Krishna Raj Raja,  is the CEO and Founder of SupportLogic, the world’s first continuous service experience (SX) management platform that enables companies to understand and act on the ‘Voice of the Customer’ in real-time to immediately improve service delivery and build healthy and profitable customer relationships.

You describe yourself as a “science geek”, what is it about the scientific world that has you so enamored?

To me, science is all about curiosity around how the world works and what patterns exist in the world. I’ve had this curiosity since childhood, and it has extended into my professional life. At SupportLogic, we are looking for patterns that exist in natural language, and using those patterns to predict things and provide recommendations. This is like science – it predicts patterns, provides recommendations and tells you how the world works. Much of what I have learned in my life is self-taught because it came from my natural curiosity, not from formal education.

You went on to get a degree in Chemical Engineering, but then ended up pursuing opportunities in computer science and machine learning. Could you discuss this pivot?

My dad was a successful business owner (in steel processing) and he hoped that one day I would take over the family business. Chemical engineering seemed to be the right training to have in order to be ready to do that. But computer science had been a side passion for me for a long time. In fact, my thesis for my chemical engineering degree was creating a software program for a chemical reactor design.

After I finished my chemical engineering degree, it became obvious that computer science was really my true passion. My computer science skills are completely self-taught, and when I joined VMware all of my colleagues had PhD’s from Ivy League schools in computer science. I was the most underqualified person on my team.

You were an early employee at VMware, where you worked in support & service as a product support engineer. What were some of the potential areas for improvement that you observed?

I was a software engineer that turned into a product support engineer. I joined VMware because their technology was fascinating – they were dealing with operating systems and I had a special interest in that. I was helping other operating system developers use VMware products on a daily basis. Because of my background, I was looking at things from two different angles: 1) How do I make this customer happy and resolve their issue; and 2) Why does this problem exist in the software, and how can it be fixed? I was looking at the product aspect of all the support issues. One of the first things I realized was that when product teams develop a product, they really don’t know how it will get deployed and used, so they don’t foresee a lot of things during the development process. However, the support team has a good handle on those issues and can give valuable feedback to the product teams as well as other departments in the company. The problem is that this feedback usually gets lost because the support team is focused on fixing a customer issue and then quickly moving on to the next issue. That important information does not get passed on.

Could you share some details on the genesis story for SupportLogic?

When I started SupportLogic, I looked at the market landscape for the support industry and I found that all of the innovations in the support space were focused on case deflection. This means that the best way to deal with support issues is to deflect them – away from support engineers, and away from the customers. This was in conflict with what I saw as a product support engineer – every customer interaction was an opportunity to learn about your customers and about how the product works and doesn’t work. But I was surprised to find that there were not tools out there to solve this (learning) problem, so I saw a big opportunity there.

Also, I noticed that support was often being treated as a cost center, which I thought was a short-sighted way of looking at things. When you look at support as a profit center or as the central nervous system of a company, you can really transform an organization and make them truly customer centric. That’s what lead me to start SupportLogic.

What are some of the different machine learning technologies that are used at SupportLogic?

When the company started, I was naïve in assuming that we could use publicly-available machine learning APIs. There are a lot of them – from Amazon, Microsoft and HPE – and they all provide machine learning APIs as a service. To my surprise and disappointment, many of these machine learning models did not work with the kind of data we were working with (customer support data). But I realized this was an opportunity and said, “why don’t we build it ourselves?” We started building our own from scratch using existing ML technologies from open-source projects, like spaCy from Stanford University, and Google BERT, and then added some of our own “secret sauce” on top of that, using an ensemble model approach. We also fine-tune the model for each customer and their specific data set, rather than using a one-size-fits-all philosophy.

Could you discuss how SupportLogic enables companies to better connect with customers by using key signals?

One of the key things that we do is extract customer context using NLP. The context is very important because context often gets lost in the tagging process of ticketing systems. You can only tag a limited amount of information in those systems. We excel at extracting customer context, such as what are they frustrated about, what’s their impression of your product or your support, or what are they trying to do with your product. There are a variety of signals and context to extract. By doing this in real-time and creating workflows in our platform, we allow companies to act on customer signals and preemptively fix issues before it is too late – meaning the customer gets very angry or goes away forever.

What are some of the other capabilities behind SupportLogic software?

Once you start extracting customer signals from interactions, those signals become very powerful for analytics. We have an analytics module that tells you what the voice of the customer looks like, based on all of the interactions. Then we go one step further and use the data to start making predictions. We can predict what will happen with a particular (customer) account. We can also predict – based on the customer situation – who is the best subject matter expert in the company to help fix the problem, and then match the customer with that right person.

And we can look at both inbound conversations and outbound conversations to give service agents guidance on what they should be doing more (or less) of in their day-to-day interactions with customers. It becomes a great coaching tool to help service agents develop their soft skills and improve their overall performance.

Is there anything else that you would like to share about SupportLogic?

One of the common misconceptions that people have with AI is that it is a massive investment that is very involved and complex, and that you will not see any returns on it for a year or more. In reality, AI and ML technologies have matured a lot, and can work on your existing data set. And you can see results in a matter of months, not a matter of years. So, now is the time to invest in AI because you can see incredible results in months that can yield great benefits for your organization.

Thank you for the great interview, readers who wish to learn more should visit SupportLogic.