Brian Kenny: In the 1983 film WarGames, Matthew Broderick plays a 17-year-old computer whiz who unwittingly hacks into the US military defense system and triggers a global nuclear war simulation with a self-learning computer. It's one of many films that pokes and prods at the perils of military use of artificial intelligence. The theme resonates. A 2017 survey by The Guardian found that 72% of Americans fear that robots will take over our lives. Although the idea of killer robots grabs people's imagination, the focus of AI in the military is much more about providing robotic assistance on the battlefield, getting robots to do things considered too menial or dangerous for human beings. Technology that protects soldiers in the battlefield seems like a no-brainer, but at what price? Today we'll discuss Shield AI with case author, Mitch Weiss, and case protagonist, Brandon Tseng. I'm your host, Brian Kenny. And you're listening to Cold Call on the HBR Presents network.
Mitch Weiss studies digital transformation and innovation ecosystems. He wrote the book, We the Possibility: Harnessing Public Entrepreneurship to Solve Our Most Urgent Problems, a book about public entrepreneurship. Brandon Tseng is co-founder and chief operating officer of Shield AI. He's also a Navy SEAL and Harvard MBA. Thank you both for joining me today. It's great to have you both here.
Mitch Weiss: Thank you, Brian.
Brandon Tseng: Great. Thank you. Great to be here. Former Navy SEAL.
Brian Kenny: Well, I figured once a Navy SEAL, always a Navy SEAL. Right? That doesn't just go away.
Brandon Tseng: It is kind of like that. Yes.
Brian Kenny: Yes, so listen. We love having the protagonist on the call, Brandon, because we want to hear firsthand from you who experienced this and who's living the case to give us your perspective. Mitch, I thought this case was super interesting. It touches on so many important issues. And I think the topic of artificial intelligence is one that people hear a lot about and they, I teased it a little bit in that introduction, they probably don't fully understand it. And there's a lot of uncertainty about what it is and what its intentions are. So I think people will really benefit from hearing about what Shield AI is doing. Let me ask you to start, Mitch, by just sort of setting the case up a little bit and telling us what your Cold Call would be when you step into the classroom.
Mitch Weiss: Brandon Tseng and his co-founders are two years past their founding of San Diego-based Shield AI. They built an AI-enabled quadcopter that can self navigate buildings. The flying robot, it turns out, can outpace human war fighters in some tests by as many as five minutes suggesting it can also save military and civilian lives. Meanwhile, the US president has said we need advances in AI for defense. The secretary of defense and the past secretary of defense have also said so. They've even gone out to Silicon Valley and elsewhere to try to help make that true. And no wonder why, because the president of Russia, Vladimir Putin, has also said, "Whoever wins AI will win the world." And the Chinese president has said, "China must firmly seize the initiative in the race for AI." So tell me this, the US Department of Defense spends well more, than $600 billion a year. Why aren't they showing up at Shield AI with dump trucks full of money?
Brian Kenny: That's a good one. That's a very dramatic opening question. Must get the room going.
Mitch Weiss: It gets the room going. The topic, the case, the protagonist, all of it. Yes.
Brian Kenny: You know, you and I have had a chance to talk together a lot about the cases you've written. And we discussed one a few years ago about Toronto and the smart city and a lot of the questions about, I don't know, the creepiness factor of AI come up in that case. And here, I think, some of those same things can surface a little bit. Can you tell us a little bit about what prompted you to write the case and how does this fit with the kinds of things that you think about as a teacher and a scholar?
Mitch Weiss: Well, you mentioned that I spend all my time worrying about whether we can solve big public problems anymore. I have been studying this potential answer, which is, yes, we can indeed solve our public problems if we're more inventive, if we try new things that might only possibly work. But if they did, they'd be transformative. And I believe deeply if we're going to get there, get towards possibility, get towards possibility government, we have to have new ideas, we have to be able to try those new ideas, and we have to be able to scale those new ideas. And I believe that scaling, a big part of scaling, is actually being able to build companies that sell new ideas and new technologies to governments and to invest in those companies. And this was a perfect opportunity with Brandon and Shield AI to try to understand just how are companies building themselves to sell to government. What are the opportunities they see there? What are the challenges they face and how do they surmount them?
