Abstract: A system for hybrid approach to natural language understanding is provided, comprising a machine learning configured to create a new natural language understanding model from user-provided examples, and provide a natural language understanding model dataset to other components of the system; an integrated development environment configured to retrieve the natural language understanding model dataset, and publish a hybrid runtime solution; and an interaction engine configured to: retrieve the hybrid runtime solution, receive a natural language input, and process the natural language input and generate a log dataset; wherein, the log dataset may be used to retrain and improve existing models.