Fundamentals first: The case for foundational intelligence today

Welcome to the World of Intelligence, a podcast for you to discover the latest analysis of global military and security trends within the open source defense intelligence community. Now onto the episode. Hello and welcome to our World of Intelligence panel at James, live to you from DSEI 2025. I'm Kate Cox, your host for today and the Director of Strategic Programs in James's Analysis Division. Today's session will take us right back to the fundamentals of intelligence and what goes into effective analysis. And who better to share insights on this topic than our expert panel. Welcome Sean Corbett, Chair of James Strategic Advisory Board. Hello again Sean. Hello. It feels like only yesterday we had this chat. Indeed it was. Dylan Lurkey, Director of Military Intelligence at James. Hello Dylan. Hello Kate. And finally, Tom Barton, Head of AMIA News at James. Welcome Tom. Hi Kate. So at James we produce a range of OSINT content to support analysis and decision making in the defense and security space. We cover more than 90,000 pieces of equipment, 40,000 military units, 30,000 geolocated installations, 197 country reports in a permissive format which Sean will talk about a little bit later and much more. So our expert panel will now talk us through how this intelligence is produced and used, the essentials of good analyst tradecraft and what the future of foundational intelligence looks like. So Sean, to open this up, starting with definitions, what do we mean by foundational intelligence? So I think the first thing to say, I'm really pleased we've used this as a subject because it's something that is, it doesn't grab the headlines, it's not sexy and all the rest of it, but it is absolutely fundamental to how we do intelligence stuff. You'll be surprised to know or not surprised to know that there's no accepted definition. There are differing views depending on where you come from, so foundational military intelligence has got quite a confined definition, but what I think we're talking about is far wider than that. And for me it's the building blocks upon which the current intelligence, the so what, the what if that you've heard me talk about all the time, is derived. So it's not that itself, but without that fundamental layer you can't actually come up with that piece. So to give a few examples, from a military perspective orders of battle, and I'm sure we're going to talk about that before. So orders of battle basically mean what equipment is where, what type, and what numbers, and all the way down to the granularity in terms of modes and models and all the rest of it. What it doesn't do though is necessarily say okay, why is that unit where it is, what are its overall capabilities, how is it integrated into the whole force, and what is the intent to use it with. So that is the extra bit, but without the initial bit you can't do that. Now it's worth saying that, that's on the military side, but for me foundational intelligence is much broader than that. So I go back to my times at our quick reaction headquarters when we did a lot of planning for non combative evacuation operations. So something bad is happening wherever in the world it happened to be and it could have been anywhere for any reason. So there might have been unrest that we had to get the diplomatic staff out, or it might have been, it happened one case of volcano that we just needed to extract people. Now you don't know that's going to happen, but without the foundational level intelligence you can't actually plan and execute an operation. So I'm talking here about things like usable runways. So what's the length of the runways, what sort of capability can they take? Port facility is very important obviously, is it deep water port, what sort of ship can it take, et cetera, et cetera, lines of communication, is it able to take more than a land rover, et cetera, et cetera. And so you've got to build up that knowledge before you can come up with a plan. And then there's a sort of more ethereal, you talked about permissi earlier on, but you're talking about something a little bit big from a country perspective. So what are the politics of that country? What are its main economic indicators that give us an idea of its strength in the world and also its intent on the world stage? So the politics, the economics, that sort of thing is as important, but you don't do those overnight. You've got to really work hard and its resource-intensive stuff to do it, but without it you can't do that. Okay, this means that. I like Sean your conception of building blocks, and I usually think of building blocks in a couple ways. The foundational data pertains to the physical and social structures, data that's around the physical structures like the runways, like the geography, and data around the social structures, whether it be a nation like the United States or a unit like the First Capital Division. And the reason that foundational data is so important, and maybe I'm getting ahead of myself here, is that it really helps us understand the current intelligence in a better way by filling in the gaps in that current intelligence, and by filling in the gaps in our current intelligence picture and our common intelligence picture. For example, if we see a piece of equipment and we have that piece of equipment