Information systems have developed a number of questions that are often asked to customers service with a high level of question similarity. For help it create a FAQ related to information systems. The FAQ has a lot of information so that the user is confused and it takes time to find information. More users to ask questions to customer service. Chatbot is one way to help users and customers service in this matter. Customer service can answer questions automatically and users can ask questions as if to ask customer service live. In this study, researchers will apply Nazief & Adriani algorithm used to do stemming because this algorithm is an algorithm which is effective in stemming Indonesian, and method cosine similarity to find the similarity level of questions with FAQs that are in the FAQ database. With implement algorithms and methods in chatbot with the help of the Telegram Messanger service produce a relative answer according to that the user expected. This method is an effective way to answer questions automatically. And if the question asked did not find an answer or the answer is not suitable then the question will be conveyed in e-layanan Information technology development center of Universitas Negeri Surabaya.