The Future of Multi-Domain ISR
September 22, 2025
Watch the Video
Read the Transcript
This transcript was generated with the assistance of AI. Please report inconsistencies to comms@afa.org.
Brig. Gen. Melissa A. Stone:
Good afternoon, everybody. Can you hear me okay in the back? Awesome. We have the coveted late afternoon time slot. But we have an outstanding panel of experts that I’m sure will definitely keep us on time on target and very engaged. I am Mel Stone. I work in the Deputy Chief of Staff for Intelligence in the Air Force A2 office in the Pentagon. And I am honored to facilitate a discussion today on — with a panel of industry leaders with a variety of different backgrounds, space, advanced research and development, generative artificial intelligence and advanced technology. And we will center our discussion on the future of multi-domain ISR and get a vital perspective from our panels in industry. I will do my best. If you’re playing AFA bingo, I will do my best to help you fill out those squares. There’s no AFA bingo for those of you searching through your folder. Maybe next year. Let me start with quick introductions and I’ll ask each of our panelists to give you a sense of who they are and what industry perspective they are bringing to our discussion today. I will start here to my left, Mr. Scott Jobe from PhantomWorks, Boeing, and he’s an Air Force veteran. Next to him is Mr. Tyler Saltsman, Army veteran, Edge Runner AI. Next to him, Mr. Michael Geist, SES Space and Defense. And at the end there, Mr. Mike Shortsleeve, General Atomics and Air Force veteran. So I’ll ask each to do a quick intro of themselves and their company. And we’ll start with you, sir.
Scott Jobe:
Okay. Good afternoon, everyone. Thanks for joining today. My name is Scott Jobe. Everybody calls me Frag. So if you hear that pop up, that’s the reason. That’s my nickname. Currently work in several advanced development programs in PhantomWorks. And most of our activities are focused on integrating across multiple domains from the seabed to space and everything in between. And really focused on that machine to machine and sending data through standardized messages that are going across to help build a common operating picture and develop long-range kill chains.
Tyler Saltsman:
Test test. Hi, everyone. Tyler Saltsman here. I’m the founder and CEO of a company called Edge Runner. We’re a newer startup, only about 18 months old, but we raised $17.5 million. What we’re doing is we’re building a better version of ChatGPT that never needs the Internet. It can live right on your laptop or your ATAC. And then it’s AFSC or MOS specific. So me, I was a 90 Alpha, an Army logistician. I require a much different AI than a fighter pilot or a combat medic. And the key here is how do we make AI usable and, of course, disconnected away from big tech, but localized, private, and most importantly, personalized. Before that, my background, I worked at Stability AI as head of supercompute. My job was to train these large language models, latent diffusion models, function calling models, you name it. And then it was from there we realized there’s a massive need for this for the warfighter. And then, of course, I was a 90 Alpha logistician. I served in Operation Atlantic Resolve in 2017. And it was over there where I really saw a need for AI solutions that are completely disconnected from the Internet.
Michael Geist:
Good afternoon. My name is Michael Geist. I’m the head of product and innovation at SES Space and Defense, which is the government vertical of SES, which is a well-known global connectivity provider. But we actually provide a lot more than simple connectivity. We also are working on new developments related to hosted payloads as a service related to relay services from different spacecraft in different orbits, as well as a variety of APNT and other associated noncommunications, nontraditional communications initiatives to go along with the global connectivity services that we provide.
Mike Shortsleeve:
Good afternoon. I’m Mike Shortsleeve, vice president of strategic development for General Atomics Aeronautical Systems. I think many of you know we’re the ones that build the MQ-9 and the Grego for the Army. We’re also the ones that are building the CCA that currently the Air Force is moving out on. We also do a multitude of other things people don’t recognize. We always are associated with just unmanned aircraft, but the reality is we actually do a lot of work in the sensor area as in addition autonomy, AI work. We have quite a few software engineers also. So we’re not just building aircraft, but we do a multitude of other different things. And we’re just one affiliate company under the great big General Atomics umbrella that does defense across the board from satellites to aircraft launch systems on carriers to energy nuclear fission, fusion, all of that. The company actually has a very wide portfolio.
