Understanding Army Futures Command’s Priorities
Tune into this webinar to learn more about Army Futures Command and how they assess and implement tech!
Riya:
Thank you everyone for joining this afternoon or this morning, wherever you are, to hear some more interesting insights and experiences from our two awesome panelists today.
So I want to get straight right into hearing from our panelists, so I’ll keep my introduction very short and sweet. But I’m Ray Patel, I’m the managing director of the government team at Dcode. So I think everybody probably is familiar with Dcode if you made it to this webinar, but our goal really is to be a revolutionary advisory firm that operates between the nexus of the tech industry, the government community, and the industrial base to drive better tech into mission, period. And my team’s goal is to specifically help the government be better enablers and partners of commercial technologies to include some of you on the conversation today. So I would like to introduce our two awesome panelists.
We have Vito Errico who is in charge of the Army’s Software Factory within Army Futures Command, and Matt Benigni, who is the Chief Data Officer of Army Futures Command. So, gentlemen, I will ask you both to come off mute and do a quick introduction of yourselves and then I’ll dive right into some questions that I already received from the audience. And the way I’d love to make this time we have together really be about questions and areas that you all on the call are interested in.
So throughout the entirety of this hour, please drop in any questions you have in the Q&A and the chat. I’d really like to be driven by what the audience is interested in and would like to learn more about. So without further ado, Vito, I’ll turn it over to you first to do a quick introduction of yourself.
Vito:
Yeah, sure. Thanks. First of all, thank you to Dcode and the whole Dcode family for having me and it’s a great partnership with you guys. You guys are obviously actively involved in what the Army Software Factory did here in Austin for the kickoff of South by Southwest where we had the Dep Sec Def’s comments and we had just a great participation from across DOD, Space Force, Navy, Air Force and then of course Army. But now a little bit about myself, active duty Army officer for about the past 16 years now and just an avid technologist and I really enjoy bureaucracy.
Riya:
And with that, I’ll turn it to Matt Benigni. Let’s see if he enjoys bureaucracy as much as you do.
Matt:
Hello? That’s what got us all gathered here on this call, right? So I’m Matt Benigni, Chief Data Officer for Army Futures Command, and I lead our data and decision sciences directorate, basically trying to enable the command to leverage data to make better decisions faster. So like Vito, I’m a little bit longer in the tooth. I used to have a head of hair as nice as his probably, if not better, but I’ve been at this for about 25 years now and then changed from driving around on tanks to trying to do cool things with data about 10 years ago.
Riya:
Thank you both so much for the quick introduction, and I’d love to start with giving the audience a little bit more flavor about some of the things that both of you mentioned in your introductions too.
And Matt, we can start with you since you’re already off mute, but kind of understand a little bit more. This conversation is focused around understanding a little bit better the tech vision for Army Futures Command. And I think both of you have such unique experiences and ways in which you shape that tech vision. I would love to hear, a little bit more kind of about your current role and the capacity that you play and how you define a tech vision for Army Futures Command and how you shape it in your current capacity.
Matt:
Sure. So like I said in the last question, my passion is getting our operational leaders to learn how to employ their technologists and their technology more intentionally. So I think for a long time within the Army, within the Department of Defense writ large, we’ve kind of viewed data and analytics almost like a procurement problem. If we buy the right technology or if we hire the right person, poof, we’ll have data science. And in practice I think we’ve seen an industry, and certainly in a few areas in the military, we’ve seen that really this is about business leaders, or in our case operational leaders, employing these technologists and technologies to deliver business value.
So I was fortunate enough in my last assignment to be in the special operations community where they really lean forward in getting teams of technologists forward, deployed in Afghanistan, in Syria, in Africa, and really right alongside the operator, and what we saw was real, what I think was the start of digital transformation, just the level of conversations we had with people that really started out not interested in technology really became very intentional in its use. And I think we started to see the culture change.
