The Curious Case of a Chatbot: Agile meets Design Thinking
“Bots are the new apps,” said Satya Nadella, CEO Microsoft, during a nearly three-hour keynote at the Build developers, USA. “People-to-people conversations, people-to-digital assistants, people-to-bots and even digital assistants-to-bots. That’s the world you’re going to get to see in the years to come.”
If you are wondering how to make your sales and marketing teams more efficient, leaner, highly predictable and reduce training costs, chatbots are the answer. This talk will explore why and how to go about developing Chatbots as a product for Customer Service and Lead Classification.
Business has always been about customer conversations and Chatbots are the latest platform for customer engagement. Chatbots, or conversational interfaces as they are also known, present a new way for individuals to interact with computer systems. The Chatbot uses Machine Learning, an application of AI in which continuous access to data results in adaptive responses and guided actions, enabling a real-time customer experience. Natural language processing technologies are used for parsing, tokenizing, and classifying the type of queries.
The objective is to discuss in detail the nuances that are involved in building an intelligent assistant that understands the user query and extracts the information (Intent-> Action-> Entities->Outcomes) from the query using NLU (Natural Language Understanding) for contextual conversations. Based on this, the outcome is to classify the query as a prospective Sales qualified lead (SQL) / Marketing Qualified Lead (MQL) or a Customer Service query.
In this case, the chatbot framework is built on a structure in which intents are defined. This is done using a JSON file where each intent contains:
- tag (a unique category/class name)
- patterns (sentences belong to the category/class)
- responses (response if sentence matched)
Agile methodology has been adopted for creating this Chat-bot as an in-house product for lead engagement and lead classification to address the pain points of the Sales and Marketing team as to what is a “qualified” lead. The Sprints include developing the Chatbot as a minimum viable product in iterations, by outlining user stories, problem scenarios, and value propositions and adding more features and complexities in subsequent iterations.
Outline/Structure of the Talk
Defining the Problem Statement
Business Case for Chatbot
Lifecycle and Architecture of a Chatbot
Natural Language understanding – Framework
Chatbot as a Product/Service: Illustration
- Personas: Think, See, Feel, Do approach
- Problem Scenarios
- Value Propositions
Interactive Session on Identifying Personas, Problem Scenarios for a given problem statement.
Designing an in-house chatbot solution as a Product or Service.
An understanding of Conversation Design
(Key takeaways being sharing of the product design template used for this product so that those working on similar initiatives can have a 360-degree view of the steps involved)
NLP enthusiasts ,Marketers, Analysts, Decision Makers, Product Owners, Conversation Designers
Prerequisites for Attendees
Basic ML understanding and the curiosity to know more about the application of Agile to develop AI Products.