Abe Usher: Good afternoon. My name is Abe Usher, I’m the co-CEO of Black Cape, an Innovative Data Analytics, software engineering and machine learning company.
Balan Ayyar: I’m Balan Ayyar, I’m the CEO of percipient.ai. We’re a Silicon Valley-based artificial intelligence firm focused on machine learning and national security missions. And I’m delighted to be with you.
Abe Usher: I think there are three macro trends that I’ve seen in the past five years that really are enabling a lot of innovation in the GEOINT community. The first is the emergence of open-source frameworks like Google’s TensorFlow and Facebook’s Pytorch that give people common environments for creating machine learning models, and then exchanging those models, instead of everything being proprietary. So number one, open-source frameworks.
Number two, the availability of commercial cloud compute—whether that is commercial, Microsoft Azure or AWS GovCloud or even in the classified environment, AWS C2S. Those commercial cloud services enable people to have an idea and how to take data and algorithms and combine them to actually scale that out at the size of data we have. And as terabytes of data are being generated every day in the GEOINT community, commercial cloud, which is that second macro factor, enables people to apply these capabilities in a meaningful way.
And then the third, of course, is sensor data availability. Previously, it had to be the U.S. government that owned a bunch of satellites in space. Today, a lot of innovators don’t own any portion of the satellites or communications infrastructure. They just bring algorithms. So open-source frameworks, commercial cloud capability, and sensor data availability. All three have converged to really change the GEOINT community.
Balan Ayyar: We’re not in this alone, right? We have two very different rising adversaries who view the world differently and think potentially their version of government is the right version for a structured world. And of course, we believe differently. We believe in freedom and human values that are associated with freedom. And so it’s so important for our intelligence and national security agencies to have the advantage.
Abe Usher: I have a lot of friends in the government and also the industry, and some of them tell me they’re concerned some day the United States is going to get left behind technologically. I can tell you as an engineer with more than 20 years of experience in software development, including time at Google, and data analytics, the United States is already behind China and Russia in several areas of machine learning and data analytics. And so I think we need to have a very realistic assessment that there are other nation-states who have significant technological capability. And for us to keep up, we need a new approach. The approach that we’ve been using in the 20th century and early 21st century, it’s not going to get us where we want to be.
Balan Ayyar: From my perspective, having led services companies and been in the environment, serving the nation for some time: We have to know what our innovation economy can provide to help us win and transform, and this is an area where the commercial sector developing new technologies, deploying new technologies, building a great product user experience. This is an area where our commercial innovation has been leading the world. Why wouldn’t in the defense and national security space we want to take advantage of that?
I believe that this will be the critical decision that we see in the coming year or two with these national security agencies. Where they can take advantage of commercially developed products to help them win, you will see great traction and real transformation. So we have to be very aggressive about capitalizing on commercial innovation here in this space. If we’re going to keep up with our adversaries.
Abe Usher: You know, there are a lot more people working in automation, technology, and machine learning outside of the GEOINT community than inside of the GEOINT community. And I would just say the government and its industry partners need to engage with non-traditional small companies and innovators to a greater extent than they have in the past or they’re going to continue to fall behind.
I’ve seen a lot of promising activity by NGA in terms of enabling rapid technology advancement within the community and in particular, I think their joint support and advocacy of the SpaceNet Challenge with a number of remote sensing companies has been really promising. Also, NGA has a robust effort to obtain new ideas and technology through small business, innovative research grants, and NGA research. So I think those things are definitely success stories for NGA. But there are definitely areas that I think the government and the GEOINT community as a whole could improve.
Balan Ayyar: It’s the first time in my career that we’ve had venture capitalists willing to put 30, 40, 50 million into a software that could serve the national security and intelligence environment. When I was growing up as a young general and through my time, we had to build all the software we wanted. So this thinking of a win-win virtuous cycle so you can leverage massive private flows of capital is how I want NGA to look at the world—like a big buyer.
Abe Usher: Most data-centric innovations, a lot of the time that goes into them is finding the right data, curating the data, understanding the data. And so I think it would be a massive leap ahead if NGA or a related consortium or maybe even USGIF would host additional unclassified data that could be used as a common baseline for vendors and industry partners to use in their capability development and demonstrations back to the government. The reason this is needed is right now you have the government asking for a bunch of innovation. They go, “Ah! We want machine learning. We want innovation. We want automation, tech.” And you’ve got a thousand different industry partners all using their own data and then coming to the government saying, “Look at what I did with this data,” And it’s very difficult for the government to evaluate. It’s kind of a literal digital apples and oranges comparison with every vendor and every scientific group using their own information. So if there is additional leadership and the government saying, OK, here are 10 unclassified representative data sets, they want people to use as they develop their capabilities and then come back.
