Logan Jones, general manager and president of SparkCognition Government Systems, joins host Ken Harbaugh to talk about ethics, dual use, and artificial intelligence in the public sector.
Before leading SparkCognition Government Systems, Logan was Vice President and founding member of Boeing HorizonX, the largest aerospace venture fund and innovation group. He was recently nominated by FedScoop as one of the best bosses in federal IT. Learn more about SparkCognition Government Systems at www.sparkgov.ai and on Twitter at @sparkgov.
ACME General: My guest today is Logan Jones, president of SparkCognition Government Systems. The first full spectrum, artificial intelligence company, focused entirely on national defense. Logan, welcome to Accelerate Defense.
Logan Jones: Hey Ken, thanks for having me. I’m thrilled to be here.
ACME General: So the thing that sets SGS apart, and it’s right there on your masthead, is that sole focus on the public sector. And I want you to share your company’s thinking behind that strategy because where I sit, everything is about dual use. It’s a mantra for the companies I work with and not in a bad way. It’s a diversified bet on a bigger market, but you have been extremely successful swimming against that current. Why?
LJ: Well, I would back up, Ken actually to some of the heritage and history starting from when SparkCognition, our parent company was founded eight or nine years ago. My connection to the company was, I had led the venture fund at Boeing called HorizonX. SparkCognition was our third investment. And we’re really interested in understanding how a company like Boeing, who operates at massive scale could tap into this emerging ecosystem of players who are driving technology. Artificial intelligence, machine learning was one of those classes of technologies that we were interested in. You can imagine, and you’ve probably done it yourself, but we did a bit of an evaluation in the market. And what we saw is roughly a couple categories of companies. And this gets right at the heart of your dual use question.
Number one is classic companies who claim to be doing machine learning, but they just were using it as a way to fundraise, put that one to the side. The second class was a group of companies really embracing the power of ML for applications in different solutions. Yet it was in a domain or an industry that we as an A&D company couldn’t understand, right? It operated for media or consumers and it wasn’t industrial in nature. Then we did find a class of companies really driving AI and ML into industrial applications. So at scale, security at mind, auditability at the heart of that and solving some really complex, hard problems, but fast forward. So we make the investment, SparkCognition was also very focused on driving applications within DOD and national security, but it was just like you mentioned, it was through this lens of dual use.
We deliver and deploy in commercial markets and we poured it over into, in this case, military application sets. The problem comes down to a couple of categories which led to our focus, our corporate strategy. Number one is infrastructure matters. So your ability to build a company to house appropriate data sets relevant to the government to work in partnership and collaboration with the government, things like time keeping, accounting, it matters. If you want true and full access to this customer community to understand and appreciate the mission, you have to build the company that way. Number two is really around the people.
So getting a group of operators who understand and have passion for the mission, who can sit arm and arm hand in hand with these customers and drive customized deployments, that comes with an exclusive focus on national security and defense. And then finally, the last one is dual use, the underlying core of technology. Sure it’s applicable in all markets, but how it’s applied that last mile truly matters. And I think you start to dilute your focus and effectiveness if you’re trying to serve both markets at the same time. So that’s what led us to sole focus on defense and national security as our corporate strategy.
ACME General: You brought up staffing, which has come up again and again, in our conversations with industry leaders and experts in this field. We had a fantastic conversation with Chris Lynch, which defied a lot of my and others’ preconceptions about the difficulty of drawing talent to this ecosystem. What has your experience been drawing the best and brightest to a company that works exclusively with defense? The conventional wisdom is that techies keep this stuff at arms length. We had the drama at Google and other big companies, the insurrections against government contracts. What have you seen?
LJ: Yeah. And Ken, I listened to the interview, the episode with Chris and it was phenomenal. People should go listen to it, because a lot of what Chris talks about is what I believe to be true as well. Speaking for my own history, which I think is applicable here. I, from a very early age was interested in military. I never served, but this, I feel like is a way to serve without serving overseas in a way. So what do I mean? When I grew up, my grandfather was a master sergeant, he was a maintenance chief for the B-17 at World War II. And we used to go to air shows, we’d talk about a lot and I’d always talk about, “Oh, pilots oh, grandpa you’re so cool. You’re a pilot.”:
And he’d say, “No, I wasn’t a pilot. My eyesight was terrible, but I had an important job. Don’t minimize it. I kept the thing flying.” And the aircraft that he was responsible for was called Battle Wagon and go on bond drives. But I have a photo album. I have certificates of when he graduated at Maintenance Tech School. It was what led me to Boeing when I worked at Boeing and this passion for the mission and purpose to serve something greater than yourself, I think it spans much farther than people give it credit for. Over the past 15 or 20 years. I do think Ken you’re right. A lot of people have moved into commercial markets and been focused on different applications. But over the last year, especially accelerating since Russia invaded Ukraine, I think people have come to realize that the world may not be as safe as everybody thought it was that there’s bad actors out there.
