← Liftoff with Keith: Conversations with Super Founders and Growth Experts sharing useful insights
Liftoff with Keith: Conversations with Super Founders and Growth Experts sharing useful insights
Trustworthy AI: Why Explainability Will Define the Next Decade of Enterprise Technology
00:36:55
Loading summary
Transcript75 lines
- [00:00]
A
Foreign.
- [00:06]
B
Liftoff with Keith is the one and Only Compass Strategic Advisors.com an experienced partner to help you navigate everything from cap tables to stock option and compensation plans and all types of backroom and marketing services. There is no better friend to the startup CEO than Compass. Check them out@compass strategic advisors.com. well, this is a real treat. I get to welcome Pat Condo, the founder and CEO of Seeker Technologies to the liftoff. Pat, welcome.
- [00:37]
A
Thank you. Great to be here.
- [00:39]
B
Yeah, we just established your, you're centered in the heart of the data center world right there in Virginia, right?
- [00:48]
A
Yeah, it's been amazing over the last five or six years. I think there's been, you know, more capacity built here than any part of the world in a concentrated fashion. And it's created, you know, an enormous economy here of not just engineers, but remarkably, construction, electricians, plumbing, shipping, you know, everything. And just to see the scale at which these facilities are being built and know what kind of capabilities that they're harnessing is, is amazing.
- [01:24]
B
It's, it's, it's exciting to hear because we used to only think about Virginia D.C. area to service the, the government. Right. And the defense tech, military tech, which is by the way still in vogue and maybe now so as much as ever. But to have those other surrounding industries, that's really exciting for that community. That, that, that area.
- [01:44]
A
Yeah. And as we were saying, Keith, at the, at the onset, the government has, you know, over a trillion dollars is about to be spent or is being spent in the Department of War only. And lots of very leading edge innovation is taking place here and defense technology has become kind of a term now. And almost every Silicon Valley company has now established more than a major outpost here. In fact, they're, you know, actually recruiting and building and you know, right in the area that we are in, in western Virginia, I mean we've got every one of the, you know, the Magnificent Seven and then we've got hundreds of other startup companies that are looking at everything from, you know, from space to, to kind of undersea capability. It's just enormous.
- [02:36]
B
Yeah. Well, before we dive into Seeker, which I'm, I can't wait to get into, I do a little bit with founder philosophy and kind of your, your origin story. And you have an interesting background, Pat. You founded six companies before you started Seek, or at least that was what I counted. What kind of patterns or insights from that journey made you convinced it's time to go for number lucky number seven?
- [03:01]
A
Well, I think that my philosophy has always been that curiosity. I'm A person that continues to learn. I don't think I, I know everything and so what I want to do is I'm curious about everything. I, growing up as a kid I, I read all sorts of science fiction and things of that nature. So it's always been something in me is to kind of learn more than, you know, what you see today in the news. And I think from my background in, you know, starting out really building navigation systems for space and missile platforms and working on the space shuttle, I've always been drawn to sort of what's the next thing and what's possible. And so I think that's been an overriding principle that I've had for the last 40 years.
- [03:52]
B
Well, to fill in those blanks, I mean I remember Dec, we used to call it Decor, semi Northrop, where some of your backgrounds. But now you've become this sort of serial founder. What kind of cross domain pattern matching, you know, Lane, do you follow to help create that consistent competitive advantage?
- [04:14]
A
Well, I, I learned the hard way. I was always a behind the scenes person. So working at Northrop, working at Harris, I was always in the back room doing programming or doing, you know, business analysis or engineering, all the things that you never had to be the frontman. And then in the early 90s, Herbert Allen from Allen Company, famous investment banking firm in New York, approached me and said, gee, we'd like you to come run this company. And I said, oh, interesting. And they said it's a public company. And I said, I don't know anything about running anything public. And they said, don't worry, we'll teach you. And the first thing I realized is that you need to be a salesman. And I, that was the last thing I ever wanted to be. But I realized is the definition of salesman is, is quite broad. It's not only selling to investors, but it's selling to business partners, it's selling to customers, it's selling to your own employees. And I realized that if you could combine that kind of capability with your intrinsic, say, desire and curiosity and knowledge to kind of move forward, that could become a very powerful way to build and lead a company forward. Because I always believe in the principle, you need to lead by example. You need to, you can't be sitting in the back room criticizing people. If something's difficult, you need to be upfront solving it and let them come behind it. So that's been increasingly how I've operated as a, you know, founder, CEO.