So we’re taking a similar approach within Army Futures Command along two avenues really. First, we’re transforming, how we modernize. So, bringing our business processes into a more modern data pipeline, data-enabled employment of technology.
So we view digital transformation really being more about adopting and gaining mastery of practices and technologies already established in the industry rather than making our own. And we’re starting to see that maturity happen.
On more of the warfighter mission so that the technologies we’re trying to deliver to the force, that is what Army Futures Command is in charge of, the force design for the future. And Vito’s going to give you some great insight into that. But also what are the core technologies we’re going to need in 2030, 2040, those timeframes? And along those efforts, again, we’re investing in people and platforms, but those two are tightly coupled, right? They have to be employed in a way that delivers capability. So that’s how we’re thinking of technology moving forward.
Riya:
Thank you, Matt. That was awesome. And Vito, I’d love to hear of your experiences too and where you’ve seen something similar to what Matt expanded on. I think the Software Factory is such a unique and important part of shifting not just the kind of technical component of tech vision and modern business practices, but also that culture piece. So I love to hear your own experiences and time at the Software Factory.
Vito:
Yeah, I mean, I think all of us have a perspective that’s basically an aggregate collection of our cumulative experiences. So for Matt, it sounded like his position and perspective is shaped by how he saw the Army employ tanks in a combined arms environment. And then what he was able to do with that, he’s too humble to mention his PhD at Carnegie Mellon, but I’ll mention it for him, and then how he was able to take those lessons and that academic background and leverage it for the benefit of the Army.
I think if you look at Colonel Benigni, he’s a great example of what we’re trying to do for the Army and underneath the auspices of AFC as we, what’s called prototype of future force design of soldiers with specialized experiences, aptitudes, emotional intelligence who can form an operationally minded army unit rather than just individual contributors to go aid and assist either commanders in the field or commanders in Garrison. Obviously, we’re in the military, so military unit, to speed up their adoption of tech best practices, whether that be decision support or data visualization process or automation that occurs with a level of agility in the last, what we call, tactical mile or where the acquisition system might end.
Something that shaped this idea, to answer your question maybe more directly, was at least my background as a scout pilot in aviation, in army aviation, and it really it’s manifesting itself as we see right now with Russia’s invasion of Ukraine. A lot of people don’t know the Russian way of fighting warfare is very hierarchical. Command decisions are made centrally, and people at the lowest levels are not authorized to make decisions and problem-solve.
Conversely, the American way of doing business, or fighting war I should say, is to make decisions and problem-solve at the absolute lowest possible level. That’s really what Software Factory is about. That’s what AFCs tech vision is all about in my view. How do we empower and enable our own workforce, our own soldiers, to solve problems for their commanders as rapidly as possible? Because they’re going to know those problems best. And on a battlefield where we think it’s increasingly technical, increasingly lethal, the commander that makes the faster and better decision is going to be the commander that wins. So that’s how my personal experience is, and I think Colonel Benigni’s personal experiences have shaped what we’re trying to do here.
Riya:
I’m really curious because you know, just touched on this and then Matt you mentioned kind of forced design as well and understanding and identifying core technologies that Army and Army Futures will need to be looking at in 2030 and beyond. And I would love to learn a little bit more about what that process looks like and how connected it is with soldiers on the ground or how do you determine what that technical roadmap looks like and then how do you prioritize that? I’ll open it up to either of you, whoever wants to take a crack at it first.
Matt:
Well, I’ll start and then hand it to Vito for cleanup because he is been in the command about five times as long as I have. But I think you have to draw a couple of distinctions when you answer that question. So if we think about what are the major combat systems that will be most important in 2040, there’s research and development, S&T efforts, and to build the equivalent of the next Abrams tank, that is a farsighted endeavor because industry’s likely not going to deliver that to us. Luckily there’s no market to deliver that, right? The Army and our modernization enterprise has done that for decades, albeit in a much more industrial based way.
