Podcast #170 Quality Vision

TRANSCRIPT

Ken: This is Ken Forster, executive director at Momenta. Welcome to our Digital Thread podcast produced by, for, and about digital industry leaders. In this series of conversations, we capture insights from the best and brightest minds in the digital industry, their executives, entrepreneurs, advisors, and other thought leaders. What they have in common is like our team at Momenta. They are deep industry operators. We hope you find these podcasts informative, and as always, we welcome your comments and suggestions.

Good day, and welcome to episode 170 of our Momenta Digital Thread podcast series. Today I'm pleased to host Christopher Van Dyke, CEO of Overview AI, a company building reliable, adaptable computer vision systems for any factory. Momenta's a proud investor in Overview AI. With over 16 years of engineering and manufacturing expertise, Chris focuses on optimizing customers' manufacturing operations using artificial intelligence and computer vision to accelerate product development, reduce rework and improve yields. Since founding Overview in 2018, Chris has been instrumental in building a company and team dedicated to a disruptive positive impact on manufacturing for decades to come. Before Overview, Chris served for over eight years in several senior manufacturing engineering positions at Tesla. For Tesla's Model 3, he led the 80 person battery design team from the initial battery concept through product launch to the high-volume production. He also managed the infrastructure and equipment design for the first Gigafactory, including equipment for supporting battery cell manufacturing. Chris launched Tesla's electric vehicle supercharger program with more than 25,000 stations worldwide. Prior to Tesla, Chris was a senior engineer at H2Gen Innovations, during which time he co-authored multiple patents. Chris holds a Bachelor of Science degree in Mechanical Engineering from Stanford University and a Master of Science in Chemical Engineering from the University of Virginia. Chris, welcome to our Digital Thread podcast.

[00:02:15]

Christopher: Thank you. Thank you for having me.

[00:02:17]

Ken: Believe me, it's a pleasure and very exciting. As you know, we call this the Digital Thread podcast. And it’s about your journey, what we like to call the 'digital thread' if you will. What would you consider to be your digital thread? In other words, what is the one or more thematic threads that defined your digital industry journey?

[00:02:36]

Christopher: The main thread in my career has been problem solving, trying to create new things, and doing that in a way that enables the latest technology to impact the world. In almost every case, even though I focused on hardware and the physical world, the improvements are built from digital enablers including advancements in computing, controls, even AI, and some other forms of digital transformation that enable you to make sizable improvements in hardware or physical devices. I saw this a lot at H2Gen, also when I worked at Tesla, and I landed in the digital space because it's just so ripe for meaningful technology development. Software hardware interface has so many areas where you can create new and incredible things that have a widespread impact.

[00:03:33]

Ken: And we'll certainly talk about that impact in a minute. Coming out of school, you had an early start-up experience at H2Gen as you were a part of the team that grew this from 20 to 150 people. What early lessons did you learn from this start-up effort?

[00:03:49]

Christopher: H2Gen was a great place to start my career. It was a small company, but it was ambitious in its engineering. It tried to make major improvements in a large and technically impressive, but somewhat static, field of chemical processing. I learned how to engineer things, focus on improving them, and create new things. We redesigned components and equipment that hadn't changed in 20 or 25 years. There would be some enabling digital piece, a new piece of software, that could come in and help enable computing; and, if you tack that on to the old version it might make an incremental improvement. It wasn't that revolutionary, but if you understood the underlying physics of the piece of equipment you were trying to change at a deep level, the pragmatic constraints, and you understood the new tech, you could combine those into something way better than what you had before resulting in 10x improvement while reducing cost. We had a lot of that hands-on stuff at H2Gen, and I learned it there. Unfortunately, I also learned that small companies and start-ups are always on a knife's edge. If you had asked me in 2007 or 2008 if I thought that company would make it, I would’ve been confident. I felt like they had a good foothold on some interesting things, and it was going to start to grow. And then there were some out-of-left-field setbacks, and all of a sudden, it fell apart. That was a tough lesson as well.

[00:05:24]

Ken: It certainly sounds like it. They say that working at a start-up is probably the best real-world MBA you could have because it's never the stuff you can calculate and count on that ultimately affects the business. It's always those black swans, as they like to call it. Like over the last two years, COVID. You did go from a maybe not successful start-up to one that I think anybody would agree has been a great start-up, Tesla. You joined them in 2010, contributing to several large-scale engineering efforts. As we mentioned, the supercharger program, Gigafactory, and the Model 3 battery design. What was it like to work at Tesla at such a formative time? And what lessons did you learn from it?