linked to only certain units, we can now start to try to verify are those units present. If we see a certain unit on social media and identify the insignia, and we already have foundational intelligence that tells us what type of equipment that unit operates, we can now start to infer the units that are in the area. So the relationship, and I'll come back to this often between the foundational intelligence allowing us to exploit current intelligence, and then that current intelligence then circling around to change the foundational intelligence. Current intelligence is always eroding or chipping away at the foundational intelligence, so we always have to stay on top of that current intelligence. And then, of course, current intelligence eventually becomes foundational intelligence in that old military activities go into the past, they form a pattern of life picture that then again we can exploit in order to understand if a new event of current intelligence is anomalous in any way. So that relationship is really important, but you need that foundation in order to act on anything that you're seeing now. Yeah, I think to build on that, both of those points, some of the things that we can pull through from a James News perspective. For example, if you're looking at runways and infrastructure, you can use that from an OSINT perspective to try and identify new equipment that has turned up, perhaps somewhere in the middle of Africa, and say that wasn't there a month ago, and then work back that draws from the foundational intelligence, but it also then feeds back when you guys are looking at what units were present there. Well, actually, we managed at this time and place to freeze that picture in place and got some facts in that snapshot. So how do we do this, then, thinking about the tradecraft of foundational intelligence? You've touched on a couple of important points there around verification, pattern of life. It'd be good to kind of explore this in relation to the different types of foundational intelligence that your teams look after. So Tom, coming first to you, how does this work in the equipment world? What does your tradecraft look like? Yeah, I mean, so look where we are right now. So bread and butter for the Defence Journalists, as in my team and in all of our James News teams, we'll be here looking at the new stuff, identifying what's important here for all of those reading James, and then also we'll be coming back to the stories. That is where we can stay with a story, and then that sort of, it comes from the leading edge of the news, and then it becomes part of the foundational data, and then it accrues in a way where if we didn't have all of those other parts of James, that would sort of dissipate that information in. It would, in a way, be lost, or it would have to be rediscovered, whereas there's a rigor here which keeps it properly organised and usable in the future. Yeah, absolutely. How about you Dylan, how does this map to your world? I usually think in terms of the intelligence cycle, of course, just like any intelligence cycle, we plan, and a lot of our planning will be based on current intelligence, things that we know are going on. The news team might tip us off to something that we should be watching. We have Zapod coming up here in Russia very soon, so we know we're going to get a wealth of foundational intelligence from watching that military exercise. We plan, we put together an intelligence collection plan. Next part, we process everything, and the processing of this is really where a lot of the connected data comes in use, because it's sort of self-reinforcing checks, and if you put something into your foundational data that doesn't quite fit, that tells you two things. Either you got a piece of data wrong, or something's changing. For example, we recently started seeing the Russian airborne units, maybe about a year and a half ago, starting to operate much heavier artillery than we would expect in an airborne unit. Now, when we first saw that, it was like, "This is an anomaly. Do we have it wrong? Have we connected our data in the wrong way, or are they operating in a new way?" Indeed, they were operating in a new way in light of the role that they have in Ukraine. Again, by connecting the data, recognizing the patterns, we could find novel insights. And then, of course, on the far side of this is the analysis side of the intelligence cycle, which is really about our people, and about the analysts that we always keep in the loop. That's a really important part of James, because AI and automation, we love it, we use it, but the analyst needs to be in the loop, because they're really the ones that have been able to deliver those novel insights for 130 years. If I may, an excellent expose from both of you, actually, just to pull out a couple of those threads that I think are really fundamental to foundational intelligence. One of the enduring nature of it, and you both mentioned it, is you have to be able to develop an understanding of what is normal before you can then dissect what is abnormal. So that's one really important element for me. Context is another one, so it's not just enough to see stuff and report on it. You've got to understand, and this is where the analyst comes in, and we might go down that Robert Robert Warren moment. Intelligence is great, but it still needs that