Now let’s draw a distinction between that and how useful software typically gets delivered to the market. And that’s a much more agile, iterative, let the customer evolve the product through consumption, type of endeavor. And then there’s efforts that are really kind of in between. So how do you integrate AI into existing systems for tasks like let’s say performing reconnaissance in a fully autonomous fashion? That probably sits right there in the middle where market’s probably not going to deliver that, but we probably have to do that in a much more iterative way than let’s say building a glider that can persist in the stratosphere or something abstract like that. Vito, over to you.
Vito:
Yeah, I’ll agree with just the overall sentiment of that. When you think about the scale especially, we’re talking about DOD here, right? So the scale of DOD or the scale of one of the services, we’ll use Army since we’re Army, but it just sort of prevents the use of a more commercially minded go-to-market strategy. It’s a little bit different than conventional R&D because of the amount of stakeholders and public oversight. So what we’re seeing is that there’s got to be some framework for Army, big Army, to call out these technologies that everybody is sort of rallying around right now or best guess technologies for 2030, maybe we’ll say. Hard to say best guess for 2040. And then to come up with a roadmap for when units will get touch points with these types of technologies. Then the Army’s got weigh that with ongoing priorities and contingency operations in Eastern Europe as we just talked about.
So it becomes a sort of organized way to get soldiers in front of technologies and probably not fast enough than anybody would reasonably want. But when you think about the types of priorities and the types of requirements placed on the larger army all at once, I mean, it’s probably as efficient as it’s ever going to possibly get. Now having said that right, one of the key benefits of Army Futures Command is that it speeds up the frequency of those soldier touchpoints faster than they had been before there was Army Futures Command and then one of the things in concert with Colonel Benigni here on the line and then us at Software Factory, we can informally get technologies in front of soldiers at Software Factory, at the Army applications laboratory. We can get these newer technologies in front of soldiers in informal ways faster than the larger process will allow.
That doesn’t guarantee a transition of that technology, but it certainly doesn’t hurt. And I think that we’re finding that there’s a little bit of an alchemy in this relatively new approach of just showing off what’s there for soldiers and seeing what sticks or what gets a lot of momentum.
That’s as close to maybe mirroring the best go-to-market strategy that a commercial vendor might be interested in.
Riya:
Super interesting, and I’d love to expand on that a little bit more, but I also want to make sure that I am covering the questions that we got from folks before the webinar are even started. And what’s funny is, Ronen, I was going to start with your question and you already dropped it in the Q&A, so thank you. I think this question is relevant for both Matt and Vito. So I’ll maybe open up to Matt to start and Vito to provide any additional contributions. But at last week’s special operations conference, SOFIC, there was a lot of discussion around data sharing, data collaboration and data privacy both within the DOD and with our international partners. What is Army Futures Command’s perspective on this topic and how are these challenges manifesting themselves today? And is the group looking at specific technology solutions to solve it? There’s a lot of questions in this, I’ll stop there. But they also were curious about challenges and successes you’ve perhaps seen with exploring some technology solutions to this problem.
Vito:
So yeah, I speak probably with a little bit more authority for Army right now. There has been a tremendous drive inside of Army to catalog authoritative data sources to set up common data standards, to set up common ways of just processing data, of sharing one’s data in ways that unfortunately had not been embraced before. I’ll just probably leave it at that. You could imagine a variety of different reasons why not just the Army or DOD, but any organization with a bunch of different sort of stakeholders would not maybe gravitate to doing that all at once. So that’s been something that not just Army Futures Command, but really the Army G6 and the Army Chief Information Officer. And I think you’re going to see the new Army leadership or the relatively new Army leadership, the undersecretary, Camarillo and Secretary Wormuth really start to push on breaking down what can be sometimes cultural barriers or sometimes organizational barriers that prevent us from better data sharing.