Brig. Gen. Melissa A. Stone:
All right. Thank you, gentlemen. Let’s get right into it. So you heard this morning from our chief and secretary who laid out the threat environment. And you can imagine looking in a threat environment like that how important ISR is. Effective ISR provides the Senate decision advantage to our warfighters. So for each of you, where do you see the greatest opportunities for multi-domain ISR within the long range kill chain, which of course is how we find, fix, track, target, engage, and assess likely deep in enemy territory and particularly in high threat environments where low cost resilient solutions are essential. We’ll start with Mr. Jobe.
Scott Jobe:
Okay. Thanks. So first of all, a couple definitions at least from my perch if you want to know my definition what a long range kill chain is. That’s the first thing that needs to be defined. In this regard, in a highly contested environment, normally it’s targets that cannot be sensed with an onboard sensor or is a line of sight prohibited by the curvature of the earth. So for aerial targets, think somewhere outside 350 nautical miles-ish depending on your altitudes. So that’s the first kind of definition where you need other sensors that are going to sense to tracks from the find and fix perspective. And then it’s all about data and data transport to get it to the shooters who are going to engage. So the greatest opportunity is also one of the greatest challenges. So that is not an easy task to undertake. Take something on the other side of the planet and get some information off of it. You’ll attract that target, maintain custody of that target, and then fuse that data to include combat ID, get that information data passed through to the shooter who’s going to engage that particular target, and then close the rest of that kill chain. And by close, I mean actually put an effect on the target. Could be kinetic, could be a non-kinetic effect, but those are one of the main challenges. So at least a level set of the definition in my mind is not something that you engage in a traditional manner, either from an aircraft or from a surface vessel or whatever, where you actually sense the target, track it yourself, and finish the engagement. So in the greatest opportunities and also the greatest challenges are how do you go about doing that? Well, obviously you get a lot of multi-domain ISR from the space layer. So there’s space ISR that can help you identify and track certain targets. But there are phenomenologies and physics limitations associated with any sort of sensor to include the space domain. So in my mind, that’s one of the greatest challenges. How do I take a sensor that is in a different domain and then integrate it with other sensors that are in a different domain of itself? So from space to air to the surface. In the air domain, fairly well understood and well known. Space domain is as well, but how do you integrate those is one of your greatest opportunities and the same challenge. So you’ve got to build a picture that tracks all of the targets that you need to track at scale, and you have to do that with combat ID and infusion. So it’s everything from either the space layer sensing it, data backhaul to say the distributed common ground system where you actually can fuse and develop targets, and then get that information back in a time scale that’s relevant to engage. So in my mind, you’re going to need a redundancy layer that includes space. It’s going to include some air assets as well. It could include other assets, either ground assets, so it could be forward personnel on the battle space or it could be forward sensors such as radars or over the horizon radars or things of that nature. So that’s one of your greatest opportunities and challenges that is presented to the joint force at this time.
Tyler Saltsman:
So what I’m most excited about is all this sensor data being generated. So for example, to his point on cruisers, destroyers in the Navy, for example, we’re working with them to ingest all the sensor data and make sense of it. What we learn is they can generate over 150 terabytes a day of sensor data, but 90% of it is noise. And so how do we go through all this data and make the best decision in the fastest way possible? And so as industry, what I’m excited about is bringing AI to the data. And what that looks like is taking a large English model or a hybrid model like a state space model architecture with attention layers, and what we can do is train on all this sensor data, create the Q&A pairs, but now with all this sensor data, I can now interact with it via a human conversation like we’re having right now via natural language processing. So rather than using tools that are just ingest machines that just give you insights, which is what we have today, and the problem with insights is you can interpret that a million ways to one, why not have AIs actually speak this language? Now we can translate between human language and machine sensor data language, and now I can really understand what I need to do. And that’s what I’m really excited about to add value there, but also with images. So for example, in Ukraine, we’re having issues with drones getting jammed. Well, and drones see with LIDAR and radar data, or they use a pilot. So if you jam that drone, it disconnects the pilot from the drone. So now what we can do is add vision language models on those drones so these drones can see. So now imagine a suicide drone that can see a Russian tank, neutralize it while it’s carrying an explosive payload. At the same time, though, if there’s innocent civilians around, it can see the innocent civilians and divert away. And what we can do now is create these swarms of consumer-grade drones that can be used for defense or offense. So what I’m excited about is the combination of all this sensor data we can use to make sense of now with language models, and then all the images that we’re seeing with reconnaissance drones to create a better picture of the battlefield. And now we can create that common operating picture that we can standardize across all the branches to have a more united fighting force.
