How Blockchain is Revolutionizing Manufacturing at AFRL with Keith Scheffler

[00:00:00] Keith Scheffler: Blockchain's come a long way. It's, it started off as this cryptocurrency thing. That's how people really knew it, or that's the first real application. But if you think about it from like a 10 by 10 matrix, right? You got a hundred squares on there. Literally one of those squares is cryptocurrency. The other 99 are other applications that would boggle your mind.

[00:00:40] Ryan Connell: Hey, this is Ryan Connell with the chief digital and artificial intelligence office. Joined here today with Keith Scheffler. Keith, how are you doing? 

[00:00:47] Keith Scheffler: Good. How about yourself, Ryan? 

[00:00:48] Ryan Connell: Hey, can't complain. Hey, I'm excited to talk today. Um, before we dive in, you want to just give a quick intro to yourself? 

[00:00:54] Keith Scheffler: Yeah.

Keith Scheffler with AFRL, the, uh, RX, well, man, tech division. I've been at AFRL now for four years. And prior to this, uh, was in industry for 25 plus years and manufacturing. So manufacturing all around the world and various methodologies and different products and using different machinery and such. So got a pretty extensive background in manufacturing.

That's kind of how I landed within RX or Mantec. Awesome. 

[00:01:20] Ryan Connell: Very cool. So AFRL has a stake in manufacturing. Like what's that like? That's your entire team or division. 

[00:01:26] Keith Scheffler: Yeah. So let me, uh, so AFRL for those that don't know is a air force research lab. So we are, um, a roughly 7, 000 strong group of mainly engineers, but scientists and engineers that handle some of the harder capability technology gaps that are kind of thrown our way cross service.

We handle a lot of air force stuff, obviously, but we do handle a lot of stuff that's outside of service or that the technologies that we create are transitioned. We'd like them to be transitioned into sister services as well. So we create a lot of umbrella technologies to push across rather than a spot fix.

Like, you know, if you've got a wound on your hand, is it from a, you know, do you need a band aid or do you need to go to the hospital and have larger triage done? Right. So that's kind of what we do is try to address that larger umbrella so the money goes further. So yeah, AFRL. So. If you think about it from a holistic view are our platforms are weapons platforms are, you know, they're very organic, right?

So we can't just look at the platform once it's delivered to us. We've got to look at the feasibility of sustaining that platform, the manufacturability of that platform, what does that look like? You know, we've got some platforms that are out there now. Over 70 years old, you know, and we're still moving those along as far as how do we continue to use those?

Do we update the engines? Do we need to retrofit wings? Do we need to shore up landing gears, things like that. So those are some of the things that come to us to be able to identify and to be able to move forward with or to fix or to find some resolution towards the new capability that they're looking for.

So, from a manufacturing standpoint, that's what our XM does. So a division called Man Tech. So each of the services has a Man Tech division, and that's manufacturing technologies. I sit within the advanced manufacturing technologies team, if you will.

[00:03:11] Ryan Connell: Got it. Very cool. And so, you know, my, my only experience with AFRL is mostly focused on like digital capabilities.

So kind of when I met you the first time I was like, Oh, we have a whole manufacturing area here. This is interesting. But when I'm, when we talked, you showcase a lot of, I'll call them digital capabilities for the manufacturing mission. And one thing right off the bat was just like blockchain. And I thought, I was like, gosh, I thought blockchain was a cryptocurrency thing.

So can you kind of just give a quick overview of what blockchain is and how it's potentially helping you in your job. 

[00:03:40] Keith Scheffler: Yeah, for sure. So yeah, you know, in that context, from a manufacturing standpoint, if you look at it, even 40 years ago, 50 years ago, there was a lot of just. You know, ground and pound, if you will.

You know, we just kind of got out there with machinery and we had CAD to a certain degree, but it was very rudimentary. We couldn't, can't do the things that we can do today. And we would go manufacture things, right. Pen and paper and such. And so now everything's evolved to a digital twin or a digital thread.

Right. So if you start pulling on that digital thread, what are you going to get out of it, you're going to get drawings, you're going to get certifications, you're going to get requirements, you're going to get NBSC or models basis and systems engineering stuff, system out, you're going to get those types of things.