I would also add that if you go and look at Secretary Wormuth’s security fireside chat and her message to the force, I think it was back in February, she published basically six priorities that the army can focus on. And number two, her number priority for the United States Army right now is to become a more data-centric army. So in classic fashion, everybody is now becoming far more cognizant of what data is and what it isn’t, and that’s a good thing and a bad thing. Under the interesting things that I’m happy to report that are going on about data sharing, a lot of times we hear about it in terms of an operational setting, and you’re seeing commanders in the field dictate new, or what I would call, not new practices, but basically lowering the classification of their data so that more partners can be able to use that data.
I think very clearly, the United States of America, it’s way of waging war is to do that as a coalition partner with coalitions. To be able to operate effectively as a coalition, you need to be able to share data fast. And I think we’re seeing right now a lot of the impediments of unexpected contingencies, the things that are preventing us from sharing better with our partners. And so you’re seeing for the first time commanders mandate, American forces go to more partner minded networks for everything, not just bespoke partner operations. And I think that the spirit of the question is not just about data, but about how do we handle our own information in a way that is more consumable, whether that be to a client or to a soldier or to sister organizations? I hope that answers the question.
Matt:
I will add. So in my last job we straddled the intelligence community and Department of Defense. In terms of information sharing, I don’t view this as a technical problem, this is a governance problem for the most part. So the 9/11 commission mandated that the intelligence community improve its ability to share data. So that really forced them to mature in terms of governance and infrastructure to have all of their data tagged appropriately so they can control who sees what. They’ve got basically a very robust identity and access control measures that allow you to just see the things you’re supposed to see.
So I think we’re going to lean into some of that government experience as we look to employ those things. Something I’ve seen in a deployed coalition environment, actually partners would often move their information up to the top secret networks so they could share it more easily because the infrastructure was there, ironically. Some of the technologies I’ve seen that are helpful are technologies that kind of lower the barrier for implementing group policies and access control measures. Immuta is a company that has had some success in that space. I know there’s several others.
One other area I would speak to though is as we have more technology proliferate the battlefield, which I would say is Army unique, the amount of machine-to-machine communication we expect in our battlefield networks has gone up quite a bit. So with that, the standards needed to make that equipment interoperable–I mean that’s a growing area. So we’re having to look towards kind of open source community approaches. How can you embrace a more collective way to generate standards as a group and how do you govern that? How do you do it transparently? I love the InnerSource Commons model. We’re looking at the open geospatial consortium which has done that for over a decade, but those are some of the areas we’re looking in terms of governance to leverage best practices.
Riya:
Matt tangentially related to the latter point you were making, another question in the chat around what is AFCs vision for leveraging the private sector to process and exploit open source data with specific military informational purposes?
Matt:
We embrace that idea. I think right now open source information is a rich area to contribute a lot of the NLP kind of implementations that I’ve seen are effective. So bottom line is where we embrace and lean into that.
Riya:
I’d love to hear it. And we have another question in the chat that I think is a really interesting one and we’ll segue into some of the questions we got ahead of time, but how does Army Futures Command expect to engage with the newly stood up chief data and AI office within the DOD CDAO?
Matt:
So our primary touchpoint with them is our AI integration center located at Carnegie Mellon. So they really have two missions. One is to incorporate AI into our combat platform. So things like predictive maintenance, things like the autonomous reconnaissance task we were talking about earlier. But then also how do you democratize the ability to deliver AI solutions to current problems as well. So we’ve started to see some really interesting use cases recently around information operations where we have more mature platforms and more skilled army talent and they’re able to leverage source algorithms to answer critical questions with ongoing operations in Ukraine. It’s a super exciting thing to see where some of these AI integrations aren’t necessarily pushed through a procurement arm. They’re delivered in a much more quick to market type approach, which obviously is needed with the rate at which algorithms mature and become obsolete these days. But AI 2C is our primary touchpoint with them.
