B (3:11)

So we take any incoming data and convert it to actionable data. It's the only real time pixel by pixel computer vision technology in the world. And what you're seeing here on the left is the incoming video from any type or quality of sensor. And then what we do to it on the right, it does take 3 milliseconds, but that's less than a blink of an eye. And essentially what we're doing is we're taking that incoming video, converting it to 33 million pixels, and then we optimize each individual pixel. The two red, one green, one blue, the concentric pixels all around that, and then just do that 33 million times every 3 milliseconds. But because it's pixel by pixel, you can take it every single snowflake, every single raindrop, every single piece of sand for particulate matter or whatever it is, or in this maritime solution, sun glare in the morning, rain or fog in the afternoon and night, half the time. And the end value of that is, look, if you're a safety and security professional and you can see this on the right and all the analytics that go with it, real time without being in the loop, that makes a difference, to help save lives and make the world a better and safer place, which is what our mission is. And it just makes a difference, especially with the kinds of kit that David provides with the ruggedized laptops at the edge and the extreme edge, which is where so many of the deployments are going.

C (4:46)

So, Brent, like, I'm curious as to how you got engaged with Dell. I know that, you know, in talking to you, I met you in Barcelona at Smart Cities, and I got so excited about your technology. And to learn that the thing that excited me the most was that, you know, if you look at the stack of RTX cards, right, like because of thermal management that we have to take care of in the rugged product, we can't have big GPUs, right? So we've got the smaller 8 to 500 card. And I was really excited to learn that your technology is, is thin enough or it can use a small enough piece of that, of that RTX video card GPU that, that we could take advantage of it. Like you're saying on the edge or the extreme edge with our rugged products versus, you know, a standard laptop or a, a workstation with a larger video card that could, you know, more easily be damaged out in the field. So, but with that said, I know that it takes a lot to be, you know, to be partnered up with Dell. And one of the things that you told me that really impressed me was, you know, you went through these massive processes and you became one of Dell's validated designs, which is no small step. So how did, how did you come into and get with Dell and how, what was the process like for you to get to the point of being a, you know, a Dell validated design for AI?

B (6:15)

It's a hard question to go back that far because it took a long time to get it done. It really took three and a half years to go through that process. Just think about it from Dell's perspective. If they're going to put their name on it and they're going to validate to the customers and the hundred billion dollars worth of customers that we have at Dell, that this works and it's the best technology in the world, boy, they're going to put you through the ringers, right? They're going to test and validate and test every single other potential alternative solution out there. Right. But it's still new for the company because when, how long has our AI factory been out, Logan?

A (6:58)

What?

B (6:58)

Two years? Maybe?

A (6:59)

Probably two years. Yeah, probably two years.

B (7:02)

And we are now recognizing, look, our first hundred billion came from our infrastructure, but the next hundred billion is going to come from leveraging all that infrastructure and customer relationships and providing new solutions to those customers. And we don't have within Dell all those solutions. So we have to pick a few select ISVs or independent software vendor partners that can provide those new solutions and can actually work and scale up with the Dell infrastructure. So that's what Dell has to go through. So, so the company spends billions of dollars behind the scenes to bring these new solutions that customers can trust. And from my perspective, I just said, look, who's the best in the world from an OEM standpoint? That was kind of question one. Question two was who is going to be the leaders in AI in the future? So that was the second question. And third was where do we have some personal relationships? And we did have some at the, at the top levels with Dell. And so that just led us to say, look, let's form this strategic partnership with Dell. And we got lucky enough, I think, on both sides, both Dell and Prohawk, to be able to Ultimately execute that. And fortunately, concurrently, we did the same thing with Nvidia at the same time going through their labs. And the end result is we ended up being, I think Nvidia's got like 20,000 vendors, but they have about less than 50, 15 global preferred partners. So we're one of those. So we are exclusive in that respect also. So that is what led to it. Because, you know, as a military guy, I'm, you know, not the smartest, but I, I know how to execute it and I know how to do the basics. And that was smartest thing to do. It would be, you know, partner with Dell and Nvidia and then. And the partnership's just been fantastic.