Michael Geist:
Thanks. Yeah, where I see the biggest gains and the opportunities for gain are in latency, reducing the latency from sensor to shooter, if you will. And so from our perspective, from a space perspective, as we look at things that we can do from incorporating or integrating hosted payloads into our connectivity services that we provide as a standard main day fare, and how we can move from bent pipe constructs, satellite constructs, into constructs that are either digitized or regenerative in nature and differentiated from the type of services that we offer, and even take it one step further to incorporate edge processing on the spacecraft itself, that leads us to an ability to reduce the latency significantly from traditional means of sensor data bringing back down to Earth, taking to a processing facility, and then redistributing out to the tactical edge through one or multiple hops into an environment where perhaps a tactical warfighter with a UHF radio wants to see a picture of something beyond their horizon, and they can make that task, they can make that request, an Earth observation satellite can see that, can receive that task, or can take those pictures, can translate those from their LEO orbit to, let’s say, another exchange point, if you will, digitize that, process it at the edge in space, and re-deliver that image in a fashion that makes sense for the limited capacity of, let’s say, a tactical UHF radio to the user on the ground without having to go through four, five, six hops and processing on the ground to make all of that happen. So that leads us to an ability to move from latency of information that might be 10 or 20 minutes into something that’s seconds to a minute. And so that’s a massive increase in capability for the warfighter and something that we’re extremely excited about.
Mike Shortsleeve:
What they said. No, kidding, all kidding aside, I’ll kind of sum up a little bit of each of what they’ve talked about. You know, really, when you look at what multi-domain ISR needs to do in the future is, one, it needs to be connected, it needs to be intelligent, and it needs to be decisive. So exactly to all their points here in that aspect, the connective aspect of this is you need a lot of things out there. Everything needs to be a node in one form or fashion. I can’t rely on moving data or knowledge, in that case, if we’re doing edge fusion, if I only have one mechanism to push it to one area. I need to be able to do that in multiple areas and have connectivity across the board. Doesn’t make a difference what the platform is, whether it’s in the air, in space, or on the ground, or at sea. So on the connected aspect is we need everything to be connected in that realm. When we talk about intelligence, exactly the points that they’ve been making here about how AI is going to be the enabler, it’s the force enabler, it’s the force multiplier for us. We need smarter decisions, right? There’s a lot of data out there. And I know a lot of people talk about data lakes. I always reference it as data landfills. The data just piles up and piles up and piles up, all right? I had 28 years in the Air Force doing ISR. I worked at DGSs. I know that data just falls on the floor and doesn’t get looked at, unless there’s one individual who just happens to be sort of a picker, right? Looking for that gas tank for a certain motorcycle. They’re going to go and go through the whole junkyard. The issue is there’s so much out there. So if we can get to a point where it’s smarter data, and I think having capabilities like you talk about, to put it in more natural forms, right, that we can work with, that’s going to accelerate the ability to do this when we talk about multi-domain ISR, because it’s going to be needed. And the last one is decisive. We need decisions. We need it to cue us to make certain decisions, whether that’s to actually execute a strike, accomplish a certain effect, or just to understand different courses of action that may exist out there. So we can’t just rely on human power and fusion occurring between the ears like we have been. Some of us more challenged than others in that capacity, but I would say that it has to be decisive and provide that information in a format that we can actually act on. So kind of touching a little bit about what each of them said when we talk about that long-range kill chain, that is a huge challenge and a huge problem that we have to get after. But I think one way we can do that is by being connected, intelligent, and then also being decisive with the type of capabilities that we build.