Vito:
Yeah, I would just say it’s a welcome position from our perspective because especially at the DOD CIO, DOD Chief Software Officer sort of level, we kind of count on that office to set a lot of, I would say, permissively minded policy for the rest of DOD to sort of follow off of, and ironically or those high-level DOD offices can serve to be the first movers and the risk-takers that the subordinate services are not going to have the authority to take. So the announcement of that role, that role underneath the DOD CIO umbrella and then especially so close with Dep. Sec. Def. Hicks who if you haven’t seen her software modernization memo from back in, I think it was February in the DOD CIO’s open source memo back in January, I think those are both publicly available off of the website. Software Factory’s benefited quite a bit from the policy and the rhetoric coming out of both of those offices who are just singularly minded on just broadening adoption and permissibility of these technologies.
Riya:
Super interesting to learn more about that and I think this ties into a question that we got ahead of time around just army futures and AI and ML use cases at large. And Matt you touched on this in your response around the relationship to CDAO. So Vito, I’d love to start with you and just learn from your perspective of specific AI/ML use cases that you have come across in your time at the Software Factory. The question also asked about what data types Army Futures was using, for example, computer vision, full motion video, et cetera, as well as just the process of procuring more AI ML commercial technologies. But start with the use cases.
Vito:
Yeah, okay. That’s a lot in that one question. So I think Matt mentioned it at the outset or earlier in the conversation, the use case, one of the big use cases everybody’s sort of getting thrust into is predictive maintenance for AI/ML with respect to how can we save money, like corporate America has started to do by applying AI/ML, it’s predictive maintenance use cases?
But you have to understand, there’s a lot that goes into that at enterprise to be able to do that well. And so I think whoever asked the question, I would just remind them, you can have these bespoke or insular opportunities to experiment with the technology. When you think about how data collection, data curation, data availability, and the number of disparate platforms the United States Army employs across its entire 1.1/1.2 million person force, it becomes pretty interesting pretty fast.
So a lot of that I think there’s some promising results, but we’re still trying to clean up the basics with respect to data curation and data availability. And then of course the hardware angles to that about what is the system on the system that’s recording that sort of stuff that can feed it up and help us work on that. Then I think I would also probably highlight the Army Test and Evaluation Center in Aberdeen, Maryland. They’d kind of got the lead for some of that, especially on digital twinning. And then the predictive maintenance stuff is really being spearheaded as a co-lead with the AI task force in Pittsburgh that Matt just mentioned with Army Material Command headquarters, which sort of owns sustainment and logistics for the Army.
Matt:
I’ll start out using taskforce maven, so that’s ODNI’s AI algorithmic warfare project. So that effort’s quite mature now. We’re really looking for a transition partner. But the way they’ve tackled specifically the computer vision problem is the government owns the data, the label data, and it allows vendors to compete in terms of performance for algorithms. So currently the 18th airborne core is partnering with Maven and looking at… basically, they lease the algorithm, evaluate the performance and a much broader community of stakeholders can compete. So I think we look at that example, and the more we can modularize the tech stack, the data services, the AI services, it just puts us in a much better position as we define the market as a service as to how we’re going to procure AI. I want to have benchmark data sets that are mission specific and allow as much competition as possible to show performance.
One of the unique things about our problem, and Vito spoke to it in his answer, so let’s take this autonomous or reconnaissance task as an example. To actually employ that technology out in a real combat situation, it’s likely going to be a new area. We are likely going to be contested across the various electromagnetic spectrums. So our architecture is going to include at the headquarters level probably commercial cloud resources. But as we move towards the edge, a much less connected, more bandwidth constrained sensor and decision making environment.