And then related to those 10 data sets, here are 10 analytic problems that NGA, NRO, Department of Defense are commonly trying to solve. And so if you have a solution to one of these 10 analytic problems, please take some of the representative data and show us how you solve the problem. And in that way, we could get, you know, the, I don’t actually know the number, but let’s just say several hundred companies who were involved in the innovative process, working with the government in the GEOINT community, kind of focused in the same direction. And I think that’d be exciting for everyone. Right now, we have a fragmentation with a lot of well-intentioned innovators and I think sometimes frustrated government because, you know, the digital apples and oranges problem of everybody showing a different facet of the problem in their own way.
Balan Ayyar: To keep the best trained and the most capable leaders, we’re going to have to provide them the best tools. So we’ve got to commercialize—NGA and others have got to reach out, make investments. They’ve got to make a splash. We got to have a couple of these technologies go to scale and everybody be excited about the fact that the best tools exist in the hands of NGA operators and analysts.
In the software itself, we have to create an analytic repository, so that every analyst, the senior-most analyst that’s about to retire on Russia, guess what happens right now when they retire? We lose all of that brilliance, the analytic tradecraft, the tipping and cueing, the sense they have of what’s happening because they’ve studied it for 30 years. When we have the right artificial intelligence supporting these analysts, just constantly learning from every operator’s use of the tool, we’re creating over time the spectacular analytic repository. So this ability to keep learning as an enterprise. I call this body of work enterprise intelligence. The partnership with machine intelligence gives us not only dramatic impacts for the individual in accomplishing their mission, it’s also creating spectacular opportunities for enterprise intelligence, to train and keep and develop the most powerful tradecraft and workforce. And if we don’t think about this the right way, it’ll be harder and harder to attract, more and more difficult to retain, and more difficult to train. And then at the end of the day, we’re less prepared as different classes of analysts retire into the private sector. You know, it’s like we’re starting over again. This work, listen, if it’s really as important as we say it is, we should leave no stone unturned on the kind of technology that we can bring to bear to help our analysts win. If the truth be told, if our analysts don’t win, our nation doesn’t win.
Abe Usher: To be successful at innovation, you need to understand the basic value creation process. A version of value creation a lot of people have experience with and they’ve heard of, that they’ve experienced in their own life, is a kitchen. In your kitchen, you have a bunch of raw ingredients. And then in addition to the raw ingredients, you have utensils and you have appliances, and you transform those raw ingredients by using utensils and appliances to make something of greater value. Maybe you dice up the potatoes, deep fry them, and, boom, French Fries. Not the healthiest example, but certainly delicious. And so, after you figure out how to do that once or twice, then you can combine these ingredients in this order with these utensils and these appliances, and you can create a recipe so that you can recreate that valuable thing again and again. What I see with the government, and I used to be in the government and I certainly respect that they don’t have infinite resources. And sometimes the selection of the chefs, the ingredients, the appliances, the utensils, the recipes—there’s a critical weakness in one or two of those areas. Even if you have the best chefs in the world, if they have the wrong ingredients or they don’t have the utensils they need, they’re not going to be able to create value. And I think in the same way, the government has some challenges where you can kind of multiply the effectiveness of all those things together to get the outcome. So if you have a great chef with great ingredients and great utensils, great appliances, great recipes, you’re going to have great results. But if any one of those things is a zero, you’re going to fail.
Balan Ayyar: You’ve got these analysts willing to take the time to validate software. Give them the resources. And we’ve got to align the resources to the vision if we’re going to win. Goodness gracious, if I can have an app on my phone to do almost everything that’s inconsequential, why can’t I have an app that is for things that are consequential? The software is there now, take advantage of the venture capital ecosystem and leverage this in a way that your analysts love. And guess what? The transformation takes care of itself.
One of your responsibilities at USGIF is to keep everybody honest about our progress. It’s not enough that we talk about this. We have to take some risks, we have to validate the software, we have to see what the analysts are saying. It’s just, it’s incontrovertible that human-machine teaming is coming. The question is whether it’s going to be done right. And that’s why we want to make sure that we’re looking at all sides of this, including the commercialization. Listen, we know your work is hard. It’s long hours. It’s thankless. You can’t share with your family. So here’s what we’re going to do. We’re going to give you the best damn tools … it’s going to be like some movie. We’re going to give you superpowers. We’re invested in you. That makes them want to go, “Yes. What I’m doing is important.”
In some ways, the impacts of these technologies are kind of the new arms. These are the new arms. If your operators or analysts don’t have machine learning working for you, it’s like going into battle with antiquated weapons.