[00:06:06]

Christopher: It was awesome. It was an exciting place from the first day even with under 500 people. It may feel like an inevitability now, but it didn't then. Living through that transition was powerful, from an underdog company to a fairly established company. I learned a lot, and if I had to reduce it down, it’s keeping the bar high on people. Tesla did that from the very beginning in a very intense way. That included the hiring practices and then letting people go. Positions would go unfilled for a long time, even in desperate times. If you couldn't find the right people, then you really shouldn’t bring them in. That had a huge ripple effect across the whole company. Having the motivating mission certainly would help during the tough times. You don't know what's going to happen, and that’s certainly a big part of what made Tesla both a great place to work and a success. There was this mentality that whatever you were working on, whatever task, if you were designing a charging cable or figuring out how to do the regen for recharging the battery while the car was slowing down, you had to approach it knowing that you could do it better than anyone else in the world. Maybe just as good as anyone else in the world, but really, you were shooting for better. If you weren't ready to do that, then that's what you needed to focus on; and it was improving your skills or pulling other people in from the rest of the company to do things the best they possibly could be done no matter how small the task. That was powerful and led to a lot of improvements across the company. And that's something I carried forward at Overview. We also approach our work that way. We look at the manufacturing quality and automated inspection. We're trying to build the best toolkit possible to help manufacturing engineers and operators solve the type of problems they're working on for our customers.

[00:08:14]

Ken: That sounds like a great experience. I think somebody I'd interviewed in the past has called it the "Elon factor" in some sense. Those cultural attributes you describe out of Tesla are like the ones people would assign to Elon's probably larger-than-life personality.

[00:08:29]

Christopher: Yeah. If the person at the top embodies the cultural points, then everyone else gravitates to that for sure.

[00:08:36]

Ken: Yeah, very sure. Let's talk about some of the key insights that led you to co-found Overview AI in 2018.

[00:08:44]

Christopher: I spent a lot of time in manufacturing, and manufacturing is challenging as there's a lot to be done. It's rich in problem-solving. It's very satisfying. It's a wonderful world to solve engineering problems. I don't know if that was an insight, but that's what drew me into starting a company in this space. And it's also underserved, especially given how big it is, and especially in software, as seen from the inside. We had amazing budgets and excellent engineers at Tesla. Still, we had tons of rudimentary problems, and I knew good software products could make a big difference. Then, in 2016, 2017, computer vision and deep learning improved a lot. It was a fantastic transformation there and became this powerful digital technology. That was making a lot of waves on the self-driving car side, but it wasn't anything we could access in manufacturing. It was new and hard, and people internal to Tesla were always working on it on the product side. Then the legacy companies in manufacturing had a little bit of it. It was complicated enough that it was hard to work on. That felt like this technology innovation would fill some of the needs in manufacturing. That always seemed like a great opportunity and made me excited to leave the comfort of Tesla and find the start-up Overview.

[00:10:11]

Ken: Yeah, I'm sure that was a tough decision to make, especially given how much you had contributed to Tesla now becoming the de facto success, if you will, in electric vehicles. As you know, Momenta has deep roots in manufacturing. We've invested in companies like Eigen Innovations, Litmus, Sight Machine, Thingworx. Many of them do support video-based solutions. What do you see as the opportunity space for visual input inspection? And how does your solution complement these more general-purpose platforms and solutions?

[00:10:43]

Christopher: We channel our days as engineers and operators in the factory. We're trying to make our software easy to use. Some of the general ones are great and powerful. Still, they're not as tailored to manufacturing so we try to simplify the work of the manufacturing engineer. We provide a sophisticated tool, but you don't need specialists to use it. We're trying to make it fast, affordable and low effort on the customer side. That'll allow the people to put in even more systems that can have compounding results. We also really try to make the solution powerful but incremental. You don't need a major overhaul to take advantage of it; you can start with one use case and get a lot of value out of it. Then you can build up to major improvements. You don't have to get rid of anything or do any big, scary initiative. That's another distinguishing feature for us.

[00:11:39]

Ken: What stands out for me is this ease of provisioning; it takes a couple of days to get systems like this set up. And way back in my discrete manufacturing days at Philip Morris, of course, very high-speed manufacturing when you think of tobacco products, particularly cigarettes, as some of these machines produce something like 10 to 12,000 per minute. Every one of those has to be inspected. Those video systems are quite complex to set up and infinitely valuable in improving quality overall. Tell me about some of the early use cases and wins that Overview has experienced.

[00:12:18]

Christopher: Our software can spot defects on all kinds of parts. We are tailored to manufacturing, but within manufacturing, we're broad. We have good customers in medical devices, industrial food and automotive. The common themes, in cases where it's particularly good, it would’ve been parts that were previously batch inspected. They would take a couple off of the line and deeply inspect them. But not every part would get inspected, and even that was a little error-prone. We can put our systems in, and you can inspect every part, and catch every defect. You track all the data and have an interesting insight into how often things fail. We have a couple of examples where that's worked on all axes, and it's dramatically improved the component that the customer provides. We have a nice ability to fit with mid-sized manufacturers that don't have massive engineering teams but still want and can benefit from some of this cutting-edge technology, especially in the company's early days. That's a nice, natural fit. We’ve worked with some larger companies where we'll take a particular problem outside the toolset that the companies are used to working with, such as the work of the in-house experts within the commercially available systems, to address challenging problems beyond that capability. We do a repeatable software product, and we also have an awesome team with deep expertise. We can use our product to do problem solving that previously didn't have a commercial solution for it. We've had some good success doing some of those projects.