So we’re looking at a hybrid cloud edge infrastructure, but having an ML ops platform that can span kind of train a model anywhere on that architecture to deploy a model anywhere on that architecture is something we’re really looking heavily into specifically up at Pittsburgh at the AI integration center. So just kind of circling back, I think you had talked about evaluation as well in that question. I mean that is for computer vision on the battlefield specifically, that’s an incredibly hard problem. I mean those algorithms are usually built from some kind of transfer learning. It is a different environment. We are likely to employ them than how we’ve trained them in understanding performance where there’s weakness and where there needs to be retraining. I think that’s going to be part of the war-fighting art moving forward and we need an infrastructure and talent that embraces that.
Vito:
Real quick, sorry, just because as Matt was talking, I know people that are listening ideally for you guys, I think they’re hoping to do business with the government in some way, shape or form. Just an observation, I probably get 5 to 10 emails a day from somebody trying to sell me an AI/ML platform, and I would just coach people on that kind of stuff. That line, that product line, is pretty ubiquitous in the market right now and you want to think about what’s the product differentiation to the government, or what is the way ahead for that market? Is it going to become commonplace as cloud service providers mature their offerings as cloud-agnostic platforms or PaaS offerings grow their capability sets? For anybody listening that’s kind of in the product world. Just think about that from our perspective.
Riya:
I love it, Vito. Thank you for that helpful tidbit. It really contributes to our title. So I have two questions here in response to two distinct things that you both mentioned in your last response. So I’ll start with Ronen’s. In your hybrid cloud edge description, how do you envision protecting data and models on the edge while training/ tuning/ running in inference, et cetera?
Matt:
So one I would say there’s still some open questions there. I think that for a while we have thought about AI in the military like it’s an algorithm problem as opposed to a data problem. And I think we are starting to mature to the point that really the data collection processing and life cycling is probably where the biggest challenges lie, particularly in our military applications because a lot of times data scarcity is an issue we’re going to deal with.
Now this is just my opinion. I think we are going to mature over the next few years to where data backhaul is a much more deliberate thing we’re going after. So I think of the Tesla model where you’re given model three, I think it collects something like eight gigabytes a day of sensor data and driver interaction data, and when that vehicle either goes to a charging station, a garage that has the right kind of router or back to its house, it uploads that data to the cloud. That vehicle fleet of half a million vehicles contributes to their collective training data. They develop stronger algorithms for decisions at the edge and then those are deployed forward.
I think in the future we will have a robust data corpus that we use for training combat models that as our logistics pushes maybe take fuel out, they’ll come back with data and a similar model. Because I think what will win at the secondary phases of the fight will be whose models mature faster in the future. So in terms of security, clearly encryption at rest and zero trust principles and all of that type of rigor that I think industry is really leading the charge in will incorporate that. But the overall data life cycle management will become a much more meaningful part of what we’re talking about.
Riya:
The other question in the chat in response to your previous responses, you mentioned ML ops. Are there any efforts to tackle the accreditation of commercial cloud AI ML software as a solution to capabilities or is the focus only on government off the shelf?
Matt:
Oh, absolutely. Yeah. I think we can both speak quite a bit to that. So just as an example, my directorate has helped develop a secret accredited Azure cloud platform where we have intentionally chosen to, wherever possible, use platform services to ensure we’re going to be able to scale and have reliable services. So we’ve minimized custom work. I’d say the only custom work really is in the access to get to those resources in a way that meets our security accreditations. So I think we’ve matured enough to know that we don’t want to reinvent things that industry can build and are software tools that are refined by really large user bases, so on and so forth. Clearly there is user facing function that we need to generate that is specific to the military, but even where we are doing that, I think Vito, I’ll hand off to Vito in a second, but the platform services that underlie that custom user facing function, again, we are trying to couple as tightly to industry as possible. Vito, over to you.
Vito:
I mean, I’ll just tell you, you’re talking to the two right guys here. We are not the folks that think we’ve got to build everything ourselves and are certainly not advising our leaders to do that either. Why build if you can buy? Why buy if you can rent, sort of mentality here. I think if you go back and look at one of the memos I kind of cited at the beginning, the DOD CIO has weighed in pretty heavily on this already back in January. Very much pushing open source and commercially available technologies as the first choice. I don’t know how much government should really be in the business of engineering things from scratch unless it has a really good reason to do that. Usually, it has to do with classification now. But even that, that’s kind of a thing of the past.
