Making interactions more efficient with Knowledge Management and AI

NITESH DUDHIA

Knowledge management, while seen as a very useful thing, has never quite been a critical
requirement at most organizations. The COVID-19 pandemic has forced us to rethink our
approach to knowledge management as co-located teams decrease, and work from home takes off.

Where does your organization’s knowledge reside? Is it in occasionally updated knowledge-based articles, often ignored FAQ’s, or perhaps in your CRM or intranet somewhere? Let’s face it; most of the organizational knowledge resides is tacit and lives within the people who get work done. So what happens when these people are too busy to answer a question or leave the organization?

We’ve built excellent systems to keep records, but they are usually siloed and ill-suited to
storing knowledge. We have brilliant systems of engagement too, digital channels to connect with customers and employees wherever they may be, but fall short in the intelligence department. What Jerry Chen said in 2017 about the need for a system of intelligence to bridge a system of engagement and a system of record couldn’t be truer today.

Augmenting knowledge management with AI

What does a system of intelligence look like? I believe that a system of intelligence is essentially a knowledge management system that leverages systems of records to make engagement with users more effective and smarter. The challenge with traditional knowledge management has always been about who will create, curate, and maintain knowledge. No one wants to write Knowledge Base articles or update FAQ’s, and this is exactly where artificial intelligence and interactions step in.

Augmented AI Defining the New Reign of AI Advancements

Experts agree that most enterprise knowledge is exhibited & demonstrated during
conversations (in-person, over the phone or from behind the screen), and this is where we
should capture the insight, create context and save it for future re-use. It is impractical to find a conversation, let alone listen to all of it, and get to the part that you needed. Recent advances in AI can take care of collecting relevant information, insight, and intelligence from conversations -regardless of where they happen – to create a repository of knowledge that can be used to make future interactions smarter and more efficient.

Discover knowledge once, re-use it across systems

Imagine automatically using an answer to a product issue reported via email as a response to another customer who is asking for the same thing via chat. Let a bot inside SLACK or MS Teams answer recurring questions and automatically learn new answers when a human answers a question that the system of knowledge had not seen before. This scenario may seem utopian, but it is well within reach of technology available today. With more and more people preferring digital conversations, perhaps out of necessity, we will see a huge glut of data that we don’t do much with at the moment.

How many times have you heard the words  “this call will be recorded for quality and training purposes”  only a fraction of these calls are sampled and analyzed. Today, AI can listen to these recordings and create a meaningful knowledge graph out of the conversation content and relationships between the participants so that future conversations can be enriched. Natural Language Processing technology is getting better each day, and now not only do computers understand what we say and write, but they are also starting to generate natural-sounding sentences too.

Google Launches AutoML Natural Language

It’s not all smooth sailing though to use AI for creating and curating knowledge. We must
ensure that it works with tools that we are already in love with, and there must never be
knowledge management overhead on people. They just go about their business, let AI and
associated platforms create knowledge, and push nudges or answer questions when called upon. The incentive to validate knowledge or be recognized for their contribution to knowledge is critical to get humans to share their hard-acquired knowhow. This is the guiding principle of how we are applying natural language processing and knowledge graphs to capture and re-use enterprise knowledge while creating tools for our customers.