AWS Startups Blog

How SupWiz is Revolutionizing Customer Service with AI

supwiz uses ai to improve customer service experiences

Take a second and think of an aspect of a business that really frustrates you. An aspect that you, and every other person, seems to have negative experiences with. What comes to mind?

Despite the variety of possible answers, the majority will likely have an underlying commonality: poor customer experience. Long hold times, answering the same questions over and over, weaving your way through a phone maze to find the right person to talk to. It’s annoying but seemingly unavoidable—at least for now.

Founded summer 2017, SupWiz is making that poor experience a thing of the past by delivering a customer service AI platform with e.g. virtual agents and chatbots that can understand customer request in the domain language of a given company and resolve issues to deep integrations such as resetting WiFi routers.

But what does that mean? These days it seems like every company is boasting their application’s “AI” abilities, when in reality what’s under the hood and behind the scenes, is far from cutting edge. In a field that has a lot of hand-waving, SupWiz stands out through a team with deep technical expertise and experience.

All companies have a founding story. Whether it’s having a difficult time calling a cab or trying to make extra money by renting out an extra bedroom, they must start somewhere. For SupWiz, this path traces its way back to academia. It wasn’t that long ago, when the now team of 20, was just a few researchers looking to make their next move, per Søren Dahlgaard, Co-founder and Chief AI Officer at SupWiz.

“The origins of SupWiz came from a desire to make an impact outside of academia. At the time, Mathias Bæk Tejs Knudsen (CTO and co-founder of SupWiz) and I were grad students completing our PhDs in computer science, with a focus on algorithms at the University of Copenhagen. Although we really enjoyed our time in academia, we felt that many of the projects that were being worked on would never have real-world applications.”

So the two of them teamed up, with a goal of building something real. One issue arose though, as both had deep expertise in the AI field, but lacked the experience of launching and scaling a startup. For this and more, they turned towards one of Dahlgaard’s academic supervisors Stephen Alstrup. Prior to his time as a professor, Alstrup co-founded and served as CEO of Octoshape – a video streaming optimization startup that was acquired by Akamai in 2015.

Headshots of the co-founders of customer service AI startup SupWiz

From left to right: Mathias Bæk Tejs Knudsen, Søren Dahlgaard, & Stephen Alstrup, Co-founders of SupWiz. Photo by Maiken Kestner.

Having secured a team with both deep technical expertise and founding experience, they were ready to make the jump. The next step? Figuring out what industry to jump into. While some founding stories are told as serendipitous events—like being frustrated with how difficult it was to call a cab – SupWiz’ story falls more in-line with their academic research background. The team took it upon themselves to attend conferences and ravenously read reports to understand where the largest opportunities were laying. One such report came from Gartner, which stated that 89% of companies believed that customer experience is their primary basis for competition. With a market selected, the team was ready to go.

But what does SupWiz do? Applying their years in the AI academic world, SupWiz set out to build a better customer experience for large companies. As Dahlgaard puts it, “When you look at the customer service industry, it’s really surprising how much of the work is currently being handled manually. Not only is that not very cost effective, but our technology can actually outperform the current manual methods in many ways. There is also a lot of repetitive work in this line of business, which is a perfect application for computers and AI to optimize.”

So that’s where the startup operates. For each new client, SupWiz first train its internally developed algorithms on the company’s existing data tuning it to the specific domain language. The customization is needed as e.g. ISP’s have different communication styles than banks, which have different styles than airlines – and so on. Once that’s done, the team deploys its AI platform to automate portions of the customer service process. This changes depending on the need of the company, but can for example include virtual agents and chatbots to automatically handle customer requests, identifying the right knowledge article for service agents, or automatically routing tickets to the right teams.

Dahlgaard points to this approach of training company-specific language models as a key differentiator for the startup. Leveraging that deep technical knowledge of SupWiz’ team consisting of 85% PhDs within AI, statistics, and computational modeling, SupWiz’ AI platform is able to understand highly domain-specific language and outperform the current methods used for customer service, automatically resolving 75% of customer requests.

And as for the customer service representatives that currently handle these tasks? They’re actually huge fans of the SupWiz, as it takes away the boring part of the job and allows them to focus on more interesting things.

Customer-wise, the 2-year-old startup works with companies across a variety of industries, including finance, insurance, telcos, and pharma benefiting millions of end-users in more than 149 countries. If you have a lot of inbound volume for your customer service department, then SupWiz is happy to chat. That said, the nature of their product and services typically means that the startup ends up working with companies that have dedicated customer service teams of 10-20 people and up.

To power its AI solution, SupWiz turned to AWS. The team was able to build their own framework from the ground up, which includes heavy use of AWS dedicated machine learning and GPU heavy computers. They also leverage Kubernetes to help deliver the SaaS application at scale. “AWS offers the best-suited infrastructure for our demanding machine learning algorithms in the right geographical regions, making it the perfect fit for SupWiz” notes Dahlgaard.

So, what’s next for the startup? Scalability, according to Dahlgaard. “As of now, each new customer requires a bit of hand holding. To grow globally we will need to make that process a bit more automated.” If the past two years are any indication, it’s not hard to imagine the company will have continued success in its field, and, hopefully, will decrease the amount of time we all spend frustrated at customer service reps.