Podcast #224 Accelerating Plant Autonomy

 
TRANSCRIPT
 

Ken: Good day, and welcome to episode 224 of our Momenta Digital Thread podcast series. Today, I'm pleased to host Dominic Gallello, former CEO of Symphony Industrial and a special advisor to DARPA. Dominic is a technology leader with a deep background in vertical software solutions. He has a long track record of building great products, driving intense customer focus, and building global brands. His products have won over 50 awards, including two R&D 100s, considered the Oscars of innovation, and a company-level Best Work Culture Timmy award. His products have delivered several billion dollars in revenue. He was named a SaaS Top 50 CEO, an honor with an Achievement in Management Gold Award for Advancing Industry 4.0 realization. Most recently, he has led three successful public and private software companies, resulting in over $1.3 billion in exit value and a 300% plus average increase in value. Earlier in his career, Dominic held executive vice president positions leading product development at Autodesk in Macromedia. He was President of Intergraph Japan and Director of Asia Pacific, based in Tokyo and Beijing. He also serves as an adviser to the Defense Advanced Research Projects Agency or DARPA, where he contributes to commercialization efforts. A hands-on philanthropist, he funds and oversees the development of orphanages that have provided a nurturing home and solid education to nearly a thousand children. An avid long-distance swimmer, he has completed many challenging races, including Asia to Europe, the Arctic Circle crossing Sicily to Calabria, Capri to Naples, and the Golden Gate to Bay Bridge races. He holds both BS and MBA degrees from Monmouth University. So Dominic, welcome to our Digital Thread podcast.

[00:02:35]

Dominic: Thank you, Ken. It's a delight to be here.

[00:02:36]

Ken: It's a delight to have you. You gave me a lot in terms of your background and even some of the research I did on you. But what a phenomenal background, and I'm looking forward to an equally phenomenal conversation there. We call this the Digital Thread podcast, and we'd like to think about it as one's digital thread: what led them to where they're now. What would you consider to be your digital thread?

[00:03:00]

Dominic: Ken, my Digital Thread is grabbing onto and harnessing technology shifts for me. I've experienced many technology shifts, from minicomputers to workstations to PCs, multithreaded ships, and, most recently, machine learning. It's been around since the '50s, but now it's possible. Of course, you can't ignore the large language models going on. In Autodesk days, a reseller once told me, "Dominic, there's margin in mystery." When you're at the forefront of these new technologies and mysteries, you can significantly help your customers.

[00:03:41]

Ken: Now, that is a notable quotable. I love that "margin and mystery." Indeed, you've lived through quite a few waves of technology shifts that you have been able to help grow. I looked at your trajectory, and even in your early career, you went for President roles at Intergraph and EVP at Autodesk in Macromedia. If you had to boil this early time down into three insights relative to vertical software solutions, especially for industry, what would they be?

[00:04:07]

Dominic: Yeah, I would say that it's probably one important thing instead of three. If I think back on my career, I would say a self-absorbing user interface that solves customer problems better than anything. The customers have told me over the years, they say things like, "Dominic, I learned your software when I was in school, and it's made my career." It's a useful user interface and tool that customers can get passionate about. When they get passionate about it, they tell other people, so really, the focus on what I would call self-absorbing UX is very important.

[00:04:50]

Ken: Interesting. The only one you're missing then on your background would be Apple, which certainly makes a lot of claims to that. But in some sense, since you were with DARPA, you were at the very genesis of what made Apple special in terms of their windowing interface.

[00:05:16]

Dominic: That's right. I've seen the genesis of those things and why some were created, like at Xerox PARC, and a lot of that is that of necessity, right? When you have four printers and only one Window on your computer, smart engineers say, "Well, why don't I make four Windows on a computer to launch for print jobs at one time?"

[00:05:38]

Ken: Most recently, you led three successful public and private software companies, as you said, resulting in over 1.3 billion in exit value. GRAPHISOFT, with which you sold to Nemetschek, MSC to Hexagon, which is the deal we know well at Momenta, and more recently, Erecruit to Bullhorn. To what do you attribute your ability to find and develop such winning companies and help them have such great exits?

[00:06:03]

Dominic: Well, as far as the finding, I think I get found, and when you build a track record, people look for the management that can do it. But again, it's focused on the customer. If you focus on the customer, you have a few things to do. Number one is creating a vision for what customers want and don't know they want. That's equally important, maybe more important. Look at the anthropology of what they do, then try to figure out how to do it better. The second thing is a team. My whole career has been about people, putting the right skills in place, and ensuring you have all the different heterogeneous skills in one team. Then, the third is execution. I focus a lot on execution, and you must combine all three things. If you do, generally, great things happen. I've built at least four next-generation systems. It's interesting- customers want to follow. They need to know you're building for the future, not just making the products you have today work better, but creating something for the future because they want to be let.

