Chapter 3 Design • NLU (natural language understanding): This component helps your application to understand the intent and take action on it. After understating the intent, you can call APIs to interact with the external system and get or put information in other systems. • NLG (natural language generation): After understating intent and getting external information from other systems and databases, you need to generate the resultant message and send it back to the client. • You can use open-source bot frameworks such as botpress, WIT.ai, etc. or use chatbot platform services like chatfuel, dialogflow, etc. to build a chatbot, or you can your create a custom framework using ML and NLP libraries with the help of a recurrent neural network and bag-of-words model. • Integration with legacy system: If you are building a chatbot for a business, then most likely you are working with CRM, ERP (enterprise resource planning), and core banking, etc. You need to integrate the chatbot with an external system using REST APIs (external). Figure 3-28. Chatbot Integration with DXP 108