Model-Driven Chats: Enabling Chatbot Development for Non-technical Domain Experts Through Chat Flow Visualization and Auto-generation

Chatbots can significantly reduce the workload on client-facing service agents by conducting a first screening and answering frequently asked questions. Reaping these benefits, various organizations now implement chatbots. As typical in an iterative project within multi-disciplinary teams, information can be lost in translation. This paper presents the first proof-of-concept demonstration of creating an editable model for conversation flows development in the IBM Watson Assistant chatbot platform. Chat flows are auto-generating after changes are made to the model thus enabling our non-technical mental health domain expert partners to build a chatbot application by simply drawing the chat flow.