I mentioned this a little bit before. I would be interested. I think the cloud service providers are going to start jumping into this idea of how they can have AI/ML offerings as part of AWS Gov cloud, or I don’t know if Azure’s calling it Gov Cloud. And then we’ll see what happens with Oracle and GSP. But yes, the bottom line is that we are actively trying to continuously accredit those types of technologies, whether that be IL4, IL5, IL6, and to build the packets of precedents and reciprocity so that when we do it for the first time, we’re not going to have to keep doing those packets all over again. Cause I think probably most of the people on the line understand the authorities operate processes is not the most fun process in the DOD.
Riya:
Feels like the understatement of the year, and I’ll let the audience react to that. So one more question on data. Would love to know a little bit more about how both from the Software Factory perspective as well as the Chief Data Officer perspective, how you process dead or unstructured data from documents, images, et cetera? Just kind of the process behind that.
Matt:
Yeah, so I think I remember that question. Am I right to say it was a bit more of a battlefield context?
Riya:
How do they aid in processing dead or unstructured data from documents? What about images?
Matt:
Okay, yeah.
Riya:
You can add in the battlefield context. I’m sure that will be helpful for the audience.
Matt:
There’s been quite a bit of work in the special operations community around capturing material. So special operations forces or proxy forces which end up doing most of the heavy lifting at least for the past few years for us, oftentimes they’ll bring in just reams of documents off of an objective. A lot of times with certain high value captures, things like handwritten notes in folks’ pockets, being able to upload those, machine translate, do those kinds of things at scale, there’s been quite a bit of work leveraging industry technologies to help with that.
We have got a few kind of data lifecycle management significant projects going on in the command right now. So DEVCOM, which owns most of our S&P, so labs and that type of expertise, has quite a bit of effort in a program right now called electronic product data management. Whereas an R&D effort matures towards transition becoming part of an acquisition program, how do you ensure the majority of that data can migrate along with the project in a way that’s useful to somebody who takes this from a research and development focus to a delivery to the field and scale those kind of things? So I view unstructured imagery data, those type of things as just another type of artifact that we need to get standardized in uniform and how we collect it so we can automate as much of the life cycling that is possible.
Riya:
Thank you, Matt. Anything to add, Vito?
Vito:
I mean, I think Matt’s got the best probably in the Army view on how that unstructured data has been handled to this point. What I can report for at least from the Software Factory hat is the Army’s got us not going after unstructured data. What it has is it has us better connecting to authoritative data sources that had not been well-connected to this point. So we at the Software Factory haven’t had much experience with the unstructured part. Not that we won’t. But to give you an idea of where the army is, we’re still trying to leverage a lot of the data that candidly we didn’t know we had, if that makes sense.
Riya:
Switching gears a little bit, we have a question about software supply chain and better understanding how it fits into the tech vision as well as are there areas that are particularly challenging right now with software supply chain?
Vito:
Yeah, so software supply chain falls under this idea of, this more overarching idea. You’ll hear an AFC and the headquarters department of the Army talk about tech protection. Same thing. I think you’re seeing it, especially in post Covid world, you’re seeing it increasingly factor into senior leader decisions and beginning to dictate changes to army policies and processes for a variety of different reasons. As far as our experiences at the Software Factory go, we go through pretty rigorous background analysis, cataloging of any third party dependencies that we take on.
So strictly speaking about the software, I probably can’t get too deep into the hardware. But strictly speaking about the software, you’re not just looking at the software supply chain in sort of a vacuum. You’re also looking at it with respect to, what type of environments are you operating in? Is the software just for IL2, IL4, maybe IL5, CUI type stuff? Or is it presumable that the software is going to jump up into IL6 and IL7? That’s a big part of tech protect and software supply chain, and just general supply chain security is what is the use of the technology? The lower you can get it, the more ubiquitous the product can be.
