Leveraging Enterprise AI and the Cloud Ecosystem Part 2

Welcome to “HSDF THE PODCAST,” a collection of policy discussions on government technology and homeland security brought to you by the Homeland Security and Defense Forum

 Our panel  shares how that shift to AI changes our process, our culture, and our platform. By embedding prompt engineers at the mission edge, we compress requirements into live solutions, gather feedback in real time, and turn early power users into a broader base of confident practitioners. Think Excel’s early days, but with governance, security, and scale built in from the start.

 Featuring:

• Jay Alalasundaram, Deputy Assistant Commissioner, Software and Applications and Services, CBP
• Baibhav “Bobo” Devkota, Executive Director, Border Enforcement and Management Systems Directorate, CBP 
• Thomas Mills, Executive Director, Cargo Systems Program Directorate, CBP 
• Neha Saijpaul, Director, Digital Solutions, CACI International Inc. (moderator)

 This discussion took place December 12, 2025 at 8th Annual Homeland Security & Defense Forum Border Security Symposium

Follow  HSDF THE PODCAST and never miss latest insider talk on government technology, innovation, and security.  Visit the HSDF YouTube channel to view hours of insightful policy discussion.  For more information about the Homeland Security & Defense Forum (HSDF), visit hsdf.org.

Transcript

From Curiosity To AI Demand

SPEAKER_01 0:00

You know, it's amazing to see the transition. Um, earlier this year, when we we are OIT, our job is to provide solutions. And as we were working with these um various mission groups, they we would be like, hey, can you you can you add AI to it? And now, you know, towards the end of the year, it has reversed. They come to us and they're like, hey, add AI to it. So it's great that everybody is seeing the benefit and the speed is really helping. We have been able to do POCs rather quickly with CodeSys, so all in all, great. But XD Burble, I know you talk about this concept of prompt engineers or AI field engineers, which can really accelerate AI adoption. So can you elaborate on that?

Building A Governed AI Platform

SPEAKER_03 0:42

Yeah, so um there's there's oftentimes a requirements, you know, when you're doing traditional software development, there's the requirements gathering phase. And and with generative AI, it's almost that it really is where the requirements are your solution, right? And so um as we as we kind of go forward into how we meet with the the customer and and and the mission stakeholders and identify um A what the problem is and how to solve it, there they're having a skilled group of people who are closer to the mission, um, and actually deploying our our some some of our team out directly into watching the day-to-day of the mission, it's almost solving the problem in real time, right? So with prompt engineering, you you go out to the mission, um, you're watching what they're doing, as they're saying it, as you're seeing it, you really solve the problem real time and you deliver the problem real time, you get the feedback in real time. Um, and and and to to DACJ's uh comments earlier, that the I think if we if we're ever able to get to a situation where we can upscale our staff that are doing that first touch labor with that kind of support, um I think I think you're going to see adoption of this technology go uh through the roof. And and as uh CTO was was mentioning earlier, we're really in a transformational uh place with this technology. And um, I was discussing earlier how it's kind of like giving uh people Excel for the first time. So when when when everybody got Microsoft Excel, there was one person in the room who was just like a whiz at it. And you would go to that person, right? And then they would do all your fancy Excel. And and nowadays that's what's that's what AI is going to be. We we hand it over to the mission, there's going to be uh a few early adopters. Um and and our our goal here really is going to be to kind of professionalize that into the mainstream Excel.

SPEAKER_00 2:47

I think there was even a question on uh from on the previous panel about power users and and users that are just beginning to explore AI. I think that's a perfect example of it.

SPEAKER_01 2:57

Yeah, we all are going to be jack of all trades because we got AI with us. So building on this, there's also a technical side of enablement. Um, common platforms can reduce to applications and accelerate delivery of AI capabilities. Um, XTBobobo, we have seen really strong leadership from you in advancing implementation of CP Common AI service. Can you describe some near-term vents you've had and long-term vision you have for the platform?

Mission-Driven Use Cases And NII

SPEAKER_03 3:24

Yeah, so um I guess the kind of the IT department's job is to make the technology available. Um, and so I was again we we were able to leverage the infrastructure that Dak Mays and Mark James and the ECSD team built. Um, so they provide the infrastructure. The CTO's office already has the governance built into the AI that we're using. And so uh kind of in this traditional microservices architecture, we we just essentially built a microservice on top of that. Um and so in that we're inheriting the governance, we're inheriting the cybersecurity, uh, everything we we we rely on the CTO's office and uh and the cloud team for for the security, right? So so that we could just go directly to the customer and facilitate the use cases. So uh working directly with the chief AI officer and kind of capturing those use cases, we're able to as we capture those use cases and and enhance and kind of continue to build on on the platform that we already have, uh we're we're building on all of the regulatory checks, right? So we certify for PII, and uh we're we're in the AI use case inventory. And so so we're checking all of those boxes, and as we check more and more, the platform becomes uh more able to support any uh use cases that that that the mission brings. And um, and AI, quite frankly, is just going to be a a vision where the use cases come in more and more rapidly. Um and and we would just be able to deliver faster.

