Table of Contents for

Practical Bot Development: Designing and Building Bots with Node.js and Microsoft Bot Framework

Szymon has 15 years of hands on experience in the software development industry. He found a passion working on front end applications on Wall Street. The interest in attention to user interface details would take him on a tour of the different user interface technologies on the Windows, Web and iOS/Android platforms. He has managed teams of engineers on a variety of projects and since 2016, he has dedicated his time to building chat bots for clients across text and voice channels. He has built a practice around the technologies at Blue Metal, An Insight Company, and is involved in a handful of chat bot projects at any time. As Chief Architect of Emerging Technologies, he keeps cognitively flexible by reading and watching presentations about new technologies, educating clients, leading discovery sessions, scoping and guiding engagements, mentoring engineers and helping clients apply technologies such as Chat Bots, Blockchain and Augmented Reality to business problems.

In his spare time, he enjoys taking walks, reading fiction, going to the beach, playing guitar, and spending time with his wife Kim, his son Teddy, and his Golden Retriever, Chelsea.

In mid-2016, I started work on a fascinating project. A client wanted to have users with type 2 diabetes be able to get advice from an automated coach, in other words, a chat bot. It was a fascinating idea. I had many questions. Why would anyone want to have a natural language conversation with a machine? Is it possible to make this thing smart enough to accomplish its goals? How do you even begin creating a chat bot? Through which means are users supposed to interact with it? When the project ended, we quickly realized that the technologies we just utilized, natural language understanding, Microsoft’s Bot Framework, and custom machine learning models, could serve as the technical base for a wide array of natural language applications between users and computing systems. Natural language interfaces, after all, were all the rage. The Alexa-powered Echo Dot had just been released, and the general population was quickly becoming fascinated with the idea of communicating with digital assistants by talking to them. My friend and I jumped on the opportunity to become experts in this space.

We tried many of the different platforms such as Api.ai (now DialogFlow), Wit.ai, and Watson Conversation but decided to focus on Microsoft’s Bot Framework as we felt it best complemented the enterprise space. Chat bot startups sprung up across the landscape promising the best bot or bot platform. The field became saturated, and our customers started paying attention. Suddenly, I found myself talking to multiple clients a day. Initially, these were high-level mentorship sessions. What is a chat bot? How does it work? What channels does it work with? Does it learn by itself? Can it integrate with live chat?

Starting around mid-2017, these conversations slowly shifted from educating clients to scoping the development effort for all types of use cases. Clients were starting to apply the technology to solve their business problems. In the second half of 2017, while working on delivering multiple chat bot implementations, a colleague of mine connected me to the editors who would make this book happen. Taking on this project was an easy decision to make; it was a fascinating topic, it was a new space, and it had lots of possibilities. Being slightly on the naïvely optimistic side about a three-month old’s sleep patterns, I jumped on the opportunity.

I decided to write the book in the same way I would mentor engineers on the topics. It is divided roughly into three sections. First, I introduce the topics of chat bots and machine learning (ML) in Chapters 1 and 2 . Although chat bots can, and often do, exist independently of any ML algorithms, the truth is that users expect chat bots to exhibit some forms of intelligence, minimally a bit of natural language understanding. As such, I want to set the state on ML and how it may be applied in natural language conversations. Chapter 3 is a deep dive into Microsoft’s Language Understanding Intelligence Service (LUIS), which we will utilize to create natural language understanding models for our chat bot.

The second section is all about hands-on Bot Framework development. Chapter 4 introduces the concept of conversation design, which is the practice of modeling a chat bot conversation. Chapter 5 guides us through the creation of a Bot Framework chat bot connected to a LUIS model and its deployment into an Azure app service. In Chapter 6 , we take a step back to examine the features and capabilities of the Bot Builder SDK, and in Chapter 7 we integrate our chat bot with an OAuth implementation and external APIs. Chapter 8 is a deeper exploration into Slack bots, and, finally, Chapter 9 explores the ability to connect any channel to our chat bot via the Direct Line API. We marry Twilio Voice into the picture to create a chat bot we can talk to via a phone.

The third section addresses a few additional topics that are essential to chat bot development. Chapter 10 adds extra intelligence abilities to our chat bots by focusing on a select set of Microsoft’s Cognitive Services. Chapter 11 explores two manners of creating custom cards for our chat bots: adaptive cards and custom graphic renderings. We explore human handover in Chapter 12 , we learn about chat bot analytics in Chapter 13 , and, finally, in Chapter 14 we put all our newly gained knowledge to work by creating a simple Alexa skill using Amazon’s Alexa Skills Kit and then replicate the same experience using a Bot Framework bot.

The chat bot space is dynamic. During the development of this book, Facebook acquired Wit.ai and shifted its focus to natural language understanding, Google acquired Api.ai, LUIS changed user interfaces not once but twice, the Bot Framework was officially released and moved to Azure, QnA Maker was officially released, Alexa’s user interface changed after I had written all the content, and Microsoft announced tons of new features at Build 2018 (which luckily did not drastically change the topics treated in this book). We have reached a point where the subjects of this book are relatively stable. It is my hope that the content is essential to any developer looking to begin chat bot development using Microsoft’s Bot Framework.

This project has been a humbling experience and would not have been possible without a small group of people for whose support I am forever grateful. I would like to thank my wife Kim, without whose patience, kindness, support, and late-night help with editing I would have never completed the process. I would also like to thank Jeff Dodge for the collaboration on building our chat bot practice, Bob Familiar for introducing me to the Apress team, and BlueMetal for allowing me the room to work on this project. Big thanks to Matt, Jimmy, and Andrew, and my parents, Hanna and Krzysztof Rozga, for providing much needed moral support and words of encouragement during times when I felt I was white knuckling it. I would also like to acknowledge the Apress editors Natalie and Jessica for their support during the book writing process.

This book is dedicated to Teddy.

—Szymon Rozga

June 1, 2018

Port Washington, New York