[00:07:17]

Ken: It's interesting, the way you phrase this, you execute with quality in the market, in some sense, find your company. I almost would have expected you to talk a bit about your negotiation or transactions expertise and such. But truly, you focus on the operational execution, which is great. I've always looked at it that way. If you do a good job, the market and people will find you. Good companies, great companies- bought, not sold.

[00:07:45]

Dominic:, Ken, if you look at GRAPHISOFT, it is a 25-year-old company. They had no interest in selling their company. MSC software was over 50 years old, and they had no interest in selling their company. I had no interest in selling the company. But if you build and create a fantastic brand, reputation, and market, you become very attractive to inquisitive strategics.

[00:08:08]

Ken: If there was a way of productizing this or commercializing that approach you've done, your most recent effort as CEO of Symphony AI Industrial would, in my mind, fit that. Right? For those who know Symphony AI and its overall model. You consider yourselves an innovator and industrial insight accelerating autonomous plant operation. Tell me, what inspired you to join the company?

[00:08:34]

Dominic: Actually, Symphony as an investment company- it's my third Symphony company, and since my first two companies, the value creation, we sold for over a billion, so the value creation was 700 million. I guess my investors were pretty interested in me running the industrial company; number one and number two, I have a manufacturing background. It's a perfect fit.

[00:08:59]

Ken: Symphony AI Industrial describes itself as 'connecting tens of thousands of assets and workflows in manufacturing plants globally, and processing billions of data points daily, pushing new plateaus in operational intelligence.' What would you consider some of your notable wins during your time there?

[00:09:15]

Dominic: Well, there's so many, but I would say it's always about the future, and so if I look at, let's say, a Belgian company it's a Japanese company, Nippon Gases. They wanted to look beyond typical predictive maintenance with vibration and oil samples and those types of things. They had tons of data in their history; it was a pity that it was just sitting there and not being used. They picked their critical assets and are finishing rolling out to 18 plants. They had great success because what happened was, they started as a maintenance tool, but it became a process automation tool because of putting the technology right next to the DCS, the control room operator, because of the discoverability of leading indicators of what's going on was much better than what their DCS can provide and much faster. These customers want to go beyond typical vibration and oil samples and infrared and use all the process data, which is a much earlier indicator. I think another company that is astounding in the manufacturing execution is Cargill because they're running our software in over 400 plants. As we know, many projects have a pilot plant, and after three years, you've got eight plants implemented. They've over 400 plants, so having a scalable platform that can run across a fleet worldwide is important. There are many examples, but we're blessed with those customers.

[00:10:55]

Ken: Blessed is right. For the audience, can you help us understand a bit about the scope and scale of Symphony AI as an industrial business?

[00:11:03]

Dominic: The scope is- first, what's a global business? It focuses on these three areas: the assets, all the machinery running the plants, and the execution system. You're talking about materials, process, and quality. Then, finally, the people. The connected worker solution provides bite-sized work instructions to the operators, especially to the desk-less worker running around the plant. I would say it's a very broad, very global business, and obviously, a growing business.

[00:11:40]

Ken: You mentioned Cargill earlier and 400 plants. That's quite an accomplishment as we're certainly earlier smaller-stage investors at Momenta. Still, one of our biggest challenges in some of our investee companies is simply getting beyond what they call 'pilot purgatory.' You can get one or two plants, but scaling that many times can be difficult. Congrats to you guys for having done that so well. I'm curious: how do you know when an organization is ready to adopt your solution? Were some of the best practices you saw in them realizing that potential value?

[00:12:15]

Dominic: Okay, let me take it one at a time. If I think about this whole world of AI and predictive maintenance, I think three things must be in place. Number one, they have to have had some critical failures, and they're scratching their head, saying, "Vibration oil samples every six months aren't enough." Second, they have to have someone who wants to drive this process. Then, I think the third thing is data, especially in machine learning. By the way, I'm kind of- I don't call it AI, I call it 'multivariate analysis,' because what you're doing is taking temperature, pressure, amperage, all of those types of things, and putting it in a deep learning model to try to figure out what's going on and what will go on in the future. It's the case that data becomes very important.

I know you've invested in a great company in Aperio, and we found what Aperio found out. I tell all my customers, "I will tell you that as we start this project, your data is not as good as you think." Then, they laugh and say, "No, no. We've instrumented well." What happens when your data flatlines? What happens when it spikes? Those types of things. We had to build not only an asset performance system management system for the machine and the asset but also one for the sensors to take that sensor data and make sense of it in case there's a gap.