And as those become more and more prevalent, we've got to, we've got to have a tool chest to be able to use what's in our tool belt. So we transfer what's in the chest into our belt. And once those things are in our belt, then we can really use those. Very handy, right? So they're within arm's reach. We know we're comfortable with them.

It's not like some tool that you use once in a blue moon that you've got to figure out how to use it again. And now I gotta, you know, I gotta figure that one out. These are things that we use daily. So that's where we come in with different engineers. You know, we, we don't fixate, right? That's not a good word for us, but we do tend to specialize and in different capabilities for the longest time.

And I was a roboticist within industry. And then that continued once I got to AFRL and then some things were happening in and around where we're currently at today in our state with our defense industrial base, and since I've got such a. Extensive background and setting up in manufacturing facilities and global distribution in different modes modalities.

So that kind of like. Got thrown over the fence into my backyard. So that's kind of what I'm dealing with today. So with that, you know, back in 2013, Walmart was one of the first public iterations of blockchain, right? And up until that time, and that's kind of when, 2012 and really Bitcoin if you will, or cryptocurrencies started coming online and people were like, oh, you know, this is a thing.

And back when it was like, you know, $50, and today it's what? 60 some thousand dollars, you know, for one Bitcoin. So like that's virtual, like, okay. So from that standpoint, you know, blockchains come a long way. It's, it started off as this cryptocurrency thing. That's how people really knew it, or that's the first real application.

But if you think about it from like a 10 by 10 matrix, right. You got a hundred squares on there. Literally one of those squares is cryptocurrency. The other 99 are other applications that would boggle your mind. I mean, we're, we're using it for things that are just so, you know, it's, it's not the wild, wild west.

I wouldn't say that, but what you can think of blockchain as is a, it's a new way to think about a database. Right? Cause that's really what it, we're storing data on either on chain or off chain, but we're storing data in such a way that now we have a full auditable trail of where it came from, who it came from, what was the actual size of the file when we got it, what was in it.

So we tend to call that the providence, right? Of the document. How original is it? Where did it come from? All those things. So from blockchain, it's immutable, which means that it can never be deleted. And it's from a consensus or it's usually on a consensus type network Everybody has to agree to add something to the block.

So if you think of it like a locomotive, right, or a train, you've got the engine, but then if you need something else to be carried, that engine to carry something else, you add a train car, right? And once that train car is full, what do you got to do? You got to add another one, right? And so on and so forth until you got to tell you just got to add another engine, which means we might pull on what's called a side chain.

We might side chain blockchain and, you know, run it off and do another avenue, if you will. Each chain progressively, um, is built off the one previous. So the, the header and the next block, if you will, or the next block, um, contains the hash of the previous block. So that way that gives us that auditable trail.

And we know the providence of this new block because it stems from the old block, right? So when that new block is full, we spawn off a new block. So that's really, so within that, within this chain, within this blockchain, this idea of a, you know, a new type of database, you can literally store anything in there.

I mean, you know, you can store a digital artifact and you can store numbers. You could store a decentralized identifier, which I'm sure we'll talk about, but you know, it's a, it's like a digital key. So we store a lot of things on there and the, the security of the blockchain is really what's paramount or that what it's known for because it's unhackable at this point, right?

You can't get to it. And even the things that we're building now in and around blockchain, we're making sure that as quantum comes online, quantum shows a high probability of being to break current encryptions at some point in time. We're not there yet. We're not anywhere close to it, but at some point in time, theoretically, it could potentially break 256, right?

As 256. And if it does, then, you know, then everything in the past is in the wild now, right? So that's where everything that we're building and blockchain is great for that because it allows us to build quantum resistant types of things into the chain currently. So sorry, maybe I'm too far in the weeds, but 

[00:08:54] Ryan Connell: no, no, it's good.

And I'm just trying to understand everything you just said and as it corresponds to like what you're doing in manufacturing. So like, teach me like I'm 10 years old kind of thing. Right? Like, so take, I'm just going to make up an example. So we have like a aircraft engine for an example, you're talking about applying all of the data elements and fields that might be relevant drawings.

And then. Connecting that to the supply chain or like, how, what does that look like? 