Scott Jobe:
I’ll just add on real quick, General Stone. So excellent points by all the panel. And I think Mike gets to the point of where he’s talking about dynamic remission planning and integration across multi-domain. And that’s really one of the key elements, which has got to be, to Tyler’s point, it’s got to be machine to machine. There’s got to be some level of autonomy all the way up to AI, where you can now coordinate space assets with air assets with surface and ground forces that are out there. So that dynamic remission planning and dynamic flow is part of the long-range kill chain, especially when you talk about the tyranny of distances in places like the Indo-Pacom Theater, where things are going to be very dynamic from a force generation perspective or from assembling what you need to get a pulse in that operation. It’s got to be some level of autonomy or machine-driven, because the data is so vast, as Tyler pointed out, and everything is so dynamic. Having that dynamic mission replanning on the fly to present courses of action to battle managers so you get true battle management command and control, I think, is also a key element.
Brig. Gen. Melissa A. Stone:
Excellent. And just to pull the thread on the data piece, some of you had touched on this, but let me give an opportunity, if there’s anyone that would like to add to this question, how can industry help the Department of the Air Force, the Air Force and the Space Force, effectively harness all of this sensor and other data, whether that’s through enterprise infrastructure, business processing at the edge, or any other ideas? I’d like to start with Mr. Saltsman.
Tyler Saltsman:
So what we can do is provide the data pipelines to help make sense of this data, as well as the language models to train on that data. The key here is we need to go with the data gravity, not against it. And where all the data is, it’s right there on the assets in the battlefield. So what we can do is build out these pipelines. So for example, let’s say we’re using swarms of consumer-grade drones to neutralize an enemy Russian UAP. Now, let’s say, though, the Russians came out with a new drone that doesn’t exist yet, rendering all of our AIs useless because the AI can’t recognize that new drone. What we’re building is a way to rapidly construct a data set of that new drone of just a couple images. Now we can use AI to generate millions of images, throw that back in the YOLO model — You Only Look Once image classification model — compress that, and put it back on a drone, all right there in the field in the battle at the edge. Therefore, the data never needs to phone home. You don’t need to use the cloud. You can do it right there on the edge. And that’s what we’re really excited about, is providing those tools to do it. So rather than it taking weeks to months and needing lots of data scientists and data junkies to put it together, AI can streamline about 90% of all of that, which, of course, saves you tons of time and allows you to, again, execute with violence and make decisions faster.
Michael Geist:
I’m going to go ahead and add. So the Space Force strategy incorporates — or includes comments toward the tighter integration of commercial and government capabilities. And as we’re — as industry are innovating and we’re launching our next-generation constellations for commercial applications and purposes, there are a huge number of opportunities to form public-private partnerships to — in pursuit of advanced capabilities to deliver to the warfighter. So from an SES perspective, we have a contract with NASA right now. We’re one of six companies that have a contract with NASA specifically in regard to replacing TDRS as a capability, and TDRS is a very valuable capability, as many of you in the room, I’m sure, can imagine. And so there are six different commercial entities that are working to deliver a commercial equivalent of the same service, are looking at ways that we can use our current or future constellations to evolve and take over that mission. In a very similar way, as we’re rolling out these next-generation constellations, we’re enabling space on our platform — not space, literal space — on our platform, SWAP, on our platform to enable hosted payloads so that we can get after solving some of these ancillary missions that warfighters have, whether that’s space situational awareness at LEO, MEO, or GEO, whether that is sensors, earth observation, relay services. All of those things are available to you as the government in working with us through public and private partnerships. So what I would say is that where we come up with these different ideas, if these have interest to you, you simply need to demonstrate your interest, show us your interest, show us some commitment so that we — we’ll go invest. We’ll make the big investments. And as we make the big investments in these next-generation capabilities that provide better multi-int capabilities at the edge, we will offset your cost of doing this yourself. And that’s the big savings that we can make to the government.