You don't want to have to retrain models; we go back and impute. It's almost like we do a skin graft to keep it running. Definitely in predictive maintenance to take the next step as a huge unknown if I'm a maintenance engineer, and so critical plant failures are kind of a good kick in the butt in terms of getting going. On the manufacturing execution side, that's very different. By the way, many companies are still running their manufacturing execution on Excel and whiteboards with stand-up meetings every morning. There again, as I mentioned before, you're looking at materials, you're looking at the process, and you're looking at quality steps. I would say critical to success is to pick one. Pick one of those three areas, start getting that right because you can get it right, add tremendous value to your company, and then move on. I had a $4 billion company tell me that we have all kinds of quality problems, but we run on Excel for the operators regarding how they assemble and what goes first and what goes second. It's created a lot of quality problems. It's a perfect example of starting with that one task, getting it right, and then expanding from there.

[00:15:04]

Ken: I love it. Yeah, they were depending on enterprise software called Excel. I see Symphony discussing accelerating autonomy for every plant in the world regarding how they describe themselves. Now, you know we're close to Rockwell, and they've coined the term automation to autonomy, which is also interesting. Let me ask: what does autonomy mean to you, especially relative to industrial operations?

[00:15:31]

Dominic: I would say that accelerating autonomy is a moonshot goal. That's kind of what that big thing that- you're never gonna get there. I like to bring it down towards instead of an autonomous plant, an adaptive plant. A plant that can react much faster to operating conditions, and at the end of the day, what are you trying to do? You get more production. If I look at people in Saudi Arabia, if I look at modern gold, what they were doing is interesting. They were changing their set points on their control system every 21 days on their ball mill, crushing the rock that produced the gold. Well, 21 days seems like a long time; why were they doing that every 21 days? Because that's when they would do their maintenance. They basically open up the machine, look at the line or where there are other operating conditions, and then change the set points on the control system. Now, with multivariate analysis and all of that data coming in its well-instrumented systems, we can do setpoint advisories for them every 30- well, we do every 30 minutes. But that's true; you don't change a line every 30 minutes. But there, I would say updating their CIP points roughly every four to five days.

Now, what's the result? They were already running these plants, crushing rock, and turning to gold at 110% capacity. They got over 1% more throughput. What that means, for one, mine alone is 4 million bucks. $4 million in terms of grams of gold that they can get out of it. Again, adaptive plan much more frequently. The amount of data, by the way, is so large that a human cannot make these decisions and make these decisions that fast. That's what a computer can do, and that's what multivariate analysis can do.

[00:17:27]

Ken: We had a conversation recently. I'm part of the Industry 5.0 initiative in Europe, the EU Commission, and we were discussing some of the elements, what they call human centricity, and we came up with this chart where we talked about automation and augmentation. Think of things like exoskeletons as an example, helping workers lift. Then, we also mentioned adaptive and autonomy and looked at these things in a spectrum. Interestingly, you introduced adapting it pretty quickly from the autonomy perspective. There is a spectrum there, and we're rediscovering even the human workers in many of these processes. But a lot of the adaption we see happening is based on AI. We've invested in several companies that do process automation, as you say, multivariate analysis. I guess would have been hard to say we went Symphony AI versus Symphony Multivariate Analysis. I can understand why you didn't choose that route; it wasn't as good of a marketing one. But it's interesting to see all these converging. That's why Cyril Perducat, CTO over at Rockwell, captures it all when he says automation to autonomy and how these pieces come together. But GE Digital used to have this phrase of, 'The power of 1% and 2%,' you probably remember at the beginning of GE Digital. That still captures what I think is the biggest payback in many of these cases. As you say, it results in $4 million extra gold, for example, probably for less than a 1 to 2% change in the input variables.

[00:18:56]

Dominic: Ken, if I can just interject that- you mentioned people, and there's a lot of confusion around virtual reality and augmented reality and kind of the scale and where it belongs, those types of things. We've always taken a position with augmented reality, almost like a rearview mirror, right? Your hands need to be free, and your eyes need to be free. But if I need to, I can look at work instructions that maybe- not only work instructions but also quality instructions. What I mean by that is it's fine. We have a large multinational company doing the lining inspections for all the health and safety stuff. Not only the work instructions that tell the operator what they're supposed to do, but the audit trail of what they did, and so that's augmenting the worker and helping and assisting the worker, but you still need the worker.