And so from a passion and mission standpoint, what I’ve seen as an uptick in people who just want to serve, be a participant and a driver in both technology, innovation and application at the same time while serving a cause much greater than themselves, right? It’s the newspaper test. Can you pick up the newspaper and see something that you’ve worked on, an impact that you’ve had in the world beyond some narrow applications? So that’s what I’m seeing. I’m seeing it in SparkCognition Government Systems, but I’m also seeing it in new entrance within the space, new emerging startups who are focused on DOD.
ACME General: I definitely want to talk about that. Especially you’re raising the issue of Ukraine, but as a aviation/history geek, I just looked up the Battle Wagon, the aircraft your grandfather worked on and man, it’s got a story. I can’t count the number of mission stickers on the nose. I’m going to read up about that plane. My grandfather was a B-17 pilot in the Pacific and that’s a hell of a story in its own right. He came back filled with shrapnel from a bombing run over New Ireland. But thanks for giving me some reading.
LJ: Yeah. It’s just like your history, Ken, it’s maybe something too that I aspire to and that how many bombing runs, how many missions did that generation perform? And what is it that we can do to perform a similar mission set different obviously in these days? But Battle Wagon is a very special asset of my family, if you will.
ACME General: Yeah. Well, one of those contributions that SparkCognition is contributing enormously to is Joint Readiness Command and Control. I read the white paper, and I’m wondering if I imagine that was conceived and written before the Russia invasion, but the lessons from that must be enormously applicable in a negative sense, given the total breakdown during the initial phase of the invasion, how were you incorporating immediate current events and the geopolitical landscape into your thinking and your input on white papers like that?
LJ: Yeah, thanks for bringing it up Ken. So Joint Readiness Command and Control as we call, JRC2 two. A colleague of mine, David Masar and I wrote that right at the outset of creating the company. And the idea is that there’s a significant amount of focus in DOD and rightly so, around this sensor to shooter integration and being able to seamlessly connect at the day of the battle. But what we saw was it left a lot on the table in terms of the many weeks and months to prepare for a battle and integrating the preparation, the logistics, the storing, the supply chain into how you actually provide mission effect is something that DOD has an opportunity to capture and grasp. What we saw and are seeing in terms of Russia Ukraine is that it comes down to logistics.
The boring, sleepy, old topic of logistics matters in 2022, and your ability to prepare and plan for a fight, and then constantly update that in near real time, based on real world conditions, truly matters in how you can carry out an operation. And so it’s just as relevant as it was nearly two years ago when we wrote it. What has changed though, is activities within the department around data readiness. As a concept, it makes a lot of sense what makes it difficult to apply is the data exists in silos. It exists in both commercial environments and DOD owned environments. And that data integration issue has been one of the challenges or barriers to adoption.
ACME General: I think it was general Omar Bradley, who said, “Amateurs talk strategy, professionals talk logistics.”
LJ: Yes he was.
ACME General: Let’s talk about the intersection of AI and logistics. Your promo video has this great quote. “AI can peer through and otherwise opaque future and help us make out the contours of what lies ahead.” I think most of us appreciate the ability of AI to accelerate decision making and amplified perception, greater range of action. Those are our phrases I’ve taken from SparkCognition, but talk about its forecasting abilities, predictive maintenance, that thing. That’s a little less intuitive than helping the immediate decision making ability of the war fighter.
LJ: Yeah, it is. Those categories of what AI is used for, Ken, I think really matters. So we always break it down into AI can be used as a force multiplier. It can be used to improve asset intelligence, and it can be used as a tool to help optimize decision making. The power comes in when it’s all three effectively in one. One of the things that we’re quick to talk about in terms of artificial intelligence and the solutions that it’s a part of, is that it’s all about the outcome. These users don’t necessarily care that it’s AI enabled workflows, or that there’s a robust data strategy that underpins it. What they care about is that it’s helping them make better decisions or do their job in a more effective or efficient way.