Mike Shortsleeve:
Yeah, so just adding to what has already been discussed, I’ll put a little twist on it there and I’ll kind of give sort of from like a user perspective, right? You got all this data and we need machine-to-machine capabilities. We need that AI overlay to be able to interpret a lot of that so that we’re not having to do a lot of that ourselves. But I will also offer, when we talk about having a common operating picture, I think it’s more of having a user-defined operational picture. It’s not about one type of thing that everybody sees. It’s about what that individual needs at that time to be able to see. You don’t need all the data to be displayed for you. Maybe you only need certain layers based on where you’re at geographically or even just what you’re trying to accomplish or a certain effect you’re trying to achieve. So the ability for all that data to be sort of in that background and being sort of smartly kept and pushed to where it’s needed, an individual has a different requirement, you know, far forward than somebody who’s sitting further back. So we need to be able to have that data accessible and it’s going to be the machine-to-machine, the AI overlays that are part of this, whether it’s in space, whether it’s in the air, we need all that stitched together. And so from my perspective, you know, we need to take that perspective of looking at it from a user-defined operational picture to achieve what they need at that moment in time.
Scott Jobe:
Yeah, for Boeing Defense and specifically Phantomworks efforts, a lot of this is driven by what industry can provide as deep engineering and deep knowledge. So there’s a lot of warfighting expertise that lives in industry, but there’s deep engineering, which is what you really require to solve a lot of these really, really hard problems. And we also have the resources and the assets to do deep operational analytics to see if we get a user-defined picture, which industry can help develop with that too, how do we implement that and how do we get that into actual real engineering terms that can be utilized in some technology that you need? So you’ve got deep engineering expertise, you’ve got deep resources also that can help with the operational analytics, put it in the virtual environment, iterate on it to develop rapid capabilities and move with speed. The other part that industry is really work with other industry partners, not just one, because this is a complex system of systems and family of systems approach that you’re trying to solve with multi-domain ISR. So industry can also reach out to other industry partners to bring what they have to bear. The last bit I’ll throw on from an industry perspective is we have lots of very rich products, they’re all fantastic, and the data that comes off of them, from sensor data or data that’s generated off-board that is ingested into that platform, we have a deep understanding of what that is too, and so we can help bring to bear how do you tag and curate that data in a way that follows UCI standards for universal command and control interface that you can now send machine to machine and actually make the data usable.
Brig. Gen. Melissa A. Stone:
Several of you have mentioned integration, and I think a unique vantage point that you all have, and you just mentioned it, Frag, is you provide capabilities not just to the department but to the rest of government, sometimes our allies and partners as well. In terms of integration, how can you help the Air Force and the Space Force better integrate across government with our partners, especially our sister services, and even more important from our perspective, with the intelligence community and the data on the Title 50 side that they can contribute? And I’ll start with Mr. Shortsleeve.
Mike Shortsleeve:
Yeah, so we need to fight as one, not as many, and a lot of times when you look at the capabilities that are developed, they’re developed in silos even between the services, and sometimes even within the service, it’s done. So certainly industry can help with knitting together, if you will, whether it’s allies and partners or it’s other sister services or it’s the intel community. Those are things that we can do through whether it’s open systems, making sure that they’re interoperable, making sure they’re secure. But the other thing I’ll mention here, kind of going off a little bit about what was said earlier was, I mean, let’s not kid ourselves. Industry is needed to win any fight in the future. So why are we not viewing industry and manufacturing as a warfighting domain? Why would we not look at it from that perspective? And if so, what that means to me is you’re integrating, right? We can talk about partnerships and we can talk about things that you want to do, but the reality is, I’ll be honest with you, having been in uniform and now on this side, I used to think we were transparent when I was in uniform, but you’re not. And bringing in industry, whether it’s using their money, and there’s billions out there in research and development, there’s also a lot of brainpower that’s back there. And being able to have that discussion and bringing them to the table, I think, is more important. But until we actually, in my mind, view industry as part of a warfighting domain, we’re going to not just have to outbuild them, we’ve got to outsmart them with the capabilities that we’re doing. And so when you think about that, we fight as a coalition, we fight with allies, right? Not fight against allies, maybe, but industry is right there, the whole step of the way. So we have to think, in my mind, from that perspective, when we talk about how do we integrate all of this, I would also offer you’ve got to integrate industry in as early as possible, too.