- [05:47]
B
Yeah, and you're, I mean your website, the homepage says it all. AI you can trust in mission critical environments, right? I mean that's where we are today. That's fundamental.
- [06:00]
A
It is a very fundamental, simple thing. But to deploy it, it's complex. Think about when banking systems. There used to be a hundred different banking systems but nobody could trust them. So then over time they became one banking system. There used to be a credit, credit reporting systems, there used to be hundreds of them. But again, bias, misinformation, different ways to process credit. The fair credit came, act came. Now there's only three ways to process credit and it's the Internet came all kinds of social media, news, you know, corporate websites. People could say and do whatever they wanted after a while. Nobody trusts the news, nobody trusts politicians and nobody trusts CEOs. Why is that? Because they can't, they can't trust what they read, they can't trust what they're seeing. So the, so what has to happen is there has to be a level of trust built into any system for it to work effectively. And that's no different for AI.
- [07:05]
B
That's a, that's a tall order.
- [07:08]
A
Well, it's, it's, it's interesting because you know, throughout time they've always had this theory in computing. Garbage in, garbage out. And so that is the way it works with AI. If you train your models with poor quality data, you end up building a poor quality reasoning engine. If you train your day, if you train an engine with high quality data for that domain or that series of domains, you're going to have a high quality reasoning engine because nobody puts high quality data on the Internet and allows you to train for free on it. All of it's just marketing material, corporate websites or you know, news and social media. So as you.
- [07:52]
B
No, no, no, you're hitting right where I live too with on the media side, the journalism side. That's why I say it's a tall order. How much are people recognizing the need on the customer side? Certainly I understand a lot of us are seeing it. How much of us are really trying to move that into commercial application?
- [08:10]
A
Well, I think that you know, in the beginning when the large consumer engines came out, the only person affected by poor information or hallucinations was the individual. And that individual has a tolerance of risk and that a corporation or a government in certain ways does not. So it as the first wave hit and everybody saw these great chat based AI programs and it was the first time that the average person could really communicate with a computer and get it to do something without being a programmer. And that was a huge Breakthrough. But as time went on and enterprises tried it and governments tried it, they found that they couldn't tolerate hallucinations. And so that became kind of the next phase. And so now as you move forward into that domain, companies that have built, you know, on the consumer side are really trying to build models that, where they can train it with higher quality data and they can provide them, they can provide a pathway where they could say, hey, look, we've reduced hallucinations or the level of errors has been reduced. We took a different point of view and we said from the very beginning that we're only going to be a domain expert and we're going to build something that is what we call explainability. And that means that you should be able to understand how you got the result in AI. Where today it's a bit of a mystery, but what corporations want or national security want is that they want to be able to say, here's the end result. Think of it like a corporate auditor. I have a P and L, but if I going to test the quality of that P and L, I'm going to go all the way down and test 10 invoices and see how they travel through the financial system. And if I can't see how they, at the end of the day they end up in that P and L, then there's a problem. It's either an internal control problem or it's fraud, but there's something wrong. And so when you look at an AI model, if you can't get to how they trained it, if you can't then contest the source in there, and if you can't observe how it operates in different environments, then you're going to have something that in certain environments, based on certain events, it may return poor answers. And that's okay if it's your resume or that's okay if it maybe is, you know, something where it affects the effect is not catastrophic. But if it's managing nuclear plants, if it's a satellite operation, if it's, if it's running, you know, undersea drone capability, if you're flying autonomous, you know, jets using AI, well that's a catastrophe.
- [11:09]
B
Yeah, there are lots of catastrophes in that scenario. So in the idea that we see what's going on right now in the industry with large hyperscaler model labs, enterprise software, all under attack and it's going getting crazier and they're all claiming to own the stack, or they're all claiming to be sort of the, the lead dog in the race where's a specialized platform like Seeker fit in. How do you avoid being part of the expansion and not the consolidation?
- [11:43]
A
Well, I think that, you know, it's a, it's a game of specialization over time. And like the Internet developed into where there was some very large players at the beginning. And now 20 years later, there are a thousand vertical markets that are set up all across the Internet. And it's the same with AI. It's starting with three, four, five, six big players. But then there are hundreds of industries and within each of those industries there are, you know, thousands of applications that can get created. No one company can be an expert in everything. So we're picking those spots that have a high total available market that we can penetrate with what our kind of capabilities are. And then we build domain expertise into our technology so that we are the very best in these certain markets and they're big enough to bring the company into, we think, multi, you know, double digit billions of dollars in value.
