Traditional Machine Learning approaches for building cognitive chatbots need technical skills for applying supervised learning/NLP and a lot of effort to train hundreds of intents by providing, for each intent, many examples of questions. Moreover, a small change in the knowledge base requires again technical people and time to retrain and deliver even small updates. EVA by App2Check is a chatbot platform that allows to build cognitive chatbots with no technical skills and to maintain the chatbot Knowledge Base (KB) like a CMS for a website. The implemented methodology allows to start with a fully unsupervised approach and then to focus just on the improvements needed. Our solution also allows to search for both answers in the KB and indexed documents. EVA can be easily integrated with any Dialog Manager and is natively integrated with EngageOne Converse by Pitney Bowes.