Brig. Gen. Melissa A. Stone:
Michael?
Michael Geist:
Yeah, I’m going to add to this by saying that if we think about delivery of capabilities as an as-a-service kind of model, and I think about this from an innovation perspective, there are many different areas where the U.S. leads the world in a lot of innovation when it comes to intelligence, multi-intelligence sensing, and things like that. And there are other areas where other nations, hopefully friendly, lead the world as well, right? And so if there are areas that are sensitive to us that we do really well, or sensitive to other sovereign nations that they do really well, but other aspects of that that can be delivered across the coalition, if you will, of friendly forces, then we should deliver what we can deliver across all of the community to use it while protecting the things that can’t be delivered outside of the sovereign nation of ownership of that innovation. And there are ways to do that. There are ways to do that both in policy through ITAR controls, and there are ways to do that through artificial intelligence and machine learning and management of information in that way.
Tyler Saltsman:
So I’m going to go with a little bit of a different approach. And I know at industry, we often complain about the DOD taking, or the DOW, taking a long time to work with, but y’all have been moving really fast. So for example, we signed a cradle with AFRL in just three months, whereas we did a deal with a major movie studio that took a year. So y’all have been moving fast. I think one of the biggest challenges as an industry is we just build cool products and expect you all to buy them, and we’re not really understanding what your real challenges are and working backwards from them. And so I think as industry, we need to do a better job of having a shared responsibility model and being real partners. Let’s really understand what are the pain points. Obviously, we want to win wars and fights, but we also want to dominate. But how do we do it, and where are the gaps? And as industry, if we’re not really understanding that, we’re not building things that are effective for you all. And so once we do finally do that, then I think we can start to integrate much easier and in a more speedy fashion.
Scott Jobe:
Yeah, so this is a fascinating topic and a really pertinent one. So coalition and ally partners are absolutely critical to the US national security and our interest, right? That is just a fundamental fact that we have to come to grips with. I think one of the things that industry can do, which is very difficult to do, I will admit, is number one, if we have a clear understanding of what the warfighting gap and capabilities are, which I think we’re pretty good at, and then have a demand signal that sets design, engineering design for exportability. And that naturally brings in coalition allies and partners. Now, I would say General Stone knows probably better than anybody or others in the room, that from an ISR perspective, that’s an extraordinary challenge. There are security policies. There are data right policies. There are all kinds of challenges to that. But if you design for exportability, that you can now have ubiquity across coalition and partners, utilizing the same, at least the same type of equipment, the same types of capabilities, it really leans you forward into the ability to actually share capabilities across national boundaries. And so design for exportability is something that’s probably needed and often not talked about.
Brig. Gen. Melissa A. Stone:
Awesome. So, as we’ve talked about in other engagements, and as some of you have referenced here, and it doesn’t hurt my feelings at all to say, to hear that it’s difficult sometimes to do business with the Pentagon. I don’t think that’s surprising to a lot of us in the room. But I think it’s probably the most important question that I could ask you to leave this audience with, which is, what would you like us in uniform to know about how best to do business with industry and make industry as a partner, to your point, Mr. Shortsleeve, not just in name only, but early on? And in fact, let’s go ahead and start with you, if we could, with that one.