[00:19:58]

Ken: In fact, the idea of augmentation is beyond simply the intellectual augmentation that we think of or visual in the case of what you guys are doing from an AR/VR perspective. But now, getting more and more things like cobots is an example, right? Or how do I reinforce muscle lifting capability so that you can have anybody work on a factory floor, regardless of their genetic muscle mass, for example? It's interesting what's happening around human centricity and how much this is becoming almost automation plus humans and using these for what they do best. It's fun to see how this converges. This brings us to one of the funniest things you're doing. You're a senior commercial advisor for DARPA. Again, Defense Advanced Research Projects Agency. This is the granddaddy of all advanced research and projects for anybody who grew up in the US and the Bay Area. What are some of the projects that you're advising on these days?

[00:20:59]

Dominic: First, let me say that DARPA is an amazing organization. $4 billion a year goes into investment and research, and DARPA invented the internet, GPS, autonomous vehicles, and many technologies. The whole drone technology grew out of DARPA, 30 years ago. It's an amazing organization. What I do in the projects that I do, I'm kind of a venture capitalist board member for DARPA because the DARPA funds a lot of research. DARPA doesn't have its R&D center; it funds universities, funds private companies, all of those types of things, and so I help DARPA make commercial decisions on, "Who do we want to fund?" and then act as a board member to commercialize the projects so they don't die. The reality is that there are some fantastic projects and then some that died- moonshot projects and some that died. To answer your question, what I'm involved in is a wide range of things from internet security, software security, AI for defense purposes- because everything we do at DARPA is to help the battlefield warrior, and then finally, a very, very large one I would call DARPA hard project, which is a game changer. It's a top-secret project, which is a game changer for our national defense. The government can fund things that private companies just can't afford to fund. Being involved in those things is a delight to ensure they work, survive, and get commercialized.

[00:22:31]

Ken: Well, as you're headquartered in Silicon Valley, it's interesting because you have the best of large institutional government-style defense research. Then, of course, the ecosystem that is venture capital in the startups in the Bay Area, so you get the best of both, and groups like In-Q-Tel, who sometimes sit in the middle of there, right? To help in funding with them, or- what's the other one? Defense-led research and logistics agencies? Out of Moffett Field. I remember that one there, yeah. They also operate as a bit of a middle ground between the military and some startups. Pretty interesting space to be in. I know that you've recently left Symphony AI Industrial. I'm curious and sure our audience is wondering what's next for you.

[00:23:19]

Dominic: Well, first and foremost, I've got some pretty challenging long-distance swim races that I have to stay ahead of the sharks. What I'm doing with DARPA is fantastic, and I will likely run another company, so I'm always doing many different things. But I'm enjoying what I'm doing for the government with DARPA.

[00:23:39]

Ken: A good time to be at it as well, with a lot of the re-shoring work going on, and sadly, with some of the conflicts going on around the globe, very timely. It's funny. Ten years ago, I used to work at Lockheed out there, so I remember way back the military-industrial complex. It was certainly viewed as a hot space to be in. Then, we went through this where there would never be any more war, so we didn't need a lot of defense, i.e., speaking from a kind of Western global perspective. All of a sudden, our defense stocks are the ones that are bringing the most return in value these days, both in Europe and in North America. Ultimately, it's best never to rule out the need for such things, as Israel certainly has found out. In closing, where do you find your inspiration?

[00:24:26]

Dominic: For me, I'm a voracious reader, so I read a lot of books. Recently, as I started with DARPA, I read "The Pentagon's Brain," which is Annie Jacobsen's book, and she also wrote "Area 51." I read both of those books. By the way, during the '40s, '50s, and '60s, our government employees and those massive projects worked seven by 24. The most selfless people in the world weren't the private industry. The government was the private industry in those days. I just finished up Walter Isaacson's book on da Vinci. There's a lot of inspiration when you watch a guy like da Vinci, who went from basically building stage props, and that was a lot of his flying machine stuff in the early days to what he did. Books always inspire me, and I've always got one on the plane, and I'm always reading.

[00:25:14]

Ken: Excellent. Some good recommendations there, and I will certainly have to put them on my hopefully soon-to-come summer reading list as I'm sitting here in the snow of Switzerland. Already thinking about the beach, baby.

[00:25:27]

Dominic: Be careful. They're pretty thick books.

[00:25:30]

Ken: Dominic, thank you for sharing this time and insights with us today. It's been a great conversation.

[00:25:36]

Dominic: Ken, it was a delight. Thank you.

[00:25:38]

Ken: As well, I appreciate you taking the time. This has been Dominic Gallello, former CEO of Symphony AI Industrial and advisor to DARPA. Thank you for listening, and please join us for the next episode of our Digital Thread podcast series. We wish you a momentous 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]