How A Group Of Academics Open-Sourced Conversational AI, And Then Turned It Into A Business

Benzinga's Fintech Focus Podcast features conversations with the biggest names in fintech. Subscribe to the Fintech Focus newsletter to get a roundup of industry news delivered to your inbox weekly, and check out upcoming programming at Benzinga events.

In this episode of the Fintech Focus podcast, we talk about the science of, well, talking. 

Conversational AI has already changed how we interact with computers—thanks in large part to Siri, Alexa and Google Assistant—but we’re only just beginning to see the transformative power that this technology can have on our daily lives. 
 
Leading the charge in this arena is Clinc, a company that was born out of a research project at the University of Michigan. 
 
On Episode 27 of the Fintech Focus Podcast we chat with Dr. Johann Hauswald, co-founder and chief product officer of Clinc, about the company’s beginnings, why they’ve chosen to focus on financial services right now, and what makes conversational AI such a difficult technology to crack.
 
Listen to the podcast below to hear how Hauswald is hoping to help out the financial services world through the help of artificial intelligence.
 

There are conversational AI platforms out there. Siri's not new. Alexa's not that new. It's a couple years older by now. How is Clinc different?
Part of our process to build this conversational AI, we early on identified collecting crowdsourcing data. So, data from the crowd that represents how you would talk to an AI as a fundamental piece in the pipeline of building the technology.

So, what we do and what we recommend to all of our customers to do in thinking about building conversational AI is to collect data that most closely represents the use case that they're looking to solve. And to not worry about how messy the data is because the messiness of that data represents how people would actually talk. And so, our algorithms are architected in a way to support that messy language that's unconstrained where we don't need to put parameters over how a user or a customer of this technology would talk.

How receptive have financial institutions been to your pitching of this kind of technology?
What's been interesting is there are some key players, especially our early adopters, that have been hoping that there is a technology that can come and disrupt in the space.

So, for example, USAA, they've been looking to conversationally enable their enterprise for a while. And then they were one of our earliest customers and so it's been very straightforward I'd say with the customers that have been thinking about this because they know what the technology out is out there in the market, and they've done the validation to say, "I haven't seen anything like this." Because basically what we can do is we say, "Hey, we're a conversational AI technology. We're not gonna bore you with a set of slides, we're gonna show an experience." Because at the end of the day, and this is back to what I was saying early on, as to how fundamentally we think about this, it's about the experience that a customer will have.

You know, AI is typically very scripted, very single turn demos, but here we train our sales and product team to just, almost do unscripted demos of the experience to show just how natural the experience is.

What would you say is the biggest issue right now, the biggest challenge facing Clinc?
We have this very powerful technology and really one of my cores focuses and I think one of the companies is enabling these enterprises to build for themselves. We know how to build conversational AI and we know the level, the bar, at which it should be built, but we need to innovate in a way that we train and educate enterprises.

And so, we're heads down focused on creating an offering that makes sure that customers understand not just the how of using our platform, but really the why and the dark magic that goes behind building robust and advanced conversational AI.