expertise to go, "What does this mean, which is the so on, what if, if you like?" And you don't build that up overnight either, so I think that's really important. And then the final piece on that is the level of detail matters. Sometimes it won't matter that you don't know that that was a particular, you know, EW version of a certain type of vehicle, you know, the vehicle, but sometimes it really will. Understanding when it's important when it isn't is quite important, but when you do see something that's slightly out the normal, then you need to have that expertise, so back to the human element again, in terms of that is signature equipment. So for example, you know, it's not a secret that when we were watching the build up for the Russian invasion, it was only when they started meeting the field hospitals and the blood supplies that we realized that, "Oh, this is for real here." So that's the, that's, which is why there is a massive gray area between Foundation Intelligence and the analysis that comes as a result of that. So we've talked about the applications of that to trade craft in relation to equipment, military, military units, installations, kind of taking a step back from the defence side of things and looking at our security content too, so Pemissi country profiles, and actor analysis showing, "Could you talk us through the trade craft's approaches there?" Well, that's quite complex, actually, because that's a huge, so the Pemissi for anybody who doesn't realize it is a framework we use to basically deconstruct how a country works, what makes it tick, et cetera, et cetera. And again, you know, you use all sources you can, whether that's diplomatic sources, whether that is social media, you know, whether that is actually looking at things from a, for example, a satellite imagery perspective. And you've got to keep on top of all of those things, because really what you're looking at from that perspective is the, okay, how powerful is that country economically and politically, you know, what can it can't and can it can't do in terms of capability, but also what's its intent? And as we know, you know, most wars are won-and-loss by the economics and the weather, but generally by the economics, how can people sustain themselves? So I always go back to the example of Libya, when, although it was a reasonably high priority, nobody was looking at Libya when Gaddafi started, you know, getting extreme and Allied force happened. So we had to use open sources, publicly available information, to build up that deep understanding very, very quickly before all the, you know, expensive, exquisite collection capabilities were, were diverted from Afghanistan and other places to start doing the sort of, the near real-time stuff. Now that was challenging, don't get me wrong, but, you know, companies like Jane's at that stage were really valuable, because we didn't know what the all-bats were. What we had to do though was validate the all-bats. So obviously we were interested in surface-to-air missile systems. What do they have? Where are they? Well, what did we have? We kind of knew. What we didn't know were the surface-ability states, where they were located, because they do move. So, you know, we had to go through very much a publicly and commercially available scrape, basically, of everything we could to come up with that. So moving towards some of the recent challenges that we face in our day-to-day work, could you talk to us about this in the context of the equipment world? Yeah, sure. There's situations that happen, you know, Russia fires a Russian missile, or Houthi's fire an anti-ship ballistic missile. Often in those situations, as a journalist, the details can be quite limited, at least to begin with. There's maybe a couple of official statements, and you're thinking, what can we say about this? What can we reliably say about this? I'm very fortunate at Jane's, because there is more that we can usually say with reasonable confidence and be able to say with how much confidence than many others. So for example, we can look at the OSINT with various types of anti-ship ballistic missiles fired into the Red Sea, but we can check back and say, for example, aha, this particular missile, which may have been renamed by various groups three different times, it has a particular fin on it that was seen in a parade that we noticed two years ago, everyone else moved on. And we realized, aha, that's actually this missile, and that means it came from this supplier country. And we can put that together, we can't necessarily always answer the questions. We can say, this is what we know, and help our audience to fill in the gaps. Yeah, if I could follow up on that, what he was speaking to, similar challenges we face. It's a hard job. There's a lot of secrets out there. The only way we can overcome that is by persistence. We have been persistently monitoring this stuff for 130 years. We've been persistently gathering Russian insignia off of social media for the last eight years. And that has given us advantage in monitoring the forces after the war in Ukraine started. So persistence is so important to everything. You can't just decide that you're going to do a research project to put together the Chinese Order of Battle in a month and publish a book. You need to be monitoring it every day. And the second thing is novel use of the