- [12:55]
B
Yeah, Pat, is that what you guys talk about when you write content evaluation by design? Is that those lines?
- [13:04]
A
Well, so there's really a few things here. So if you, if you want to build a very successful, let's say model to analyze satellite data, you, you have a platform that allows you to ingest data, build models, you know, manage the outcome, you know, observe it and you know, contest sources, all that stuff. But on top of it, you want to build agents. And those agents are really only as good as the underlying foundation. And then those agents can then be very domain specific and can handle very specific tasks. And so that's so and so when we say content evaluation, it's like what the content could be. Real time imaging coming from a drone that could, and then we evaluate what that is, how to analyze it in the model, and then where do we deliver it? And we could deliver it wherever the customer wants it if he wants it in a cloud or if he wants it, you know, on a ship, or if he wants it in a weapon system or if he wants it, you know, in, in a laboratory. It doesn't matter. Interesting.
- [14:20]
B
You, you describe explainability as a baseline requirement. Yeah, it sounds like one of those nice to haves, but you know, it's.
- [14:29]
A
No, it's not. Because just think about in the very extreme case, let's say you're used inside of a, of a nuclear plant and that plant, something happens to that plant, the first thing they're going to say is how was the plant operated? And you're going to say, oh my AI product was running it and they're going to say, well, to me, how, how Explain to me how it got the result. It did that turned this valve in that valve and caused a problem. And explainability, does that, does that cross
- [15:00]
B
over to other industries as well that aren't as mission critical kind of industry?
- [15:05]
A
It is a sliding scale, in my opinion. So it slides towards some very basic things like I'm inserting documents and I'm looking for financial information. And that might not cause death and harm, but it could cause a lawsuit if the model's wrong. All the way to the kind of critical information infrastructures where if somebody's harmed, somebody dies, then that can cause a catastrophe of pretty substantial proportions. That's where explainability becomes an absolute must. And then think about the most explainable place that you're going to have to be is sitting inside the $180 billion golden dome that's being built, the missile defense system. How can you have an error in allowing something to occur within that system? That's the catastrophes that you want to avoid. But that's where all the investments are going in defense tech is to have absolute explainability at every level to get autonomy.
- [16:11]
B
Pat, let me just hit a quick sidebar question then for you. How big is Seeker now? And are your customers mostly on the government defense side or on the commercial side, or how do you view that?
- [16:22]
A
Yep. So we are a mix right now. So Today we're about 60% commercial, 40% government, and next year I think it's going to probably flip. And that's really based on the unbelievable acceleration in the requirement of Department of. Not just the Department of War, by the way. It's almost every department within the government where they want to either drive efficiencies or they want to implement AI for revenue leakage types of models. And we've seen dozens of that where simple things like over the last 50 years, here's the way we aggregated data and we put them in these 10 accounts. But there's not 10 accounts, there's really 1,000 accounts. But nobody ever monitored them, nobody ever looked at them. And when they consolidated all the accounts and actually looked at things, they found a lot of savings.
- [17:22]
B
Wow. Well, government business is great when you got it rolling, that's for sure. It's a long sales cycle, can be very challenging. Right. You probably know that as well as anybody. What else did I want to ask you? Inference optimization is quite a critical battleground too. What you know, everyone's thinking about inference costs. Right. And inference first infrastructure is a Word I think I saw in your, in some of your language, what's it actually look like in practice?
- [17:54]
A
Well, I think what, what, what people are really talking about is there's two sides to AI. First, there's training and building the model, very expensive, all the jeep. You know, Nvidia has made a huge business out of being the best at that sort of thing. AMD is right up there too. And then, and that costs a lot of money. Then you move to what's called inference where they just execute the end result of that. And, and that can be, that's going to be like, you know, a cost game where whoever can deliver the lowest cost to inference, you know, that could make a huge difference because you don't want to necessarily run all your inferencing on the same chip that you run all of your training. Now there are multiple ways, multiple companies that are entering the inference market that we see. So there are the big players, you know, the AMDs, Nvidia's, intel, people like that. But then as you go to inference, you might find the floating point companies, you might find ASIC companies. So you get a whole host of people, specialists that are saying, hey, we can drop the cost from X to dramatically down to Y. And that's really just about kind of the growth in the compute business where if everything is running and moving towards AI, you want to figure out how to drop the cost to execute it.