[00:09:18] Keith Scheffler: Yeah, so those are really, they're related for sure. And this is where the larger part of my research is centered around is a lot of graph theory. So they're coincidental, right? I mean, they're out there, they're connected to one another, but how distantly connected or what's the closeness of that connection, right?

To use layman's terms. And we. We're able to find connections between things that we normally you or I would just look at and be like, wow, that's happenstance like now that's nowhere near happenstance. And here's the six different ways of, you know, that it's related, but, you know, from a manufacturing standpoint in using blockchain, we can use it in a lot of different ways.

Because if you think about manufacturing, you know, there's a few instances and overseas when. An adversarial nation was able to gain documents, you know, from some programs of record that they shouldn't have gotten right. And those documents kind of walked off. It's a very public thing that had happened and well reported and company got fined and all that mess.

So it's not saying anything that people don't already know, but like all of that's avoidable now because some of the methodologies that we use and we employ. So like. Because documents, we don't look at vellum anymore, or we don't look at a piece of paper. We don't machine off of a piece of paper anymore.

We feed digital instruments or digital documents into a machine and it performs its function, whatever it may be. So because that document is digital, I'm able to secure it. You know, we're able to secure it through multiple methods. One of those methods is through blockchain, right? We can put, like, if you think of it like a baseball.

I can take a baseball and I can put it inside of a glove, or I can put it inside of my hand, or I can just throw it, right? And once it's in the air, it's plain, it's open, right? It doesn't have anything around it, but we still know it's a baseball. When it's in the glove, The idea is you still know it's a baseball, right?

Because you've seen it. Well, what happens if it's not really a baseball? What happens if it's a tennis ball painted to look like a baseball to trick you? What happens then your mind is like, well, it's still a baseball. I saw it go in the glove, but upon further inspection, it's bouncy. And all of a sudden it's fuzzy.

Like, wait a minute. What is this? How, how did this happen? Right. So with these different cryptographic hashes, like a synaptic hash, right, we can take an object, a digital asset, and we can wrap it with a synaptic hash or wrap it with a cryptographic hash. And then on the inside, we can hit it with a synaptic hash or the inverse.

And that way we know that once we send something out, we know that it is indeed what I sent. Right? So there's no fake news going on, right? There's this isn't a fake document. You can't forge this document anymore because it's been signed with this cryptographic hash. And that's one of the great things about blockchain to prove providence of a digital artifact, especially manufacturing.

So what that means is that if I'm working on a document or I'm collaborating with a supplier in the defense industrial base, Um, they can rest easy and know that the document that I just sent them to manufacture is actually the document and it came from me. It didn't get hijacked mid, you know, midway through the stream or it's called a man in the middle attack, right?

It didn't get plucked out in the dataverse somewhere. And, you know, some meantime before failure was, Programmed into the part and then reinserted into the dataverse and then sent on to me. Right. So some file that is not what I originally authored, made it to the manufacturer. And now they're manufacturing a part that has a fatal flaw built into it.

Right. But at what rate is that fatal flaw going to show up in the platforms that it's going into? The answer to that is it's at no state of frequency. There's no distribution for that because there's no statistical distribution for that because. All the aircraft fly at a different rate, right? Some aircraft fly at a thousand hours a month.

Some of them fly at, you know, it might be a national guard aircraft where it only flies at a hundred hours a month, which means it takes it 10 times longer for that fatal flaw to show up if it's programmed at a thousand hours. Does that make sense? 

[00:13:20] Ryan Connell: It does. Yeah, no, I appreciate that. Maybe this is just my ignorance when I ask this question, but is the purpose for all this really the cyber security element or is it really the understanding the supply chain element or is it a combination of?

[00:13:33] Keith Scheffler: Yeah. So I really don't know in today's day and age, right? I don't know that we can, can we really separate the two? Can we pull them one thread without the other? Because that conversation always erupts, right? When I start talking about manufacturability of a part, you know, people are like, well, I can manufacture that.

Well, how are you going to do that? Well, I can use additive. Okay, great. How do I know that that additive part that you're creating is the one that we originally sent to you? How do we know that? Right? Because from, uh, you know, all of our, obviously anything that flies and I says it has to go under airworthiness testing, right?