A (9:03)

So, you know, one question I have is a little, maybe a little more technical, right? Like, you know, in the video that we showed, you know, being able to light up computer vision models, I mean, I think there's huge, you know, I mean there's huge implications that we'll talk about, use cases and all that. One thing I was interested at when I was kind of reading a little bit more about your technology. You described this kind of AI data, Flywheel. Right. It's like a self learning system that improves the accuracy of the downstream AI models. Right. So I'm, I'm pretty familiar with AI. You don't have to disclose anything proprietary or anything like that. Say you're using YOLO, you know, V8. Right. Which I don't know if you are, you aren't. But when you say you're kind of constantly improving the models, is that from a kind of a reinforcement learning? Is that an ongoing fine tuning? Can you walk a little bit more how the Flywheel works and how it kind of makes your product better over time?

B (9:51)

Yeah. Can I answer a different question, Logan? The question I want to answer is, look. Yeah, And I'm gonna, I'm gonna answer the question with a question back to you. What is the Achilles heel for every single AI application?

A (10:07)

Well, I'm gonna say the human at the end of it, but that's probably not the answer you're looking for.

B (10:11)

Well, it's not the answer I wanted, but obviously we haven't rehearsed this, but David, what would you say? What's the Achilles heel that for every one of these AI applications or softwares or agentic to work or work better? I think it's the data, don't you?

A (10:26)

Yes, I would say it's the, it's the data, but I'd also say it's the use case. It's being Used for. Right. Which ties back to the data ultimately. Right. Like for example, in your use case, let's say, hey, we're trying to see someone coming over a wall for sure, like that type of data, important, etc. But like another use case might not be someone like nefariously trying to break in, it's someone not wearing their hard hat. Very, very different data set. Right, But I agree, I mean, but it's, it's also kind of the human at the end who has to make interpretations and a lot of times they're incorrectly interpreted, which I think your platform does kind of help because it makes it easier to understand what they're actually seeing. But anyways.

B (11:04)

Exactly. Right. And, and if you can give a confidence level of what they're seeing, such that that human in the loop, ultimately, like you said, Logan, and you said David, is, Has the confidence to make that decision, that's critical. But for me, it's all predicated on the quality of the incoming data. And if you remember going back to the Dell AI factory framework, that incoming piece on the far left side that we don't talk about that much is that quality of the data and the incoming data. And this now to your question is all pro hog, does it cleans up that incoming data from any incoming video source to make the AI flywheel and to make all AI applications work and work better. So here's a simple example. If you see an incoming video and let's say it's degraded and it's snowing, and let's say you have a soldier in a Northern European conflict zone with a Dell ruggedized tablet, and there's battlefield smoke and it's pitch black at night and it's snowing, and if you can then restore that video real time so he can see his objective interest, that's step one. But step two, what the software knows how to do is cycle back because it's operating on that Nvidia chip and refreshing every 3 milliseconds and optimizing 33 million pixels every 3 milliseconds so it can go through and identify the object first and then classify the object next and then give a confidence level next, all within less than of a blink of an eye. And so this each time is improving the output for the user to make a real time decision and to be able to decide at that critical point in time, in some cases, when lives are on the line.

C (13:07)

Yeah, that's a great answer. So Logan's more the AI expert, I'm more the use case guy. Right. Like I'm really intrigued by all of the different ways that this software can be utilized, whether it's a fixed, fixed camera asset, you know, in extremely low light, or I'm assuming also that we can do or you can use it with drone technology, whether it's a tethered drone or a mobile drone. And how so the live feed, I mean, talk to me about like how that live feed is working because obviously you need bandwidth for video, right? How are you taking advantage of limited bandwidths, if you will?