- [19:16]
B
Some interesting tokenization models come to mind too.
- [19:19]
A
Well, so I, I have a view on that. It might not be exactly shared by everybody, but I think that, you know,
- [19:26]
B
nothing is these days. Pat, it's okay.
- [19:29]
A
You know, the idea that tokens is the way to pay for things, you know, has been pioneered in from the very beginning on kind of on the consumer side, but on the, but in the enterprise side and in the government side. I mean it's question mark because the use cases in the way it's used is very different. I mean you could have a power user that might use thousand GPUs to solve a problem, but are you going to charge them on tokens or you're going to charge them by user? Or you could have, you know, a consumption based model where there might be, you know, thousands of users only using one or two GPUs. So are they going to pay for all that data? So I think over time you're going to go away from some of those kind of models and more again towards is it a user or is it a volume of data? There's going to be Some combination there. But right now, you know, the way that it's operating, you know, is a, is a great amount of money to some people that sell based upon the token model. But I, I see a definite shift away from that over time. It's just too expensive.
- [20:33]
B
Well, and you're probably picking up between your enterprise sales cycles and your government sales cycles. It's the path of least resistance. Where is there a natural progression versus you want to fight them on a new way of compensation or payment or pricing. So what's the best from a go to market standpoint? And you're talking about trust and compliance. How do you make this conversation go as fluidly as possible?
- [20:59]
A
Right, that's it. I mean, the end of the day, the value is to get the customer. And so how do you get the customer? Well, they're getting pretty smart now. And in the early days they had no idea like how, how much they were going to be spending to do AI compute or how much they were going to spend to run an app or how much they were going to spend. No one knew. Now they have a very good idea. And you know, I think it's, it's going to, you know, go back to the, to the way it's always been. There's some driver of the use of that and they're going to figure out how to, how to, how to, you know, put that back on track because they're not getting, if they can't get an roi, they can't see an roi, they won't do it.
- [21:42]
B
You're pitching an AI platform on trust and you know, they've been burned with other deals that have come before you. How do you get in there? How do you, I mean, obviously you're leveraging a relationship, I imagine, but in those cases where it's not a personal relationship, what's that first step to sort of rebuilding that conversation and gaining that credibility and that advantage to, to avoid the skepticism that's, that's, you know, pervasive.
- [22:11]
A
I think in the very beginning we would walk in the door and they would say, seeker who? Let's talk. We're going to talk to OpenAI, we're going to talk to Anthropic, we're going to talk to X, Y, Z. Because my boss told me that I need AI and I don't know who you are, but I know who they are. A year later we'd walk in and they'd say, we tried these guys, we don't know what happened. We had all these issues. It's cost Us a fortune. Can you solve this problem? So our best customer is someone who's already tried them and it hasn't worked.
- [22:47]
B
So does that move you into a trial pilot? How does, what's the.
- [22:51]
A
It's always that there's. There's always that it's. And they're getting shorter. I mean, in the beginning, they would. They could be, I don't know, plus three, four or five months and a lot of work on our part. To be honest, there was trial and error on both sides because you're finding out things about the customer, his data, whether he really knows what problem he wants to solve. Oftentimes they don't. Now it's very in. The consultants came in. So the, the second wave was the wave of consultants and they got in the middle. And, you know, there's a lot of obfuscation when they're in the middle because their work, they're trying to learn at the same time as help the customer. And that caused lots of, let's call it, anxiety amongst companies our size. Now it's a lot more straightforward. We walk in, we say what we've got, they tell us where they were, they tell us what didn't work, and we say, okay, we can solve this problem or we can't. In 30 days we're up and, you know, already produced a model, already can tell them whether they're going to get an ROI or not, can tell them how much it's going to cost and can tell them the next 10 apps that they should be building, and it goes much, much more smoothly. Customers also become smarter, upgraded their knowledge. Even the vernacular has become simpler. So in the beginning, you know, AI vernacular was like a whole dictionary of words no one heard of. Now it's boiled down to exactly what it always is.
- [24:26]
B
So let's, let's dovetail off that for a second. In this AI era, for somebody who's been around a little bit, how are you seeing the idea of running a company today versus some of your previous enterprises and larger companies, smaller companies, et cetera. But now it's a whole new world, too.