Brian Kenny: Awesome. So Brandon, let me turn to you for a second. It'd be great to hear a little bit more about your background beyond just the Navy SEAL. Tell us a little bit about the sort of journey that you've taken to get to this point.
Brandon Tseng: And so I've always had this fascination with engineering, with technology, but also very similar to Mitch, a strong sense of service and wanting to solve big problems, wanting to be part of something bigger than myself. I think that's what attracted me to Mitch's class. After the Navy, attended HBS, and at the same time concurrently founded Shield AI, founded alongside my brother, who's also a serial entrepreneur, also an engineer. And it was really this idea of how can we bring leading edge technology, leading edge autonomy and AI capabilities to a customer that needed it very badly, and at the same time, build a large organization, a large company that could marshal the resources, marshal the teams to make an impact for our customer and for the world.
Brian Kenny: So put a finer point on what Shield AI does. Can you just sort of describe what the product is and how it works?
Brandon Tseng: Sure. I think the easiest way to think of Shield AI is it is self-driving technology for the US military for aircraft in general. And that is at the heart of what we do. And if you've ever had the chance to ride in a self-driving car, it's a really magical experience. A lot of engineering, a lot of technology goes into that. And when I show people Shield AI's products, it's a really similar feeling. At the core of it, the computer takes in sensor information and really begins to understand what it is looking at, where it is in the world, and making decisions about how to navigate the world, and how to solve problems within the world. And so that's a very simple analogy to a very complex technology, but it doesn't operate that much different than human beings do. We have sensors. Our eyes, our ears, our nose, our sense of taste. And our brain is our computer that processes all this information. And that is really what is happening in the self-driving technology world with autonomous cars, but also where Shield AI operates at, which is autonomous aircraft.
Brian Kenny: And in a very different setting. So people think about autonomous cars and they think about like Tesla and, "Wouldn't that be cool?" But you guys are operating in a different theater and the case describes the notion of this being the kind of tool that could enter a building and sweep a building looking, I guess, for bad guys. And I'm guessing that... Obviously, you've got some firsthand experience having been a Navy SEAL. Can you just describe for our listeners what it's like for a person to do an exercise like that so they get a sense for why the Shield AI product is so important?
Brandon Tseng: It was interesting that you started off with this Hollywood example because I think a lot of people, when they think about the US military, think about these Hollywood examples. And then, you can go to, you name it, the action movie, where you have the protagonist clearing a building and shooting a gun in very close combat situations, and there being no sense of fear within the protagonist. I will just tell you personally, there is nothing more terrifying than going inside a building and clearing it when you know there are real bullets flying at both ends of the spectrum; from the protagonists or the antagonists. And it is why we decided to tackle that problem first because it is such a meaningful problem. The stakes are incredibly high. And it was something that I was familiar with in terms of being able to articulate that problem very succinctly to our team of engineers. I was Shield AI's first product manager to help them be able to understand and empathize with the customer.
Brian Kenny: So it certainly seems like the kind of product that people could get the behind. Mitch, I want to turn back to you for a second as we think about artificial intelligence in the broader picture of things. You know? What does that landscape look like? Where is the United States in the race, I guess, for supremacy on artificial intelligence?