Mike Shortsleeve:
Yeah. I mean, I think it’s bring us to the table. Let us be involved in understanding exactly the point of what are the challenges? What are those pain points? Sometimes it’s alluded to in an RFI, or maybe there’s some industry days and things like that. You know, you have an industry day, and then maybe a year goes by or two years go by. I’m talking about a consistent drumbeat, including industry into this, and help us figure out, you know, what it is you want to get after. And I’ll tell you, there are examples of that. There are consortiums that have been put together where there’s a multitude of different industry partners in there, and a problem’s been thrown to them, and they come up and figure it out. But I will also say that by signaling to us exactly the kind of things that you’re going to be needing doesn’t have to be specific. But like I said, if the Air Force is already fiscally challenged, right, you heard about that this morning, and you probably heard about it for years, about how there’s not enough money, you need more Air Force. There’s billions of dollars on the industry side doing research and development, and each of us would rather be building something that’s going to be of use, or at least lead you down to the path to answering the solution or creating the solution you need. So from that perspective, I mean, that’s kind of how I look at it, is like, bring us in, but don’t just bring us in one time. Be consistent. And what I mean by that is, if you tell me something, I’m expecting you to tell each of us the same thing, right? I’m not saying that you’re favoriting certain people or whatever. It’s like, have your standard answer, but let us all know that.
Michael Geist:
I’m going to add to that. You know, it’s no secret that industry is innovating at warp speed right now, and you actually said it. We want you to tell us what you want. Don’t tell us how to deliver it. Just tell us what you want. And give us as industry the opportunity to ideate and come back to you with different solutions. And enable us to bring those ideas to you in a way that’s both meaningful for industry and meaningful for the government. What I mean by that, I referenced earlier a contract that we have. That contract is a cost avoidance contract for NASA. That’s a $4 billion cost avoidance contract for NASA. And so NASA went out and spent $300 million to avoid $4 billion and turn themselves into a customer buying a service capability. We’re looking at all of the different bespoke systems that the government buys today. And I could name them. You know, I work in the space industry. So I work in satellite space. And there are billions, tens of billions of dollars of bespoke programs that if through a joint partnership or through a public private partnership, you would pursue cost offsets associated with commercial industry and what those bring, you could take a $10 billion program and turn that into a $2 billion program. And then that gives the government more money to spend on other things that they need. So doing more with less. So just bring us to the table. Tell us what you want. Let us ideate. Let us bring those ideas back to you. But when you see those ideas that are important to you, you have to get out your wallet. You have to spend in those ideas to make that savings happen and allow you to do more with less.
Tyler Saltsman:
You know, I’m going to give myself a shameless plug. But as we define industry, I know we think of the big primes. We think of Palantir’s, the Lockheed Martin’s. But also think about the garage bands like us, the new startups. We’re young, but we’re hungry, and we can move fast. And we can build things that maybe you didn’t know that existed. For example, like completely offline AIs. So think about these new startups. And I understand that there’s concerns that we won’t be around long enough to be effective well. But work with us, and that won’t be a problem. So as you move forward and work in industry, think about us. And also think about how can we partner up with different folks where we all have different core competencies and different swim lanes. And I think as we all sort of come together, rather than fighting over a single contract, we need to be better at working together as a joint forces operation ourselves.
Scott Jobe:
Yeah. I agree with all those points. I think they’re really well taken. I think the last one, I’ll — my specific spin on it. Part being at the table, 100% correct. Michael’s right. We’ve got to continue to look for opportunities to do things smarter. And one of those ways that you can do that is you can do lots of heavy prototyping, which Phantomworks is driven towards. You can also do lots of experimentation. Probably a lot more to be had in that arena where you can iterate on something you’re actually doing and seeing and get after, you know, speed or look at the different capabilities. So I think iterating and experimenting is probably something that needs a little bit more emphasis. And we are all cognizant of the fact that this is a very fine-tuned balancing act. Because at the end of the day, the government’s still going to want competition. And we all get up every day wanting to compete. And to compete for what? To deliver the best capability to the warfighter as fast as we can. And that’s all that we think about. So one of the ways you can do it is lots of rapid prototype and experimentations. Some of those, in fact, many of them, if you do it right, will not work. That’s okay. That’s called learning. And that’s what we should be doing. Rapidly spinning through experimentation and prototypes until we find the thing that you actually want.