SPEAKER_00 4:57

I mean, one of the reasons for our success is that we we don't do things for the sake of doing it, right? We don't want to just say AI because we're doing AI. Our mission partners have been super educated about this, and they're the reasons why the use cases are coming to us, right? It's it's mission drives what we deliver oftentimes. And that's what XD was talking about, right? So it makes it easier for us to implement those solutions because there it's the demand is coming from them.

SPEAKER_01 5:24

Very well said. Um we talked about enablers um for AI and how fast everything is moving. Now let's talk about how it actually becomes real operationally. Um so how do you prioritize when to invest where to invest first to balance mission urgency with long-term enterprise value?

SPEAKER_00 5:43

I mean, that's the key, right? So we just talked about it, where uh our mission partners drive where we need to focus on. That's where the enterprise AI comes into play. Right? So where CTO CNL talked about chat CVP in general, right? That's a tool that's available to all the individual users. Where do you take that AI models and apply it towards specific use cases like CKS, which uh you guys built for our OPA team? Several use cases, like NII is another one that is a big huge priority for CVP. CTO Mother Guru touched on it, BIBAFS team is working on it. But as a software organization, every one of my team is going to take uh advantage of the benefits of NII, whether it's in the passenger processing space or the cargo processing space. That's where some of those specific use cases come into play, and that's driven by our mission partners. And they give us those requirements. Having enterprise chat and enterprise uh AI is one aspect of it, but the individual use cases is what's gonna help us with the from a mission standpoint.

SPEAKER_01 6:44

Yeah, so building upon that, how do you how do you measure success? How do you decide whether to take this proof of concept forward and scale it to the enterprise? Like, what's that litmus test?

Measuring Proof Of Value

SPEAKER_00 6:55

Yeah, and that's that's the NII is the perfect use case for it, right? It's working with our mission partners and measuring the successes based on time saved, right? Baiba talked about human-in-the-loop architecture that we built, like CTO's organization set up the organization structure. It's as we look at OIT and gender, like it's a very cohesive unit. We have infrastructure teams, software teams all working together under our IAC's leadership with the CTO organization embedded and providing the architecture. So we don't look at a proof of concept. I never, I don't like the term proof of concept. We look for proof of value. Anytime we invest in something, we want to look at proof of value with scalability in mind. Like how do we take this and how are we gonna operationalize this in a real-world environment and then be able to measure the science time savings from the end user standpoint? How much time is it gonna save for our officers and agents in the field? Some of the technology that we invested last year is uh seamless border entry on global entry, leveraging TVS as a as our uh facial biometric service. People can buy global entry travelers now can get off on select airports now, and that's still a pilot, um, where we get off the airplane, they can walk through with the cameras placed strategically and can process people on the move without them having to do anything about it, right? Those are some of the technologies that we are investing in, and we're actually measuring a success based on how much time it saves for the officers. How does it improve our national security? Right? And those are the areas, and that applies to the trade space as well. And uh, and XD Mills can point out some of the examples in the trade space.