- [24:46]
A
Yeah. Well, I would say two things. One is it's, it's an order of magnitude faster pace. Number two, it's a lot more focused on really the right talent pool. You know, people increasingly who know these things and can walk in the door make a huge difference.
- [25:06]
B
Yeah.
- [25:07]
A
And number three, you know, for somebody like us and probably everybody, to some degree, it's a money raising and you have to simultaneously do all these three things while going out and getting customers and you know, convincing people that, that the AI boom is a boom, not a bubble. And you know, when the stock market goes down, you spend a week, you know, making people feel good. And when it goes up, it's like, you know, how come you're not bigger?
- [25:38]
B
Time to raise more money.
- [25:39]
A
Always. Yeah. Well, that's it.
- [25:41]
B
Did you just close a huge round too?
- [25:43]
A
Well, we did. We've raised almost 280 million. We've built an exceptional team of over 100 AI engineers and we're rolling out products across these different dimensions. But it is unfortunately the money that's being raised as you raise a lot of money, which in relative terms 280 million is an enormous amount of money. But next to someone who raised a trillion, it doesn't look like anything. I mean, those are the.
- [26:15]
B
We are, we are talking some insane numbers too. I know, it's, there's the haves and the have mores. Yeah. And a few have nots.
- [26:25]
A
I guess the, the ambition level is not just confined to the planet Earth. It's like I want to own all the space of the solar system. Right. I don't, I'm not just satisfied with Earth. So we've gone from where you used to worry about dominating, you know, a region or dominate, dominating a country. It's like not just the Earth is not big enough anymore. I've got to go to the moon and I've got to go to Mars and they're spending the money to do it.
- [26:55]
B
Yeah. I have one more formal question before I hit you with a few lightning round questions. But you know, you have a very sharp view and seekers got a unique space in the marketplace. You've accomplished a great deal so far. You know, what's, what's that one piece of advice that, that you constantly see in your business and in this industry that's a bit contrarian or underappreciated and you go, why don't these people, you know, understand or what's that piece that's missing? Did we touch on it already a little bit or no?
- [27:30]
A
I, I have my, I have my view on this and sometimes it's, it's this, that if you are a policymaker in Washington D.C. and you are trying to set policies for AI kind of moving forward and thinking about how to create a level playing field, how to accelerate the development and adoption, all the important things to make the nation bigger and better and greater in the economics here. The pet peeve that I have that everybody, that people don't see the hypocrisy I guess in is who do they go to for advice? The very firms that they are fighting as monopolies today. And it's like, why would you go to the monopolies and ask them how to create a fair and level playing field? Why would you go to the monopolies and ask them about what the great future is of these products? Why don't you expand your point of view and pick people who are in different kind of aspects of AI or in education or in just areas that are not concentrated on. They're the biggest AI companies because they are the big AI companies. They're also being. All of them have been, you know, accused of monopolistic, you know, practices. And yet the hypocrisy is they put them into positions where they can continue being a monopoly.
- [29:00]
B
So yeah, you're touching on PAC money and lobbying and some of that stuff too, which you've had to live a lot more than I have.
- [29:08]
A
It's all connected. Yeah, I mean, they pay more money to those kind of, you know, influence peddlers than any company can afford because there's a relationship.
- [29:22]
B
Yeah, they're just following the biopharma playbook. That was a decade ago, a generation before thing.
- [29:28]
A
To me, that's the most.
- [29:29]
B
Okay, well, I love that. I think that's fantastic. Let me jump into some again. I love this industry because we are at such a unique moment in time and I love the fact that you have such a broad perspective. The AI capability enterprises are most under investing in right now would be what
- [29:48]
A
areas up most underinvested. Yeah, well, I would say manufacturing. I would say that if, if we want to really up, you know, improve the level of, you know, let's call it skill set in America and be able to regain that manufacturing capability we had. The more energy you put into AI to build digital factories, to build these automated systems and it'll create so many more jobs, it'll bring so much, so many more benefits back into that environment. And I think we're missing that. We're all guilty of going for these things that are far into the future, let's say the hype cycle. But the reality is you want to revitalize. I don't know, pick your city that was destroyed during the 80s and 90s. If you want to revitalize it, go there and build a factory with robots and all kinds of digital kind of twin capability and you will get people around that and then an economy will grow.
- [30:59]
B
Yeah. Not to mention the city's governments can do a Lot with the AI, automation, the robotics, just their own infrastructures getting them more AI powered.