So in terms of predictions, there’s a number of expert systems out there that are making predictions around supply and logistics every day. Where we think AI and ML has a right to play is the fact that these data sets are so large with so many variables so many factors that play into future potential outcomes that it’s moving beyond what expert systems are able to achieve on their own merits. This is one of the lessons that we’ve learned in our commercial heritage is that applications in fraud detection in banking, for example, or approving loans or insurance. All of those lessons in terms of optimizing a decision outcome are relevant in terms of logistics and supply chain. And that’s what we’re applying to improve predictions within this marketplace. Now, what makes it very difficult is a lot of the supply data that we’ve seen through different government organizations and agencies is really spotty.
It’s tough. You get maybe one widget ordered one time, or maybe infrequently at best. And it’s really difficult to make a prediction based on that. So the way that we’re thinking about and applying decision making is not just leveraging that single data set around, “Here’s my historical demand now forecast what the future could look like.” But it’s around incorporating the context that drives that demand signal.
So what does that mean? It means where the assets are being used. It means information and context around the operators themselves is a pilot, a new pilot is the pilot an experience pilot. What forces are acting upon the machine itself and on and on and on. That’s how you improve the robustness of prediction. And that’s where AI and ML really stands ahead of the field.
ACME General: You’ve got a couple other categories on your Solutions page that I was trying to figure out. And they’re really interesting to me. There’s the typical enabling of operational insights and situational awareness. I think we all get that, but explain, capture, retain, and operationalized tribal knowledge, that’s pretty novel and upscale and augment team members through prescriptive recommendations. What are you doing there?
LJ: Yeah, I’m happy to. Actually, it goes back to my grandpa in a way, right? When I’d listened to his stories about maintain an aircraft he was experienced, he was a crew chief. So he ran a number of maintainers. What does that mean? He would’ve been, what’s called level nine. And at a level nine, when you have that level of experience the maintenance manual of the fault isolation, manual, whatever it might be, may recommend based on these criteria. Here’s what the problem is with an aircraft, but people like my grandpa and there’s many more nine levels out there, they’d say, you know what? I got this, the hairs on the back of my neck stand up. And I feel it in my bones that it’s not that it’s this problem over here and the way that they troubleshoot based on symptoms and actually find the root cause is very unique.
And it takes many, many years of not just training, but doing. That tribal knowledge is locked up in this case in maintenance logs. So when they go in and they diagnose a problem, they run through a course of action, they write it up and then there’s a fix and then the aircraft lives, and you can see how the fix did at work or did it not.
So what we do in this case is we use a class of AI called natural language processing, that can cut through the idiosyncrasies of how a maintainer describes a certain event. One may call a symptom of vibration, another, a shutter. Another may describe it as some really loud noise. And it may all be the same thing, but natural language processing has a unique ability to cut through those nuances of human described events, and then it starts to correlate the way that those symptoms were described and goes back and ties it to previous examples, the fault isolation manual, how do you diagnose it? The job guide, how do you actually apply the fix? And then the tech pubs, what’s it look like in context of the broader system or environment. And it prescribes a course of action aligned with policy that the air force in this case has. So our solution is what’s called digital maintenance advisor, and it uses AI to help diagnose, troubleshoot, and then apply a solution to the aircraft maintenance event.
ACME General: That’s awesome. Logan, you were just nominated by FedScoop to the best bosses in federal IT. And I want to talk a little bit about leadership in this domain. We’ve touched on the ways you attract talent, but AI poses its own set of ethical challenges as does any emerging tech, but this one has the potential to disrupt, not just an industry, but civilization. I’ll leave it there for your initial reaction, but then I want to dive deep in a couple of areas.
LJ: Sure. Thanks, Ken. And first off, the nomination to an award like that is more a reflection of the team that we have in SparkCognition than it is about me. The only one who would disagree with that’s probably my mother, but it really is about the team that’s in SparkCognition. Now in terms of what we do and the ethical context around it. One is, it matters. We are a company that takes these conversations and incorporates in a decision making process. We have many engagements with our board around ethical use of AI. It’s a subject that we’re not going to hand wave away or move around without thinking about it, without forethought. At the same time, I tend to be of the mind of really two ways to look at it.