Matt:
I mean, this is definitely right in Vito’s wheelhouse. I guess the only thing I’d add is, I’ve kind of grown up as a data scientist. There’s plenty of jokes about data scientist code, but we really have gotten Draconian about dependency management. So I think if I remember this question came out of someone from ION channel. And JC Herz, who I don’t know if she’s still there or not, kind of demoed that capability to me a while back and to just really impressed upon us if we are standardized in how we build our dependency mappings, then automating the process of ensuring the kind of open source components of our software are being actively maintained. And if not in ensure that we’re trying to get those removed from our systems.
It is important. But again, I think this, again, speaks to what Vito’s team is doing in similar efforts is that the government has got to have competencies in this area. The ability to understand software security, good development practices, good data management practices is now a core competency if we want to be a data-centric army like the Secretary of Army has called us to be. So this type of supply chain management, especially with solar winds and the colonial pipeline, I believe both those were supply chain attacks. We have to lean into this. That attack space will be active, I think, for our entire career.
Vito:
You’re right. This push towards the open source, I mean there’s a supply chain element to that obviously, and there’s only so many people inside DOD and the services that understand what open source means, what it doesn’t mean, the difference between open source and open standard. So it’s very much a balance with the security side of things. I wish there was one easy answer to it all, but unfortunately there’s not.
Riya:
Thank you both first so thoroughly asking such a broad question around software supply chain. And that leads us to, I believe our final question.
Kevin, from Renewable Edge, I’d appreciate any information you can provide on specific pain points around energy availability and resiliency. For instance, I imagine that specific metrics and the performance of microgrids could factor into critical time sensitive military decisions, yet information overload might be an issue as well. So careful pre-processing and presentation would be very important. In general, how can we, and I think in general just industry, especially if energy is not your particular forte, better provide guidance on what soldiers need and how industry is setting a standard here? So big question on energy resiliency and availability.
Matt:
Yeah, I’ll start. So the first thing is don’t reach out to Matt Benigni or Vito Ericco for that, but do reach out to Army Applications Lab. So just last month they held an event called Vertex, centered around energy needs in the future for the Army. So I think it’s over a hundred companies. They had I think 25 different universities participate as well as members of DEVCOM. So our S&T community, the acquisition core and headquarters department of the Army.
So there was some discussion on microgrids, and I’m asking folks to promise for no follow up questions, but some of the major takeaways were kind of efficient fuel consumption and kind of a minimization of moving parts that we think that on a contested battlefield, some of the current solutions and industry look brittle.
So I would ask you to reach out to Army Applications Lab if anyone needs a point of contact. I can shoot one to you after this, but definitely leaning into this space. And the goal of that event was for us as a modernization enterprise to help put a more collective signal around where the S&T community is going for the acquisition community.
Riya:
All right. I think you also safely covered what Vito might have to say about energy. So I won’t turn into you, Vito, to drum up something additional around energy resiliency. But I just wanted to say, I don’t think we have time for one more question, but Vito and Matt, thank you so much for your generosity with your time and especially I feel like I just peppered you with a whole host of different questions running in the gamut of deeply technical to very broad kind of culture and bigger picture of what’s going on in the Army Futures Command. And you guys took it like champs.
So really, really appreciate your brain and your insight. And I think one thing that I took away that I hope the audience takes away too, and it ties to both of your introductions, is that we talked a lot about specific technology areas and technology at large. But Matt, I think you highlighted the importance of people, process and employment. And Vito, you also talked a bit in the beginning about culture and governance and that there is more to this tech vision and just implementing and adopting relevant and critical technology than just the technology in and of itself. So I took that away and I really appreciate you all sharing more than just your technical prowess and expertise with the audience.