Mitch Weiss: Perilous. There was a National Security Council on AI commission co-chaired by Eric Schmidt, led by many other capable Americans, which concluded last March that, "The US government is not prepared to defend the United States in the coming artificial intelligence era." That's a quote. And further, they went on to say that China possesses the might, the talent, and the ambition to surpass the US as the world's leader in the next decade if current trends do not change. So their conclusion at least is that we must change if we're not to fall behind with these technologies. And if we don't, the consequences could be devastating. There are things that we could do to change. They involve recruiting and training new talent. They involve investing in these technologies. They involve showing that there's a model for this use in a democratic society and democratic ways of using it. So there are things we can do to accelerate and improve our situation, but I would describe it as perilous at the moment.
Brian Kenny: And does the US have its own sort of R&D? If we focus maybe on the Department of Defense, do they have their own internal R&D resources that are looking at this? You know? Is it necessary for them to go out outside and look at companies like Shield AI to sort of help what they're doing?
Mitch Weiss: Well, absolutely. We have some of the most brilliant people on the planet working for the Department of Defense, inside DARPA, inside the various armed forces under many commands. But the scale of the investment by our adversaries and the pace at which they are doing so suggest that we are going to need help inside our government and outside our government if we're going to solve this problem.
Brandon Tseng: If I could actually just build on some of what Mitch was talking about. Paralysis is a good description of kind of the state of things. But I want to empathize with the customer here, with the government here. It's a hard problem to solve. It's hard to navigate a large bureaucracy. And I am thankful that there are people in the government that are willing to take on those challenges because oftentimes, I know this personally, it can be very frustrating for them. But I can't understate how important it is that they do solve these: work within government, work within industry, work with academia to solve these problems. And because they're not going to get solved unless people actually take the to start solving them.
Brian Kenny: So Brandon, how do you even begin to get at that issue? How did you and your team figure out who to talk to and who to bring your pitch to?
Brandon Tseng: Trial and error, hard work, grinding, talking to a number of different people. I don't think our entrepreneurial journey is that different than many of the entrepreneurial journeys that so many companies and so many successful companies go on, which is great because it gives me hope that there will be other companies that can work with government to solve these problems. Because, I'll just say, the government tackles some of the largest problems in the world, Shield AI is not going to be the only company in the world to solve those problems. They need all the help, all the partnership that they can get from industry to tackle these problems.
Brian Kenny: Mitch, I remember we talked about your book on a previous episode of this where I think you gave some great examples of public sector enterprises that are really trying to find ways to partner with private industry and be creative about how they do it. And this case talks about, I think, something similar in the DIU. Can you maybe describe what that is and kind of the role that they play in the case?
Mitch Weiss: The DIU was founded under then Secretary of Defense, Ash Carter. It was an effort to try to bridge some of the distance maybe that existed between the Pentagon and startup world. It was meant to bridge that both personally by placing people in places like Silicon Valley and Boston and elsewhere. It was meant to help people like Brandon and other startup founders navigate inside the various programs inside the DOD, find money where it might exist. It was meant to and did help deploy creative procurement mechanisms so that companies could get to revenue earlier, and then use that revenue to raise private venture capital.
Brian Kenny: The DOD is enormous too. I mean, the case talks about the sort of size and scope of it. Brandon, can you talk a little bit about all the different groups that you need to be in touch with to sell a product like what Shield AI has?
Brandon Tseng: You're absolutely right. The DOD is enormous. There is a question of, "Where do you start?" You know? At the beginning it was, for us, again, not so much different than any entrepreneurial journey, is identify a champion organization, a champion individual that is going to champion your product, help build a bigger picture for their organization in the sense of the Defense Department, and then lean in and work with them to move things forward. And so that was where we were in 2015, 2016. And it was DIU. They were this champion of ours that was able to get us our first contract. And from there, we were able to build and show the department the value that Shield AI could create and provide. And one of the things that I think is really exciting about where the government industry is and why business-to-government, B2G, is such a exciting space right now is, I never before, it was probably a pretty bold statement, but I would make it never before has the government been so interested in working with small companies, small startups that are bringing technology solutions to the table. They are trying to figure out ways to work with them, ways that they can help solve their problems. And what that has done is you've seen a massive amount of investment into the space, whether that's with Shield AI or other venture back companies. But that has really only taken place over the past six years.