One is, if you look at all of the problems within this market set, there’s a significant portion of problems that are pretty mundane in nature, where ethics does not even come into the calculus. As an example that up-skilling a maintainer by unlocking the insights that are locked up in tribal knowledge. There’s not many ethical trip wires in that, this is not a smart munition. This is literally maintenance logs and maintenance manuals and helping somebody find information in a more appropriate way. How many different boring or mundane application sets exist like that across the department? It’s a huge fraction. If we would focus as an industry on solving those, there’s so much progress to be had where ethics, it’s not moving into that very dangerous area or gray area. Now the second category are the application sets where it starts, it does become a very real debate.
And what I believe is that we, as a nation, we, as an industry, should be first to solve those problems with the ethical framework and the value system that we have. I’ve heard it on this podcast before Ken, we have a framework as Western society. We need to be out there driving and innovating. Otherwise, the vacuum will be filled by our adversaries. And the way I look at it is we need to provide the leadership and first mover position in this marketplace. So that those decisions aren’t made for us in a way.
ACME General: Yeah, I think we’re moving beyond a loose framework. We don’t have the policy we need yet, but as part of the prep for this read the National Security Commission on AI’s final report, which has this statement of principle, which I think is really powerfully articulated. It says, “The American way of AI must reflect American values, including having the rule of law at its core. That’s a profound statement about an emerging tech that it needs to be tied to American values. I’m sure you’re familiar with that report.
LJ: I’m very familiar with it. The chairman of my board is secretary Bob Work, who was vice chair of that commission. And that values based is used in making does flow into everything that we do. So one of the other solution sets that we’re working on is around domain awareness, driving into and proving decision making, and when you have a solution that is perceiving the threat environment and then teeing up decisions for a human operator to make, ethics and values goes into everything that we’re doing. And I would also say Ken, that one thing I’m really proud of about this country and about this community is that it’s not left just to companies like SGS or me as the president of the company to drive this. These are conversations that we have with customers on a frequent basis about keeping it at the forefront of how this is deployed in application sets. And I’m glad you read the report. I would recommend anybody truly read the report. I think it’s 700 some odd pages. There is an executive summary out there.
ACME General: That’s what I read.
LJ: You read that. Okay. Well, it’s outstanding and it details a lot of the issues that we have around the talent base and government being able to scope out and acquire solutions enabled by AI. It talks about stem education. It talks about our acquisition systems. It talks about how we put pools of funding in place for early stage pilot programs. And it talks about ethics. So I would recommend everybody read it. It’s an outstanding report.
ACME General: Yeah. It is. You mentioned Bob Work the chair of your board. He said this about you, “Logan understands that this is a values competition. He really understands that everything starts at the ground floor, that the technologists who are developing AI applications have to fully embrace the idea of responsible AI.”
LJ: We do. You can’t just have a group of people who are working in national security and defense treated as I guess, benignly as a social media app in a way that may not be a popular statement, but the way that we are thinking about this and the gravity that we apply to the decision making process, matches the gravity of the mission of our customers. And this isn’t, again, this is not just SGS. What I’ve observed over the many years I’ve been working in this industry is that no company’s differentiating themselves on an issue like ethics and responsibility, is that this is a unifying theme across American industry, across this ecosystem. And it’s working its way into the applications that we’re deploying collectively into DOD. So, Ken we’re on the right track as an industry.
ACME General: Bob Work also talked about your focus on the relationship with the Warfighter and the criticality of building up that Warfighter Trust. We had a great conversation with captain Michael Shaw from task force 59. And I wanted to get your perspective on the role of software in enabling an initiative like that. All of the glitz is around the kit, right? The hardware, the robots, the drones, but none of it works without code. Talk about that.
LJ: Yeah. Well, Ken, it’s one of those, I’ve faced this, I think many of us have faced. This is how do you compel? What’s the attractiveness around the enabler, which is software intangible, oftentimes buying the scenes when everybody can see and touch and feel the hardware itself. It’s a difficult thing to do. I think the Commodore has done a very good job of yes, showcasing hardware and how it’s being adopted into the mission set. At the same time, working in parallel of applying and investing in the building blocks of software that actually makes the hardware even more effective. So it this virtuous cycle. I think task force 59 has done a really good job of that. The way that we’re thinking about this is we also invested in experimentation facility.