resources you have of every little bit of intelligence, social media intelligence, imagery intelligence, and applying that in really novel ways. And this is where you rely on the people again. For example, the ability for us to actually use synthetic air, synthetic aperture radar, and see the echoes and the shadows in SAR that is caused by the early warning radar systems associated with SAM sites, that was something that was a novel discovery a few years ago. We've been able to use that in order to find surface-to-air missile sites. And recently we used that to discover that a Russian airbase had been reactivated, an airbase that had been idle for 20 years, was now in use again, so thus, again, building the foundational data using some current intelligence and adding a new installation to our dataset. Just pick up on the reverse engineering piece of that, because it's really important actually. So, great example that I use quite often is the, it was an aircraft, a Russian aircraft shot down in the early days of the crisis, and there's quite an iconic photograph of the tail playing actually in a building. Now, very quickly, because the actual tail number was already referenced from an ORBAT, it was very easily to go, right, that is SU-34, but it's from this regiment, by the way, this regiment was actually in Syria up until a couple of months ago, and we now know that it's resubordinated to this particular base in Russia. Therefore, you know, it confirms that they've taken some of their best pilots and their best trained organisations, and they're using it in anger there, but it also confirmed what it was. So, that, you couldn't do that without that baseline intelligence to start with. Great. And before we move towards our takeaways from the session, one final question from me. So, yesterday we talked quite a lot about the role of technology and AI in particular. So, as technology continues to advance, do you think this will affect decision makers' requirements for foundational intelligence, and also will it affect the way that foundational intelligence is produced and used? So, the first thing to say is that it doesn't affect the requirement, because the requirement, for all things reason we said, is absolutely there. What I think is quite hard for the analysts to do is to get across to people who live in a very busy world, the instant gratification cultures I talk about, to actually put the resource into doing that more methodical bit, the bit that takes a longer time, because it is time-intensive, it's resource-intensive, et cetera, et cetera. So, the requirement is absolutely there, and we ignore it to have that peril. But things like AI are there to help. There's a great example that I've just been looking at right now. So, when we are doing order of battle analysis, it's quite traditional to go into commercial satellite imagery or other imagery, and go one, two, three, four, five, six of those, ten of those, et cetera, et cetera. The technology is now to the extent that you can do that in an automated way. So, automated object recognition is well developed now, and with something like a 99% accuracy, as long as you've got the training data and the images to go with it, you can just go, okay, that's the order of battle of that particular airfield. You can equally scale it. A great example is Russian strategic aviation. Everyone's doing it. So, we can build up the entire air order of battle for this Russian strategic aviation in an automated way by making sure the satellite goes over at whatever period obviously you want it to, running the algorithms and saying, right, there's that. But it can go further than that as well, because if you put in the right alerts, say every Tuesday morning, one or two of these aircraft go and do a training evolution, you know that all that's going to be reduced by a couple, probably, during this time. But say if you set the parameters, say, right, if five aircraft now are disappearing, you can put an alert, and then the analyst can get onto it and go, something has changed. So, it's that change detection that is significant. You can do it the other way around as well. You can be observing airfields where there is nothing there, and suddenly, you know, two T22s turn up. You go, right, okay, something significant has happened. So, AI can help in terms of deciding what's normal, doing the very manual heavy work in terms of just identifying the stuff that, okay, a good analyst will be able to do it quickly as well, but not at scale. And then alert the analyst at the right time when they need to engage. So, it's back to the human machine teaming piece. And they say, right, okay, now the analyst can get involved, say, right, so what? And with their background knowledge, they go, yeah, okay, yeah, we know that there's actually a couple of aircraft that have been up for modification, or equally, oh, hang on, we think they might have deployed to Syria, let's go and have a look there. So, that's the way that automated stuff can help. But we ignore foundational intelligence and our payroll. Great, well, as we move to the close of the session, I will now ask our panelists to leave us with one takeaway from today's discussion. And then we'll open it up for a bit of audience Q&A. So, there will be time for a question or two if we have any from the audience afterwards. So, Tom, would you like