B (13:56)

Well, ultimately this is where the Dell 5G capability might come in, right? Which is becoming, if not already best in the world. But we take any of that incoming video feed and Even with Dell 5G on that video transfer, it's going to be degraded in some way, right? We want to be at the extreme edge because compression is our enemy, right? So we want to get that raw data, that raw pixels, and we fit right after, as soon as it comes into a device, let's say a ruggedized tablet or laptop, it gets muxed in that device and we fit right in there, right. As a plugin before inferencing. And like you said, David, this is only using about 500 to 700 CUDA cores of the Nvidia chip. So if it's not a light touch on the Nvidia chip and if it's not a low power requirement, this only requires 5 watts of power. It's just the technology, it's just a waste of time. It's not a real end use case, right? Because there's so many applications where we need low power and a light touch in terms of using, let's say CUDA cores versus Tensor cores, which so many other applications use, right? So we're so fortunate that the software is that efficient and optimized to be able to run and to be able to execute and make a differ. You know, when I talk to customers, they want two things. Give me a two page solution brief against my specific use case, whether it's energy like you talked about, tethered drones for transportation, or autonomous driving or safety security or federal or whatever it is. Give me a two page solution brief against that and give me an ROI calculator and give me a video evidence of how you've already solved the problem. That's all they want. They already know about our infrastructure, they already know about all our capabilities, our storage, our everything. But they want to know what can we do with it. So that's really, I think, the magic of this combination of a software that actually works, actually Brings a new use case to customers that is installable in about three minutes. That is deployable edge server, client or cloud. Right. So agnostic with whatever the customer wants and is $9.99 a video stream. Right. So the ultimate strategy here is command by putting this out and getting this out on as big of a installed base globally against this $300 billion addressable market to establish that barrier to entry for whatever might come in the future.

C (16:47)

That's an incredible price point to. For entry.

B (16:50)

Yeah, it's kind of like the go daddy of the industry now right. When we used to charge $2,000 for it and we just said no, let's just go, let's just go dominate. Right. So there's got to be a better word than that, but that's really what

C (17:05)

we want to do.

A (17:06)

10 bucks a month per video stream or 10 bucks forever.

B (17:09)

$10 a month for video stream. Right. And then, and guess what? Our resellers get, get service revenue on top of that. And you know, our, our, our associates within Dell they get their fair share of margin. Our associates within the, the resellers and TD Sinix, they get their fair fair share of the margin. And the, the remaining tiny shekels are left for us, which is fine. But you know, what's our mission? Our mission is to help make the world a safer and better place and as a result help save lives.

A (17:40)

Yeah, I mean it makes sense. So let me, let me kind of ask you a quick question here. So I mean, so 10 bucks a month, whatever, 9.99 for the stream isn't

B (17:49)

that surprising though Logan?

A (17:50)

I mean it is surprising. I mean. Well, let me say this. For something that a is as good as you provide the relationships you have, the fact you've scaled into Dell and Nvidia very great price, there's OEM solution or there's you know like freeware solutions out there. I can, I can crawl around GitHub not the same as yours, but, but like so, but for that. Yeah, I mean considering some of the stuff that I talk about. Yeah, I mean it's, it's pretty inexpensive. I'm not gonna say cheap because cheap devalues it. But here's kind of the question is that it's more of a, I guess a tech functional question here is that completely understand running on you know like a Dell Pro rugged. I think it's the 14 David that has the RTX 500 in it. But I get that you know, if you're doing that but most of the time like let's say public safety. Right. Like some of the videos, you're not going to just put a rugged, you know, out on the roof and just let it sit there. So tell me a little bit more about how it would work in more of a bigger deployment. Do you know what I mean? Where it's like, hey, we're deploying this on a building that we want to protect. Right. Outside of the rugged use case. I mean, how many streams can run concurrently? All of these type of things for like a bigger use case.

B (19:04)

Yeah. So, and it all depends on the hardware and the servers and everything that we're using. So the, the ultimate calculus is based on numbers of video streams number one and number two, the numbers of pixels that are coming in because we have to process those pixels. Right. So it's a little bit more requirement for a 480p camera versus a 1080p versus a 2K. So the calculus is different for every use case. But we have that kind of T shirt of you need this kind of memory, you need this kind of power, you need this kind of Nvidia chip based on those numbers of video streams. And I'll just give you a use case. There is a country next to Venezuela that has just discovered a large amount of oil. And the quote we provided for them was, look, you need these numbers of devices at the edge to protect your border. You need these numbers of Dell servers in your tactical operations centers and you need these numbers of workstations and other kit in these locations. And so we t shirt size for each one of the different applications and linked Prohawk with each one of those things and the right Nvidia chip. And like David and I know, you know Logan from some of your blogs in the past, not all Nvidia chips are created equal. And you need the right Nvidia chip for the right use case and the right problems. So you, you deliver against that expectation. Right. So that is a perfect example where we did this. I don't know what the right English word, but the French word would be melange of software and hardware suite to meet a complicated use case to protect the the borders of this particular country that has just discovered all this oil.

