Out -innovating yourself in an AI world: Q&A with entrepreneur Devin Bhushan

As the founder and CEO of Squint, a Series B, AI-powered manufacturing intelligence company, Devin Bhushan (’14, B.S., Computer Science) advises student entrepreneurs to keep trying and maximize their shots at a goal. Bhushan built his startup in the manufacturing space, capturing institutional knowledge and creating the ability to quickly and directly apply that knowledge on the factory floor.

As AI transforms our lives and society, Bhushan sat down for a fireside chat with Gerald Wilson, Director of Entrepreneurship, University of Illinois's EnterpriseWorks at Research Park, about what AI means for the future of entrepreneurship and higher education.

Devin, tell us a little bit more about Squint, what problem in the marketplace you're solving, as background for the discussion today. 

I left U. of I. and … I worked on Fantasy Sports for about three and a half years at Yahoo. During that time, I found a mentor, his name was Tim Tully. He was the head of engineering at the time, and he took me under his wing. From there [Yahoo], Tim became the CTO at a company called Splunk. They were a public company, sold to Cisco a couple years ago now. Splunk did data analytics, and when Tim went over to be CTO he thought, hey, it would be really great if we could bring consumer product thinking to this enterprise space and try to fix the interface that Splunk has with frontline operators. People either on a factory floor or out in the field, performing work, interacting with data in a big database, they didn't have any way to query against the Splunk database. And so Splunk had sold all these deals into a lot of companies. And they had multi-million-dollar contracts with Siemens and BMW, and their factory workers could not leverage the data that they had pumped into this system. … My job there was figure out how to close that bridge. 

That’s really where I learned more about the problem that led to starting Squint. As I talked to factory operators, I kept hearing “… Your app's really great; but it's just telling me there's zero revolutions per minute. How do I fix that?" … It turns out they don't have that documented anywhere, and so what they needed was a system that captured kind of the institutional knowledge that had been passed down over generations. 

… There's a bigger problem here to solve than bringing data to the front line. It's capturing the knowledge and the context and building a knowledge base that can be accessible by frontline operators. And so, I started Squint four and a half years ago now.

For the first six months I didn't write a line of code … All I did was validate the idea. We built out prototypes. We went to market, we tested pricing, we did all this stuff. And then finally when we had essentially people saying, "Where do I sign," … we went and built the product and have been in market for three years. 

One of the things I'm really interested in, particularly within the context of a conversation about the impact of AI and entrepreneurship, is what was happening at the founding of your company — the technical and business landscape level — that made the way you're creating value for your customers now possible?

For us it was being very focused on understanding [the problem]. I never worked in a factory.  But our customers’ business is manufacturing. 

… In this case, manufacturing, the biggest pain point was we don't know what to do and how to do it just in time. We innovated in that circle. We found a way to capture that context from the people who knew. With Squint, you can record a video of anyone doing anything. So, if there's someone out there setting up the buffet, we could record a video of them setting up the buffet. Squint's AI would watch that video and three seconds later spit out a step-by-step guide on how to set up a buffet exactly like that person did. 

For manufacturing, that meant we could start instantly documenting work. In countries like Japan this is like a massive issue because they're about to lose all these operators and they've never written down how it works. Chasing the problem was the main thread. 

 

With a better understanding of the problem, you developed a relevant solution. Did you develop some kind of proprietary technology to solve the problem that contributes to the defensibility of your position in the market, and if not, how do you accomplish that? 

… The point was to solve this problem and build the best package for the customer such that it creates value for their business. But yes, in the process of that, I think you run into problems that no one has solved, and that's why this thing doesn't exist, right? And so for us it was computer vision and being able to watch someone do something, then break that down into frames, then break that down into a semantic understanding of what's happening, understand the FDA certified way to operate machinery for food … and then fit the video you just watched into that format and build that out automatically. 

