developed. An algorithm for the model's operation was designed and reviewed. The model must be capable of communicating with users in various languages. Communication is guided by a predefined scenario based on prompt templates. Additionally, the model is programmed to respond to questions that fall outside the provided scenario or algorithm, to questions posed in unknown languages, and includes prompt-based safeguards against incorrect answers that AI occasionally generates. For the chatbot's implementation, technologies were employed, including Whapi.Cloud services (a platform for automating interactions with the WhatsApp messenger API) and Vultr Cloud Provider (a provider for hosting the server-side application). The chatbot's business logic was implemented using programming languages and frameworks such as JavaScript, NodeJS, and Express. The practical implementation of this development confirmed the feasibility and effectiveness of the proposed approach, as evidenced during testing within the WhatsApp messenger. The proposed article is of interest to software developers working in the field of generative artificial intelligence.