What is Conversational AI?

• Conversational artificial intelligence (CAI) refers to technologies, like chatbots or voice assistants, which human users can communicate with.
• CAI for businesses has a very sharp focus: help business users (customers, employees) solve specific set of problems
• Compare this to general CAI where the focus is on engaging users in chitchat about some domain (e.g., sports)
• As a result of these completely different focuses, the evaluations of the CAIs also varies
  • Businesses evaluate to what extent the problems are solved
  • General CAI evaluate to what extent the system is able to mimic conversation with a human (the so called Turing Test)

Why Implement Conversational AI?

• Whether you call them virtual agents, digital assistants or chatbots, they are already driving significant changes in many organizations.
• Consumers are becoming more demanding. They now expect an immediate service response anytime, anywhere and on any device.
• They’re looking for self-service options that enable them to get information and solve problems via virtual agents without waiting for human intervention.
• Virtual Agents are becoming the key to employee productivity and happiness by answering employees in the moment they have the question

What are the Challenges of Conversational AI?

• There are 2 main challenges for businesses implementing CAI:
  • Quickly spinning up multiple virtual agents (for supporting different business functions such as customer support, HR, etc.) without investing in a large technical team
  • The real power of a virtual assistant is the knowledge base (KB) that it consults to answer user queries. The ever changing content of these knowledge sources needs to be fed to the bot so that it responds dynamically with the latest information
• LearnITy™ Conversation (LC) is a Conversational AI (CAI) Platform addressing these challenges. It is built for businesses wanting to implement CAI rather than to support general CAI
  • Quickly spin up a chatbot
  • Focus on dynamic knowledge

Quickly spin up a chatbot

• LC supports quickly spinning up a chatbot by uploading a simple Excel file that contains the bot responses and the happy path navigation
• Happy path navigation is a technique now adopted throughout the industry where users mostly click on buttons to interact with the bot rather than enter text
• This avoids the frustrations with incorrect intent and/or entity recognition (the infamous sorry don’t understand syndrome) that often results from user entered text
• Modification of the bot responses as well as navigation is made very simple: update the Excel file and reupload
• LC supports testing of the bot inside the Admin module before deployment to a server

Focus on Dynamic Knowledge

• Constant changes in the underlying knowledge sources (e.g., Government regulations for BFSI) is a reality in today’s business conditions.
• Human agents need to continuously train themselves by consulting the latest documents so that they can give relevant and correct answers to user queries.
• It would lead to a very inefficient system if the virtual agents also need to be manually changed every time to reflect these changes
• LC comes packaged with document understanding capabilities whereby a virtual agent is automatically provided with the updated information from the version of the uploaded documents

Digital Self-service Platform

• LearnITy™ Conversation (LC) automates various human labour intensive tasks related to building a digital self-service platform
  • Quickly convert your FAQs to a ChatBot
  • Implement multiple virtual agents (Chatbots) serving different stakeholders (customers, employees) or different business areas
  • Easily connect your Knowledge Bases and Databases to the Chatbot so that it answers queries based on the latest updated information
  • Get detailed analytics of the service requests and their fulfilment by the virtual agent from rich dashboard

LC is a CAI Platform

• LC is a Conversational AI Platform
• It is provided as a collection of capabilities that are not tied to any specific business function or business vertical
• It may be used to support the CAI requirements of any business function, Finance, Legal, Operation, HR, …
• It may be used in any business area such as Banking, Healthcare, Manufacturing, …
• LKE empowers organisations of all sizes to quickly implement digital self-service, without requiring them to big investment in AI/ML expertise

How LC Works?

• A self-service workflow that need to be implemented by a bot is very easily created using an Excel file.
• LC converts the contents of the Excel file into a Dialogue State Machine (popularly called Skill in the industry).
• Intents are used to represent the various operations that users wants to perform; Entities are used to represent the various data elements that users select in their queries
• LC processes the intents and entities using NLP (Natural Language Processing) and ML (Machine Learning)
• It communicates with backend applications, databases and document stores for fetching the information required to service the request and reply to the query

LC – Under the Hood

• LC converts a self-service workflow into a Dialogue State Machine (popularly called Skill in the industry) and stores the representation of this machine in a format designed for efficiency.
• Sets of entities and intents are defined in the system and associated with the Skills (the same set may be associated with multiple Skills promoting reuse)
• In addition to text, button, card, and other standard types, users may create their custom response types using rich media (images, video, audio)
• LC generates the glue code for integrating the Chatbot with a website or other channels