“…Once a record has been set, it's going to be broken ...Once you show something is possible, people will keep doing it. Every day you have to show up, and you have to out-innovate yourself; otherwise, someone else will.” — Devin Bhushan

It’s called a work instruction in a factory, but one work instruction takes about 12 hours of human time to produce, and we do it in about six seconds, so it massively accelerated that process. Then we also built the piece, which was like now using the iPad camera to watch people work, and then verify against the work instruction. So now you can have someone out there who's never done this thing before. They follow this amazing guide, and then it catches their mistakes in real time, in seconds.

Those were key pieces that didn't exist before that we had to build. 

 

AI democratizes innovation. People are able to leverage it, be creative, build things. At the same time, it lowers the barrier to competition. How would you advise founders in the room to think about putting together a strategy leveraging technology to defend your position in the market, particularly given that that's going to be an important component in potential investors evaluating their businesses as opportunities? 

Once a record has been set, it's going to be broken. Until someone ran a four-minute mile, people thought it was impossible. And then once someone did it, it just kept happening. I think the same exists in technology. Once you show that something is possible, people will keep doing it. Every day you have to show up, and you have to out-innovate yourself; otherwise, someone else will. 

One of our company values is called "Dunk the Three." … Dunking a three is something that's never happened in an NBA game. Only Michael Jordan did it in Space Jam. Our thing is we're going to dunk the three. We're going to do the impossible. But you have to do that every day; otherwise, someone else will do it. 

 

 As you look a little bit into the horizon, you're creating value for your customers each day. As you think about their needs, or maybe even as you look at the broader landscape, what are some technological capabilities maybe that might be missing from your perspective to unlock the next level of value creation that AI could potentially bring? 

… Now we’ve given this superpower of creation to everyone. … You can build an app for anything. I think what's changing and where the gap is … how do you bring a product-builder mindset now to non-technical people in your organization? For instance, our finance team and our office manager, folks who wouldn't write code, are using [an AI tool] and are building things to automate their workflows. But there's still a lot of product thinking that's missing. So how do you systematize that for a non-technical audience? I think there's like a lot of open space there, that companies are trying to leverage AI, and they just don't know like how to package it. 

 

As a leader in this space, where do you see some untapped opportunities that could be really exciting problems for the next wave of entrepreneurs to be tackling? 

The one I mentioned is probably one I see right now. There's a lot going on in medicine that is pretty exciting — or in healthcare broadly —specifically on the research side, applying the new models … robotics is super exciting, but I still think we're a way out from like general purpose. … It's just kind of a cool sci-fi world that we're living in right now. 

 

How would you advise students to prepare to enter the labor market given the dislocations that are happening currently? 

For non-information workers, we are bringing AI to manufacturing. People in that space are asking the same question, "Are you like automating people's jobs away?" And I think the answer to that, our stance, is very much, “no.”

“If you're going to school at U. of I., and you're going into the workforce, just make sure that you're literate with the AI tooling.”
— Devin Bhushan

I've never met a manufacturing company that is running at like perfect efficiency, and you know if they could just do it faster, they'd stop working the rest of the time. No, they all have backlogs. I think that paradigm applies to most work, because if you created more time to do things, if you just did things faster, you would do more. 

… If you're going to school at U. of I., and you're going to go into the workforce, just make sure that you're literate with the AI tooling. It's such an important piece of every interview process we have now. … because everyone could be more productive and faster, and it just makes us look better. So, I would orient towards that. 

 

For the students who participated in Cozad who are perhaps less than enthused about the idea that they didn't make it to the finals, what would what would you want to tell them given what your journey has been since your participation in Cozad?

It's all about the reps, right? Just get more at-bats, keep going, keep doing it, and maximize shots on goal. That's been my observation on most things in my career. Looking back at 2013 Cozad there were a lot of great things I learned from it, and there were still missing pieces that I needed as a founder that came later. Some of that I picked up by working at a big company like Yahoo or Splunk. And some of that I picked up by other failed startup ideas I had along the way. Not failed, but I didn't end up pursuing full-time. And then Squint happened, and it just all kind of fell into place. But I didn't know that before. So, keep going, keep maximizing chances.