So a part that is machined has certain characteristics, certain material properties to it, right? Versus a part that is used additive, produced via additive manufacturing. It could very well have like extremely similar, like very, very close and pass airworthiness testing. But at some cases it wouldn't be, there's some materials that we just don't have in our inventory yet to use for additive manufacturing.

So the point there is, is that they're so intertwined that when we start talking about manufacturing, especially with like robotics and automation, right? I have to send code to a robot. To do some function to do some discrete task or some continuous task, whether it's, you know, drilling a single hole or whether it's painting, right?

Painting would be a continuous task. Drilling a hole is a discrete task, whichever it is. I have to send coordinates to that robot to that robotic arm for it to do something to change a bit to change its tool head to perform the function and to do this thing right? That takes a digital file. Now, if you think about it, Uh, a bad actor or nefarious actor can step into this conversation at any point in time, because think about it, you know, you almost have to be in a wartime type situation to bring down a jet aircraft now, right?

I mean, you've got to fire a missile at it. You've got to do something. You've got to something kinetic usually has to happen to bring that aircraft down or does it? Like if somebody is able to get into a manufacturing facility and there's a robotic arm sitting there and I've got a hundred million dollar aircraft sitting there, that's carbon or it's carbon composite.

What can I do to it? Right. Pretty easy to grab ahold of an arm, you know, to program an arm to just go. Like the inertia from an arm is so ridiculous. Like the amount of energy that it contains and the damage that it can do, take that arm and run it into the fuselage of that aircraft. What have you effectively done?

You just trashed a hundred million dollar aircraft. Well, how did you do it without firing a single shot? You know, it's, it's a different way to look at combat. It's a different way to look at a threat. So they're so intertwined with one another that you really have to, you have to consider both. Right.

Does that make sense? 

[00:16:19] Ryan Connell: It does. Yeah. So like, I mean, I've never even thought about the concept of, I'll call it manufacturing warfare. Like I've never really contemplated that. So that's super interesting. Um, I got to regain myself here. Okay. So help me. Uh, I'm going to kind of pivot a little bit and talk on the acquisition side.

So I have experience in, in a previous role where I was constantly getting, I'll call them like legacy, but it's not really legacy, but like current platform. So not really emerging tech. Um, and we use the word commercial often. Conflated with emerging tech, but you're like, Hey, this is a commercial clamp or a commercial nut, bolt, screw, whatever it is, wiring harness, like any of those things.

And we get ourselves in, you know, the sole source arrangements with these huge manufacturers of platforms and, you know, the subcontractor or whatever tier that offers. Whatever that product is, you know, asserts that it meets the commercial definition and, you know, it's going to be the, you know, we start in Congress, right?

Like the 37, 000 client, right? What do you know about that? And in terms of like, what alternatives do we have to be able to find whether it's an additive manufacturing product, or if there's enough details in the data to actually identify similarities between another product that might exist somewhere else, if that makes sense.

[00:17:34] Keith Scheffler: Yeah. So that's a large part of my research as well. Right. So we're, and that involves graph theory and how we do it and look at the coordinations and the closest of those coordinations, how are these companies? Similar. Do they have similar capabilities? Do they, do they have the ability to manufacture X to how big of an autoclave?

Like if I know that they manufacture a certain carbon composite. Part, right? I can infer from that that they have an autoclave, right? And depending on the geometry of that part that they're manufacturing, I can infer how big that autoclave is. And then from that, I can infer, okay, I know, I know how long it takes to cure that part.

Oh, but wait, I also know that before we ever get to putting that part into an autoclave, that part has to have some layups around it. Meaning jigs or forms, you know, to support it while it's curing. So they have to have some type of machining capabilities to create that jig, to create that tool. So we're able to infer a lot of things from that standpoint.

Now, when you bring up commodity based, that's an interesting conversation, right? Because there's no, there's nothing new under the sun, right? There's no new thoughts. There's no, like, it's just, we're reinventing ourselves. We're finding different ways to do things. Or we're meshing things together to, to make a hybrid, right?