[00:13:59]

Ken: How do you know when an organization is ready to adopt your solution? And probably along the same lines, what best practices have you seen in helping that company realize that potential value?

[00:14:13]

Christopher: If they have a quality pain point, then they have to be making something in the medium to high volume. If they're doing that, there's usually a fit somewhere. It's sometimes when defects can't get caught until much later in the process, where they've already cost the company a lot of money, one of our systems can be put in to catch the defects sooner, avoid an expensive manual inspection or the worst case where defects make it to the customer. If companies have those situations and are willing to try something new, it could be a great fit. It works when the company is excited about the potential of new technology and is willing to integrate it into their system. A lot of companies are ready for this now.

[00:15:02]

Ken: Let me ask you to put your prognosticator hat on. What do you see as the greatest opportunity for manufacturing, given the background in your company, in the next five years?

[00:15:15]

Christopher: I'd say the word in that question that makes it a pretty hard question is 'greatest.' There's a lot of opportunities in manufacturing. As I was saying, seeing it from the inside, there are many areas where things can be improved. And then, think about it from a long arc; eventually, all factories will be fully autonomous or very close to it. I think that's as inevitable as a lot of technology arcs. But you can walk into plenty of factories today, and you can see they aren't even close. Still, that's the direction they're moving in; it just highlights how big the opportunities are with technologies like ours, where you have AI driving some new capabilities. I believe those are big opportunities in manufacturing, but I also really think 3D printing is great, and there's some amazing robotics now that can do dexterous and sophisticated things. And I continue to believe there are just tons of places where manufacturing will evolve and improve.

[00:16:15]

Ken: A common element, I hear, is mass customization, and certainly you would have seen this at Tesla where every individual item that you're producing is somehow customized to a customer's order. Given that your system is so easy to provision and effectively learn, I would think that's also a key trait that you guys will help contribute toward and help companies be more efficient in this idea of mass customization.

[00:16:41]

Christopher: That is a challenge. But obviously, there is value too. Our technology and the way the underlying AI defect detection can generalize does make it easier to operate than in some traditional systems. Those are challenges that everyone will face.

[00:17:04]

Ken: I noted that you guys were a Y Combinator graduate. And of course, they're quite famous as an accelerator. How did this program help you define and scale your company?

[00:17:14]

Christopher: We enjoyed YC; it was great for us. We're a very technical team, the founders all being engineers, and we feel very comfortable with hard technical problems. YC helped us understand the commercial side; we didn't know the term product-market fit before going there and didn't know how to find any sales leads. Getting a crash course in how to run a start-up or understanding the non-technical side of a start-up was valuable for us. YC does a good job of explaining that to people.

[00:17:46]

Ken: I fully agree. We do like their graduates. And they've been a great program, along with Alchemy. In closing, I'm curious, where do you find your inspiration, Chris?

[00:17:57]

Christopher: I think it sounds cheesy or something, but I get a lot of inspiration from working in the technology field and living where I do. I live in San Francisco in the Bay Area, and it's been getting a bad rap recently. I've lived here for 15 years, and I'd meet people that early on would tell me about some idea, then I'd bump into them a year or two later, and they had started a company. I’d see them again after that and realize the company was big, or even a company I'd heard of, and it happened again and again. I lived through all the Tesla growth that is so inspiring. It pushes you to create things and gives you confidence that you can succeed if you take risks. I've been inspired by working in this field, in this moment in time, by the peers and community around me.

[00:18:49]

Ken: Mike Dolbec, who leads the US side of our ventures, lives in Menlo Park. He describes it as the modern Camelot in terms of the creativity and everything that is Renaissance building there in the Bay Area, and I suspect despite some of the recent challenges that will continue. Chris, thank you for sharing this time and insights with us today.

[00:19:12]

Christopher: No problem, it was great.

[00:19:14]

Ken: I appreciate it as well; what a great conversation. This has been Christopher Van Dyke, co-founder and CEO of Overview AI, the next-generation computer vision technology for complex manufacturing. Thank you for listening, and please join us next week for the next episode of our Digital Thread podcast series. Thank you, and have a great day. You've been listening to the Momenta Digital Thread podcast series. We hope you've enjoyed the discussion. And as always, we welcome your comments and suggestions. Please check our website at momenta.one for archived versions of podcasts, as well as resources to help with your digital industry journey. Thank you for listening.

[The End]