- [31:08]
A
Yeah, and you'll, you know, what's, what's always been proven is that if you can, you know, show people that they're going to, that you're, that there's an investment in this area, people will go to it and they will build their skill sets in real time, not just sitting in a classroom. And then that will spawn all kinds of economic boom. I believe I do too.
- [31:32]
B
Hey, Pat, what's the one metric you like to look at that your, your, we'll call it, your peer group might not be that focused on, but it's sort of your favorite metric that, that you look at after the obvious ones.
- [31:47]
A
Yeah, so I look at really, you know, kind of where I am right now is that I look at hiring. I don't have to hire the person who is the absolute best in the world at what they do. I look at someone who can work in a team and, and can effectively work with many other people. And I find I get more productivity out of that than having a whole bunch of superstars who can't work together, can't be in the clubhouse together, can't, you know, can't relate to each other. Yes, here and there you need the superstar. But if you don't have a functioning team that can operate together, at some point that breaks apart. And, and I think you see some of that happening where somebody is a superstar and they leave all the time. They go from one company to the other because they're lured by big pay packages. But what happens to the company underneath is their team is disbanded or the product doesn't get built, all these things sort of occur. So I look at who we can bring in that can really create that leadership. And the leadership has to go across multiple different functions and downward as well.
- [33:08]
B
That's, that's a refreshing perspective in this wake of, let's just say the Mag 25, where it's just a game of musical chairs, it seems, you know, so
- [33:19]
A
small company cannot, cannot sustain that kind of disaffectiveness where, where everyone's a superstar, a prima donna, and everyone can walk out the door at any time. And, and when they do, their whole entire team disbands because then they go raid them. I mean, it's, it's impossible to sort of operate in that environment. Even big guys at Silicon Valley have tried to fight that.
- [33:42]
B
Yeah, it seems like that's a lesson you might have, has evolved over your six companies and you go, okay, wait a sec. I figured out I'm not going to necessarily draft Jordan or trade for Jordan if I got, you know, of five other players that really know how to move the ball.
- [33:59]
A
Yeah. And, you know, it's the, it's the, it's the, it's the definition of winning. Right. Too. And, you know, we look at winning when everybody in the company is economically rewarded, you know, disproportionately to where they might be somewhere else at this point in time. And it, and it, and it, you know, and it can be a great motivator, but also the, you know, are they doing something they really love and making a difference? And that makes, and that also is very important. Increasingly, over the last 10 years, there are very, you know, intelligent people that are in this space that want to do some good at the same time or want to do something for their country or want to do something, you know, for a particular cause. And I think being very careful to weave all that together in, in the fabric of your company is important today.
- [34:55]
B
Yeah. So. Okay, Pat, final one, kind of the obvious one. What does winning look like from our last point, say, say by the end of the decade or five years from now? How do you, how do you evaluate that?
- [35:09]
A
Well, well, I think there's a couple of things. One is that, you know, we have an unbelievable set of investors who have believed in the company from day one when it was, you know, let's put it this way, it was a real speculation as to how you could take this idea of something called trustworthy AI and could it even be possible to make it a business? So I, I really, success is for them to be, to earn. You know, all of them have worked hard, put their money in. I want everybody to, to benefit. Number two, it's successes where my, all of the employees here do well. And then number three, I think the ongoing work that we're doing be part of whatever bigger company, you know, comes along or whether we're become a public company. But in either case, I'd like to see it endure because it doesn't change, you know, over 40 years. The key has always been what's the quality of data you're processing to get the end result. So there's no different than it was when I was building Gyrus, you know, inertial measurements systems for spacecraft. It's the same thing. You have the wrong data, you're going to go to the wrong planet.
- [36:21]
B
That's a good one. I like that, Pat. Well, listen, it sounds like you're on the right path. I don't think you're at anywhere different. I just appreciate you taking the time to answer these questions. And I know you're super busy with a company that's just on the back of a rocket ship, too.
- [36:38]
A
Well, now, thank you. Yeah, I really enjoyed it. And we do it again sometime.
- [36:43]
B
Oh, I love it. It'll be less than five years, though.
- [36:47]
A
I hope so.
- [36:48]
B
I will. Enjoy.
- [36:49]
A
All right, Keith. Thank you.
- [36:51]
B
Thank you.
- [36:52]
A
Bye. Bye.