Brian Kenny: What do you think's driving that?
Brandon Tseng: I do believe Shield AI's been a trailblazer in some senses. There've been a handful of other companies. I think if I had to point to some of the originals, trailblazers: SpaceX; Palantir were trailblazers as well. I think the government and their admission that they need to change, they need to find ways to work with companies that want to scale technology companies, rather than doing business the same way that they've always done, they've made policy changes to enable faster contracting, faster resourcing. And so it's a combination of all those things that has made this space so exciting.
Brian Kenny: Mitch, you mentioned earlier that we're in a race with China among others to excel in this area. I would imagine this process unfolds quite differently in a state controlled place like China. Are they looking to private industry in that country to try and augment what they're doing or do they basically control all the levers?
Mitch Weiss: The separation between private and public there is of course different as we would understand it here. I will say, for example, SenseTime, which probably still is one of the world's largest AI-based companies based in China is almost certainly an engine for the Chinese surveillance state. And so, I think we can safely assume they leverage every tool they have.
Brian Kenny: And money is probably not as much of a factor as it is... And this is one of the central themes in the case is that the Shield AI product is not inexpensive. It's a $60 million investment. How does that break down, Brandon? Is that for one unit or is that for-
Brandon Tseng: No. So that was for a research contract. That wasn't the price of the AI. That was to develop a capability for the US Department of Defense. When we think about strategic capabilities and the cost of strategic capabilities... aircraft carriers, nuclear submarines, stealth, bombers, stealth fighter jets, AI and autonomy. And then, if you were to compare and contrast the level of resourcing that goes into an aircraft carrier, or a fighter jet, or a submarine, and see how the DOD is resourcing and getting behind the AI, there's a mismatch in terms of, "Hey, this is a strategic capability versus the level of resourcing and funding that's going into it."
Brian Kenny: Yes.
Brandon Tseng: And that's one of the fundamental challenges. We're starting to see it shift, but it's hard because the DOD has to reprioritize how they spend their money across these legacy programs.
Brian Kenny: Yes. Another challenge, obviously, that entrepreneurs face is finding people who believe enough to invest in what they're doing. Mitch, this might be for you, whether it's more difficult for an entrepreneur who has a product that they're essentially building and selling for the government. Is it harder to get private investors to support that kind of effort?
Mitch Weiss: Well, it was harder. Right? So with this case now is, I wrote this a few years ago and Brandon will recall these meetings personally, but he had to go try to raise money. And literally, dozens of investors said, "Oh, please just do this with energy companies," or "Oh, please just do this for consumers. It'd be so much more investible." And it was quite hard to raise venture capital for govtech, generally for defense tech, several years ago. It's getting easier. It's getting easier. I wouldn't say it's easy. It never is. But certainly in govtech, broadly defense tech, the money is coming out much more substantially than it used to. There are many more investors who are focused in these areas than there used to be. Brandon can certainly tell you how that's worked out for them, but it was difficult. It's getting easier. And I think there's work to be done to make it even easier yet, but we're seeing an evolution in the space for sure.
Brandon Tseng: First, like Mitch said, raising money is always hard for any entrepreneur. But I'll say, back in 2015, for us, it was impossible. We were unsuccessful in raising money. Right before I started at business school, went up to 30 investors, 30 different funds. 100% said no. In 2016, we were able to change that narrative and get a couple yeses. And then, I think to Mitch's point, we've seen a number of investors start to look at this space and see that it is going through disruption. It is transforming. The government is transforming as a buyer. You brought up a Tesla analogy earlier around self-driving cars. A lot of people thought the automotive industry wasn't going to get disrupted or transformed. And then we all have seen what Tesla has done. A lot of people probably still don't believe that the defense industry can get disrupted or transformed, but we are starting... You know? I think more and more people are saying, "You know what? It is possible for industry transformation to happen in these legacy industries that for a long time people thought couldn't be touched."
