preface

Conversational AI is an exciting technology that helps users fulfill their needs faster and helps companies handle user inquiries with lower cost. Conversational AI solutions (often called chatbots) have exploded in popularity, especially since the COVID-19 pandemic. There are many books and blogs on how to get started with conversational AI, but most of these books stop at building your first chatbot and do not describe how to improve a production solution. Many enterprises use this technology so that their customers can self-service on a scale that may be prohibitively expensive or impossible with a human workforce. Unfortunately, a significant proportion of these AI solutions underperform.

There have also been plenty of hype and resources on generative AI, including prompt engineering and small demos. However, these are often small-scale in nature, such as proofs of concept and prototypes. There are few resources for maintaining and improving these solutions at an enterprise scale. Generative AI has reignited interest in this space, but it is not a panacea, especially for enterprises offering end- to-end task completion.

We have delivered many conversational AI solutions to production in the past decade. We have worked with a variety of chatbots: question-answering, process-oriented, and routing agents. We have seen the joys and challenges of conversational AI up close.