to lead us off? What's your takeaway from today's session? Yeah, I mean, I will leave you all shortly to go back on and wear out my shoe leather on the floors of the defense show, which is what we do. I think it is important through AI, all of the amazing stuff that Sean's teams and Dylan's teams can add to our reporting is that we are part of this defense world, James, and we talk to people. We talk to people every day from governments, from companies, and we know a lot of what we know through this wonderful old-fashioned way of talking and doing proper journalism and analysis and understanding that from the access that we get and, you know, we work hard to keep throughout this world. Great, thank you. Dylan. I'll just reiterate where I started, really, where foundational intelligence is essential to fully appreciate and understand current intelligence and fill in all of those gaps that current intelligence is going to have in it. You, metaphors are, you know, you use them at your peril, but you can't understand earthquakes without understanding plate-tech conics. You need to have a foundation in order to understand what's going on in the satellites in the activity on the battlefield. So, again, understanding that relationship, appreciating that relationship is the most important thing we can do. Great, thank you. Sean. And I'll just triple down on that, really, say that you ignore foundational intelligence at your peril, but what that means is that we owe it to the decision-makers, and particularly the people with the deep pockets and the money, to articulate why it's so important, why it's resource-intensive, and why it does need to be invested on, because without it, all the, you know, the good stuff, the headlines just won't happen. Yeah, great. Well, a couple of things stood out to me. One was the importance of the continuous monitoring of trends and events over time. I think that really kind of came out from all of you. And final thing for me is that what you see as the finished piece of intelligence is really just the tip of the iceberg. There's a whole lot of really good tradecraft underpinning it all, which takes a lot of time and knowledge and expertise. So, yeah, that shouldn't be understated. All right, well, that's it from our panelist, but over to you as the audience. So, are there any questions that anyone would like to ask? So, that question was, so, James prides itself on foundational intelligence. What makes James stand out in comparison to some other companies in this regard? We love examples. So, recent example is Ukrainian forces talked about using a new way of refuelling, maintaining some of the aircraft that they've got. And it looked like a pretty, pretty simple, almost logistics kind of bit of information, almost a footnote. But we were able, because we have stuck with the trends throughout defence over decades, we recognise that this was part of what many people call ACE, Agile Combat Employment. It's an old Cold War thing where you disperse your aircraft so that they can't all be destroyed in the first few minutes of a conflict and that this was evidence that Ukraine was learning and perfecting some of these techniques, which are being done now across Europe. People are dusting off their old Cold War handbooks and learning that again. We were able to connect those dots where I think many others would have just seen quite a boring piece of logistical information. I could add on to that as well by comparing us to one of our largest perceived competitors right now, which is AI. And our differentiator there really is our tradecraft. And I'll highlight in particular our sourcing. The fact that we have a huge foundational database of sources that we go to, that we've rated, that we know if they're biased, we know if they're accurate. So we know where to put them in our calculations of the ground truth. Whereas AI in contrast is prone to hallucinations because it doesn't rate these sources in the same way. So that's an important part of our tradecraft that I think is a differentiator and it's appreciated because it's in line with intelligence community directives. And so our work can go into their workflows in a much easier way. Yeah, great points. And I'll just say, you know, it's back to the data. It's all about the data. You know, we talk about democratization of data. It's a dreadful phrase because not all data is equal. Some is much better than the other. And if you go back, I think James, 128 years or whatever it is, that data has been built up, validated, verified, assured and developed over all this time. So you've got a historical trade to work out how things have developed. I have to say that actually AI is very useful in terms of managing that. You know, it's a tool that's absolutely not driving it. And I think that human machine team, which I mentioned earlier, is really important this, but you've got to have that data to be able to come up with the good stuff. Great, thank you. So I think that's all we've got time for today. But thank you so much to our panel, to our live audience today, and of course to our podcast listeners. Thank you and goodbye. Thanks for joining us this week on The World of Intelligence. Make sure to visit our website, janes.com/podcast, where you can subscribe to the show on Apple Podcasts, Spotify or Google Podcasts, so you'll never miss an episode.