As the main train service provider company in Indonesia, PT Kereta Api Indonesia (PT KAI) has many customers who need information. In order to maintain customer loyalty, PT KAI must respond quickly and be adaptive to technology to provide the best service to customers. Limited human resources make PT KAI unable to serve customers simultaneously, so customers often have to wait for a response. In order to provide the best service, automatic messages are needed in order to help customer service performance respond quickly and at the same time with no cost, access anytime and anywhere. This study proposes a new approach with chatbots as a medium for conveying automatic information quickly and simultaneously. This chatbot is made with a computational language that focuses on natural language modeling and cosine similarity as a method for calculating the proximity of inputs and databases. This research can help PT KAI's customer service workers to answer customer needs automatically.