A (21:04)

It really makes sense. I mean, so it's really, you know, camera screen. Okay, so that makes total sense. So it's not just a ruggedized solution. It can scale all the way up to a server, to a Delpro Max Del Pro precision workstation.

B (21:16)

Any amount of video streams for. Just think, you know, you're protecting a school or you're Protecting a large building and you got 15 screens. That user is going to sleep if he's trying to look at 15 screens. Wouldn't it be better to use AI and AI analytics to illuminate whenever, with, with a hundred percent confidence, whenever there is an anomaly, because then it gets his attention, right? That's essentially simply in that kind of use case, what Prohawk would do.

C (21:49)

You know, you talk about protecting lives and, and obviously, you know, in that oil use case, it's protecting property. You're, you're protecting the border, you're protecting the property. What are some of the biggest, and I know, you know, obviously certain things you can't, you can't discuss, but what are some of the biggest, you know, like in your mind, in your heart, like the biggest saves that your technology has ever made, whether it's been saving the lives of multiple people on a battlefield or, or controlling, you know, helping to control a wildfire that was about to consume, you know, your camera's ability to see through smoke and, you know, things like that. What are some of the things that come to mind that are some of your biggest wins, I guess you could say, you know, with your technology.

B (22:36)

I don't know if they're biggest wins, but they're the ones that impact me the most. David. So the FBI came to us and said, hey, we have this cold case in Indiana that's 12 years old where a student went missing, and here's all of the apartment and ring camera footage we have from 12 years ago. And we were able to identify, and essentially pitch black at night, a vehicle that was a Dodge like Ram truck, its unique camper shell that it had on the back, a person crouching behind that vehicle and then that person walking with the person of interest and identify the person of interest, the victim in this case, and the perpetrator from essentially the most degraded video that was 12 years old to help the FBI solve a problem, right? So those are the kinds of things when we can make that difference for police departments when they haven't been able to do it. And there's just a plethora of those kinds of things that really impact me. Or from a, you know, I would just say one other thing. From a, from a, from a safety standpoint, think autonomous driving, right? Autonomous driving algorithms are fantastic until there's night or snow or inclement weather. But essentially with Prohawk, because it's the only pixel by pixel technology, take out all that snowflake in that field of view, take out all that rain or fog in that field of view. And simply what the technology does, it allows us to see 20 times farther and 80 times faster. And if you can see objects of interest 20 times farther or 80 times faster, to then put it into the algorithms to vector your car from autonomous driving around that obstacle. That will help save lives. So we're really motivated by those things. And I like what we do and we try to take that principle in every single use case.

C (24:53)

Just a follow up to that, because that's one of the things that when we met in Barcelona, I said, so how are you restoring the video? And when you talk about, especially on the law enforcement side, chain of evidence, you know, things like that, I said restore and you said, no, no, no, we don't restore the video.

B (25:12)

You said enhance. And I said we restore. We don't. Enhance. Yeah, enhance.

C (25:17)

That's like enhance. I said enhance. Yeah, that's right. So, so, yeah, so, so just kind of, you know, hit on that. It's not an enhance because going 12 years back, you're not enhancing any video, you're restoring it.