Brig. Gen. Melissa A. Stone:
Terrific. And we have a few minutes left. So I want to maximize the time we have with you experts. And if you’ll forgive a question that was not prescripted, I’m confident all of you can handle this one. And just to get him thinking about it, I’ll let Mr. Shortsleeve know we’re going to start with him. But if you could leave us with any closing thoughts. But I’d really like to hear what you’re most optimistic about with the future of the department.
Mike Shortsleeve:
Yeah. I mean, I’ve always been a believer that the competitive advantage actually lies with the people. Not just the Airmen that are doing the job today. But if you go back to what I was saying about partnering with industry, there’s a ton of smart people. I can tell you in general atomics, I’m blown away every day at the type of things people can come up with. But sometimes you just can’t get it over the line or get it in front of who needs it. One of the pitches I will make, getting to the point, you really triggered something in my mind about these smaller companies, these startups, right? They don’t always have access to the facilities for some of these things, right? They can’t afford to have a classified facility to do some of the work. So the government being able to help in that aspect I think would be, you know, tremendous in offering them, you know, bring them into a location that they can work on classified or they have access to models or simulation that perhaps others don’t. And getting to the point of the startups and things like that, I mean, if you went back 35 years ago, general atomics, aeronautical systems was a startup, right? So we’re sort of laying that path that you can get to this level and be able to contribute in meaningful ways and continue to do that going forward. But that’s because somebody took a big bet on us, right? We brought a capability that somebody looked at and said, you know what, that can make it. So from my mind, it goes back to that partnership, and I’ll say it again, you know, industry is a warfighting domain, and we have to view it that way. And if you do, you’re going to put more attention on it than you would just viewing us as other vendors that are out there to try and sell something.
Michael Geist:
So I’m going to use, I guess, my closing comments here in the unscripted question to actually give you all in this room a bit of a shout out. And you deserve that, because if I were sitting on a panel like this 10 years ago and having this same kind of conversation, at that point in time, so much more was bespoke, and it still was fully described in seven-year requirement documents and eventually building a program, and maybe the program is late, and maybe the program is overrun. Ten years later today, I talk about fostering more public/private partnerships. You guys already are. We’re involved in more and more public/private partnerships every day. And so kudos to you, because you’re taking advantage of that, and what I’m telling you is there’s more opportunity. Keep feeding into it, and you’ll be doing just great.
Tyler Saltsman:
You know, I’d say what I’m most excited about is building AI specifically for the warfighter, for you all. But conversely, proceed with caution and be careful with it. So, for example, OpenAI and Anthropic and these big AI labs, they’re not the arbiters of truth. They’ve trained these models by transcribing the entire Internet into tens of trillions of tokens to create something that helps you, but at the same time, there’s a lot of poison on the Internet, and we need to understand these biases. And obviously, not to get political, but we can’t use AI that’s woke or that’s heavily leaning one way or the other, because those warfighters were bipartisan and were objective. And then lastly, be very careful with AIs, because the forces at be want this to be the arbiter of truth, and it’s not. And one of my biggest concerns is if you study history, you look at the rise of Stalin and atheist communism with Chairman Mao, and what they do is they remove God and they worship something like the state. In our case, I worry we’re going to try to remove God and worship AIs and not to get overly religious or philosophical, but these are things we need to be mindful of. For example, with chat GPT, if your kids are now prompting chat GPT and it’s giving them answers and they disagree with the parents, that’s a problem, and it’s dangerous. And with AI, they want you to be the product. They want your prompts. They want to learn from you, which is why I’m passionate when we’re building an edge runner, because we’re completely offline and we bring the AIs to you, and it’s safe and it’s secure, and that’s how it should be. So just closing remarks, be careful, use it, but proceed with caution. It’s not a silver bullet.
Scott Jobe:
Very wise to leave me 30 seconds for my closing comments. What makes me the most optimistic is since I traded uniform to come over to the defense industry, I wake up every day with some of the finest Americans that are trying to solve your hardest problems. What makes me optimistic is America can do anything it puts its mind to. We are Americans and we can solve problems. We have the right values and we have the right sight picture. Thank you.
Brig. Gen. Melissa A. Stone:
Awesome. Let’s give our panelists a big round of applause.