Mitch Weiss: I should say, I suspect at this point, there are some of those investors who said no in 2015, who have watched what Shield AI's accomplished and looked at their latest rounds and evaluations and regretted those decisions back in 2015.
Brian Kenny: Yes. Brandon, was there ever a point at which you thought this just isn't going to work? This is too big of a thing for me to tackle. Did you ever feel like throwing the towel in?
Brandon Tseng: No, I'm stubborn. I'm relentless. You know? I'll just say being a SEAL, I don't believe in throwing in the towel. You find ways to solve problems. And I can empathize with every entrepreneur out there. It's extremely challenging. You get the stress, the anxiety; are very large. But the way that we think about it is, focus on the problem one step at a time, continue moving forward. And if you really get down to the first principles, normally you can find a way to solve that problem. And the day gets brighter when you do that.
Brian Kenny: So, Mitch, let me ask you: If there's one thing that you want our listeners to take away from this case, what would it be?
Mitch Weiss: The one thing is that you can sell new products and services to government. So my big belief coming out of this case is that government selling is government buying. And if you want to do selling to government, your task is to try to understand how government buys. And in the case, Brandon's brother says, "Look. We just have to figure out the building blocks of how government buys, and then we can make this work." And the thing I want people to remember reading the case, sitting on the sessions, even listening today is there's just three basic building blocks. All governments undertake some process to ascertain their need; all governments undertake some process to understand how many resources they have, where they might get them, and what they might spend them on; and all governments undertake some process to ultimately decide from whom to buy, what to buy, how to deploy it in the field. If you can understand how your government is ascertaining need, what decisions they make about resources, and how it is they decide to do acquisition deployment, you can sell to government too.
Brian Kenny: Do you think that based on what Brandon has been able to do with Shield AI and the discussion that we've had today that looks like government is trying to find ways to get better at this, with the DIU and other things like that, do you think that this puts us in a better competitive stance in terms of this race for AI and the challenge that the US faces in the global landscape?
Mitch Weiss: We're better off for agencies like the DIU and efforts like that. We're better off for startups like Shield AI showing how you can sell them to government. We're not where we need to be yet. We need to get to be better buyers and better sellers, all of us together. And then ultimately, I do think we can do what Brandon said, which is combine that instinct for entrepreneurship and public service and solve big problems anymore.
Brian Kenny: So Brandon, let me ask you this. With the benefit of hindsight, you've been on this journey for a while, are there things that you would do differently if you had to do it all over again?
Brandon Tseng: That’s a fantastic question. I'll start by saying you make so many mistakes, so hindsight is 2020. But if I could really just point to one thing that I think would've made the biggest difference, the biggest impact. When you're an entrepreneur, when you are an early stage startup, there are so many things going on. You're down in the weeds, you're fighting fires. Picking your head up, taking a strategic view around the market, around what you're doing and taking that tactical pause, and focusing on the strategy will pay massive dividends down the road.
Brian Kenny: Mitch Weiss, Brandon Tseng, thank you so much for joining me on Cold Call to talk about the Shield AI case. I've really enjoyed having you.
Brandon Tseng: Thank you.
Mitch Weiss: Thank you, Brian.
Brian Kenny: We are excited to be celebrating the 100-year anniversary of the case method at Harvard Business School. If you want more on the history of the case method, visit our website: www.hbs.edu/casemethod100. Cold Call is a great way to get a taste of the case method, after all each episode features a business case and its faculty author. You might also like our other podcasts: After Hours, Climate Rising, Skydeck, and Managing the Future of Work. Find them on Apple Podcasts or wherever you listen. Thanks again for joining us. I'm your host, Brian Kenny, and you've been listening to Cold Call, an official podcast of Harvard Business School, brought to you by the HBR Presents network.