B (25:31)

The Daubert standard, which is what you're talking about from a forensics and law enforcement standard, means you cannot alter evidence. And if you're enhancing or using gen AI or guessing in any way that's considered altering evidence, you can't guess at all. You can't put something that wasn't there. And so what we do is we just restore the actual pixels that just aren't seen. And that's what we need as close to the raw data as possible so we can restore and find that data using physics of what is actually there. And so we don't use any inter, we don't use any gen AI or guessing because that technically qualifies as altering evidence. It's the same on the federal side. You can't use gen AI for the current US rules of engagement. You can only use access actuals. And because we're only using actuals, you know, people ask me, well, is there any kind of hallucinations or false positives? Because I'm only using actuals and restoring actuals. There is no risk of that hallucination or, or end result of a false positive because I'm only restoring the facts versus guessing of what might be there. So it's, we're, we're fortunate in that respect that we're, we've met that Daubert standard and, and we, we ultimately have to address that question in courts of time. We like doing that. But I mean, there's so many use cases like, you know, we were even on the coast of Oak Island a couple weeks ago on two or three different episodes because they want to find the artifacts from whatever it is, you know, thousands of years ago or whatever.

A (27:19)

And of course, yeah, they're not finding anything. They're not finding anything.

B (27:22)

So here's the truth, Logan. They find a lot of stuff. But then once season 13, 14, 15, 16 and 17, so we found a lot of stuff, whether it's gold here or stuff underwater, because, you know, underwater is just low light and turbulence for us. Right. And particulate matter. So we just get rid of all that so we can find all sorts of stuff. Whether the ultimate audience sees those things or not is a different story. So they can have season 14, 15 and 16. But we like doing fun stuff, fun things like that to find treasure or even using seismographs to find gold, because the refraction and reflection of the S waves from the ground will show different hardness of metals. And so we're trying to test those things out too. So we try to do some fun things in big use cases also. But our real passion is save lives by making the world a better and safer place.

A (28:20)

Place. Before we start winding down, Dave, any final thoughts or questions you have before we wrap it up with Brent?

C (28:28)

No, I'm just, I'm so excited, you know, like in meeting Brent, I mean, we hit it off immediately. We're both, you know, former military guys and. And his technology just really floored me. The fact that you can use it on Adele Pro Rugged really excites me and I'm really glad. Logan, thank you very much for the opportunity to have this discussion with Brent and I, because I think that his technology is really worth bringing out to the public. I think a lot more people need to know about it. So, Brent, thank you for being here. I really appreciate it. And yeah, thanks for answering all the questions. Appreciate it.

A (29:06)

It's great. So, Brent, what I like to do with the guest to wrap it up is pretend someone just came into the episode. You have to give a 30, a 30 second to one minute elevator pitch on Pro Hawk AI and what they should take away. You're on the clock. Go.

B (29:22)

This is a conduit for all of the AI stuff in the future that we don't even know about. It's going to make all of those AI applications work and work better. But today in computer vision is where we can all make money, make a difference, Drive Roi. So it is a way to get started in AI and a way to start to learn in AI with virtually a guaranteed ROI for $9 or less than $10,000 with a Dell GB10 or close to that, those same kinds of prices with a Dell ruggedized tablet. All of those things means our customers and the end customers can get started with AI and be confident that they're going to be able to drive that return on investment. And importantly, that investment is going to be supporting all of the other things that they want to do and do in the future with AI.

A (30:34)

So last question for you is if someone wants to learn more about Prohawk AI, obviously go to Brohawk AI.com where can they find you on social media? Where can they go to learn more?

B (30:45)

Yeah, so I'm on LinkedIn. I try not to do any other posts on any other things but so I'm definitely on LinkedIn and they can reach me through the company's website where they can also email me at Brent willisohawk AI and we're here to drive some Dell hardware, drive some Nvidia chip sales. We know our job and that is our job. Just doing it in a different way by providing new solutions. And it is new solutions picket across industries that we can make a difference for.

A (31:22)

I love that. So we'll link to your LinkedIn and then the Prohawk AI website down in the show notes below. So Brent, really appreciate you taking the time. Great episode as always. Time flies when you're having fun. So with that, check out ProHawk AI, check out the Dell AI factor and special shout out to my boy David Ford and the Delpro Rugged that doesn't get enough time in the sun and is now doing big things with AI. So with that signing off, reshaping workflows with Delpro Precision and Nvidia RTX Pros GPUs and we'll see. See you on the next one.

B (32:18)

This podcast was produced in partnership with Amaze Media Labs.