We call it HyperWorks. It’s just north of Austin, Texas. We know that operators within this segment must be able to see, touch and feel not just the software, but the entire solution. So we built out a 50 acre proving ground where we show how AI in the real world comes to life, how it operates and integrates with hardware itself. We believe in a future that is much brighter when you remain focused on software and the other side of that coin is hardware agnostic. You work well within the ecosystem of providers. So we have partners that we work with, through experimentation at HyperWorks companies like Raytheon, Boeing, and others frankly, around how AI can enable that to come to life. And Ken you said it, comes down to building the relationship, advancing the trust, and that trust is actually enabled by showing how it comes to life in the real world.
ACME General: You mentioned Raytheon and Boeing. Do you have any words of wisdom for the non-traditionals, the upstarts, the disruptors, trying to figure out a way to contribute within this ecosystem. You had a quote as SGS was emerging from COVID about government sticking with incumbents and the challenges for non-traditional entrance. What’s your prediction for the next few years and what can non-traditionals do to grease the skids?
LJ: Yeah, Ken. It’s a great question. When I was in a traditional, if you will, my job for the last four years of it was to make, in this case, it was Boeing, to help Boeing bridge to this emerging ecosystem of new entrants and leverage its resources to help those companies, but also effectively help its customer base. Tap into it in a natural way. What I’ve observed and I guess my advice would be, it comes down to my belief system in a way about this marketplace, about the topic of defense and national security. I believe that this has to be an ecosystem. This is not going to be a walled garden, totally vertically integrated solution that wins the day that helps to move the needle for national security.
I think that our customers and our cause is promoted best by working well within an ecosystem of providers. So what does that mean for emerging companies? I believe that you should look at your go-to-market in different ways. Two of those ways is build direct trusting relationships within the customers that you want to serve. You have to do that. You have to get to know the user community so that you have a ground level understanding of the problems that they aim to solve. Number two is also work with the OEMs, the primes, and the SIs as a way to accelerate your ability to help the end customer solve a problem. And if you have an open and clear mind about it and know that it will take time, I think it pays dividends, not just for your company, but it actually helps customers leverage the capabilities coming out of this community in a much more effective way than just trying to break down the acquisition system and disrupt if you will. Does that make sense, Ken?
ACME General: It does. I want to ask you to look out, let’s say 10 years, and if this ecosystem matures the way you hope it does, what are your brightest expectations? And I guess I have to also ask then what are your biggest fears? What should we be guarding against and feel free to factor in the ethical implications to that answer as well?
LJ: Sure. Well, actually, let’s start on the downside. The downside is status quo that we don’t actually change. That very exciting. And I think well-positioned initiatives like CDAO that’s really on great footing at this point. Don’t take hold. I don’t think we as a country and we, as a group of allies can afford current state applying into perpetuity, or maybe put another way taking what worked in the cold war and applying it to a future scenario. I don’t think it works. What I think success looks like Ken, I wake up every day focused on this. This is invigorating for many of us in industry is to think that the future will be brightened by a number of new entrants, a variety of emerging companies driving real innovation into the space. I’ll give you just one example to bring it to life a little bit.
And I have to caveat this. I’m a big fan of Christensen, The Innovator’s Dilemma, disruption theory, and there’s a segment out there around maintenance. Again, I’ll pick a mundane example, that is pretty easy to grasp. If you look at a GAO report from August, of 2020 is when I think it was. It starts to detail some of the problems within this community. They’ve completed 38 of 51. So that’s about 75% of maintenance periods late for aircraft carriers and submarines, 75%. That’s a combined total of 7,424 days of maintenance late in that period. If you added it all up. The four shipyards completed maintenance periods in average of 113 days late for aircraft carriers. Ken, is that acceptable?
So if we dug into why this is? One is yes, it’s an extremely complex problem. I got it. But it’s also a reflection of the status quo not working, the maintenance and repair business is a lucrative business. The incentives built into long term contracting, actually disincentivizes change. So again, what do I take as a bright future? I take a bright future as DOD really investing in new software based solutions in our case, aiming at fairly sleepy mundane and traditional problem sets, that change our mindset where we don’t think that 75% late is actually acceptable or a level of success that’ll help us in the future fight. And so how are we going to do that? It’s with a combination and many new entrants to this space competing and driving innovation through the incentives that exist in this industry. And so that’s what we’re applying for. I do see signs that it’s happening, but it’s going to take a number of years to make it so.
ACME General: Well, let’s end on that bright future note. Logan it’s been great having you. Thank you so much for joining us.
LJ: Thank you, Ken. It’s been a pleasure being on the podcast.
Find us in your favorite podcast app.
Find us in your favorite podcast app.