Trade Analytics And Risk At Scale

SPEAKER_02 8:28

Yeah, so I think uh as far, you know, the good thing that we're seeing with uh the adoption of AI and the adoption of uh these cloud technologies is that the uh proof of value, that the the proof in whether it's good or not is happening quickly. Um and I'll give a good example. Uh, you know, in the past we looked at uh shipments very transactionally. We focused uh more on the shipment and and and the container, what's in the container and the goods. Uh what AI, what analytics has uh given us the insight uh and ability to do is to focus on the entire entity. So we're not looking at just the uh shipment anymore. We're not just looking at the goods in the container, we're looking at all the parties involved in the goods. We're looking at uh uh you know who manufactured it, where it was manufactured, uh, what materials were used in the manufacturing of that, um, and and being able to build a holistic risk assessment uh that we we we couldn't have done in in the past, right? Um because it's a huge amount of data. You're talking uh, you know, lots, lots of data that's coming in. And uh in in the past, very transactional. It was tied to entry of goods, uh clearance of goods. Uh but uh what what we've been able to do now is to uh use predictive analytics for risk uh assessment. Uh it's really about identifying those high-risk shipments, high-risk entities, um, and uh and you know to get a little bit more technical, you you you make it harder for the uh bad guys uh that you know now because we're we're able to find them a lot better uh to import goods and easier for the good guys to to bring in goods, right? Um and uh what we've also been able to do uh with um uh within our trade analytics workspace is uh now being able to do automated document analysis, uh automate our anomaly detection. Um I I think you guys have heard uh the whole uh de minimis concept is is gone now. Uh so uh you know, low-value uh goods, low-value shipments, which have traditionally been uh one of our blind spots, uh we we shut that down and we're able to shut that down uh quickly because of our uh ability to leverage a lot of these technologies. That that that that's you know, we we were given that mandate. Uh we we were not given much time to implement that mandate, uh, but we were able to do it. Um uh in a large part because of what we've got now. So I think as far as like measuring the proof of value, um, you know, what what what's good about uh AI, what's good about cloud is um that we really can get to the point of uh failing fast. Um, right? We know right away, hey, we we've run something, we've put it into our algorithms, and we got a result. Um or we we didn't get anything, and and we know it uh rapidly. We just we just in in the past it was iterate, iterate, iterate, then you know you could waste months in in looking at something and find out, oh, there was really nothing there. Uh but we're seeing the results uh pretty pretty immediately.

SPEAKER_01 11:34

That's awesome. I thought we were just getting started, but I got the indicator that I need to start wrapping up for QA. So sadly. But before we end, looking forward towards 2026, um, if you had a magic wand to take off one barrier uh from progress, what would it be?

Barriers, Change, And 2026 Readiness

SPEAKER_00 11:52

Let me put my handy potter ad on. No. Um no, it's it's technology moves faster than the speed in which we can implement it, right? There's always barriers, whether it's uh it's technical nature or uh our our money or when you're a CR contracts. All of this takes time. And if I had to remove all the barriers, we have an exciting year coming up, 2026. One of the biggest sporting events in the world is gonna happen in our country, soccer World Cup, right, or football, right? So it's gonna be exciting times for our country. And we need to be able to get those technologies that we're building in CPP out there to help our officers and agents to be able to process people coming in. It's not just gonna be athletes coming in, right? It's gonna be athletes coming in with a bunch of equipment, support staff, and then there's gonna be the audience that are gonna, it's a global event, right? And it's an exciting time for our country, and people are gonna be coming in. And we need to be able to give our officers and agents the tools that they need in 2026 to be able to be able to make this event a success for our country. It's a it's a question of national pride, right? And it's our job to make sure that we are able to give them those tools that they need. And I would say that's our focus, and uh and we would be willing to work with our mission partners and and industry to remove some of those barriers that you talked about and see how we can make this successful.

SPEAKER_01 13:12

XTs, would you like to add something?

SPEAKER_03 13:14

Uh no, so I I think uh I don't know if I would ask Congress, but uh I think uh, you know, I'm really looking forward to to manage it, change management, right, with with the actual um with the end users and really kind of adopt AI, right? I I think that it's uh there's a lot of opportunity there, and with that comes oftentimes a lot of change, and uh hope hoping that change can be smoothened out would be great.

SPEAKER_02 13:45

Yeah, and I I think uh as one of the programs that still has um a good amount of stuff sitting in a in a secret undisclosed data center in uh in northern Virginia. Um I I really look forward to um using a lot of these uh new you know AI technologies to kind of help accelerate um us moving that gear out of that uh building as much as I love that building. Uh but uh uh really um uh doing what we can to accelerate us getting out. Um and then uh one of the other things that I'm super excited about is the uh pilots that we have with the commercial solutions opening pilots where uh we could actually really, really make a huge impact on how uh goods are are imported and uh and and cleared um in into the US. So um that that's kind of where I'd like to go in 2026.

Closing Thanks And Next Steps

SPEAKER_01 14:34

Okay, so I I think I have to wrap it up, otherwise somebody's gonna turn off my mic. But thank you, Dad J, XD Bobo, and XT Tom Mills. Um I work with you pretty much every day. And you know, CBP is such a dynamic, um, critical mission that we're always either sprinting or running. We're never walking. Um, there's no time to pause. So it was good to pause today and hear your thoughts, but unfortunately, we're short on time. But I really thank you uh for being here and um um allowing me to moderate this discussion, and thank you to our audience uh for listening in and happy holidays.