So when you say there's a commodity part, what does that really mean? Does that mean that yes, it's a bolt screw or washer that DLA typically orders, you know, an IDIQ that that's an indefinite contract that they're ordering this part that spans five different platforms, right? This lock washer is on, you know, six platforms and they just order them in bulk and then they, you know, from a distribution center, they distribute them as needed.

So you, let's say you, you've got Connell Industries and Connell Industries wants to break into the defense industrial role, but. That's pretty hard to do because there's programs of record out there. You got to get in with that. And then, or if you go bid on something within DLAs acquisitions piece, right?

You you've got a bid against all these other people. And then there's this whole other side of it, of like set asides for service disabled veterans like myself and such. So like there, there's a lot of different things you got to compete with. Right. So what's the next logical step? Right. What if, what if companies been doing since literally forever, right?

If somebody has something that you want and your company's not able to manufacture it, yeah, you gobble them up, right? It's, it's, it's either a hostile takeover or you just, you just straight up offering them a bag of cash and they walk to the beach and be like, see ya. It's, it's not worth that to me. So I'm, you know, I'm gonna go, I'm gonna go kick my feet up and, you know, and drink a umbrella drink, or whatever it is you do.

Right? I mean, that's, that's the most, that's the quickest way to it. And the funny thing about that is, is that if you look at the fence industrial base, we know in part, so we report in part, right? What's the one part of the defense industrial base that we're missing? We call it the P dip or the phantom dip.

Right. So the phantom div, we kind of coined it as the P div and the P div is all the people, all the secondary and tertiary, everybody, all the nth tier, all the way down to the last, very, very last tier that doesn't have a cage code for those listeners that don't know how to do business with the department of defense.

You have to have a cage code, right? You got to prove who you are and all these other things. You got to get this code. So if you don't have a cage code. You can still be a DoD supplier. You're just not, they're just not giving you the contract for it. Right. So these PDs are out there and they've got all these capabilities and you've got these people that do have the cage code, right?

They've got the secret decoder ring and they want to supply these parts. So they're going to some of these, you know, secondary tertiary manufacturers. And if they don't have the capability, they're buying up those people. Right. If somebody is outside the dib, the defense industrial base, and they want to break in, it's the inverse of that.

They go buy into it, right? Just like investing, except they're taking over. Right. So they, they buy up that capability. So that commodity piece, yeah, it's a commodity, but it's not really the commodity that we think it is anymore. Right. So you really got to change the way that we think about a commodity piece.

It's a commodity, but not everybody can manufacture that because. Our capabilities are shrinking. The defense industrial base is shrinking. So we've got to do all that we can to be able to identify that shrink. And that's again, part of my research is that, you know, using statistical modeling and different distributions of, we know that there is a sector or some part of the defense industrial base that is either dwindled or has died, right?

Just straight up dead. I can take that, find a distribution that fits it or fit a distribution to it. And then I can take that model and I can push it against some other industry that we think might be declining, but at what state of decline is it in? How do you know, we take something that we do know, find a distribution that fits it, and then we model against it.

Right? So then that's what helps us within this graph theory is be able to identify. At what rate or who do we need to get involved to be able to understand, like, this is potentially an issue or this is something we need to look at. Right. Just the same as when you go and you, when OSEA gets involved in, they're like, Hey, foreign involvement.

Yeah. We're, you can't sell this company because it's critical to national defense, similar to that. 

[00:23:01] Ryan Connell: I'm going to try to say it in my own words, make sure I'm tracking, but you take something that is potentially scarce, scarce in, in America, uh, mercury or, or some metal or something like that. And, and you say, Hey, there's.

One or two companies that sell this in this form and you are trying to predict how much that industrial base is shrinking the likelihood of one company taking over the other and you're doing that by and I'm just going to assume that they're analogous saying like, hey, nickels also shrinking and here's the rate in which it's shrinking and we can take that model and apply it to mercury, something like that.

Yeah, 

[00:23:34] Keith Scheffler: for sure. 

[00:23:35] Ryan Connell: We can absolutely 

[00:23:35] Keith Scheffler: do that. We, and that's essentially what we're doing is, is just, we have to identify and truly we don't even have to identify it, right? It's a scan type principle because graph theory, you can, it's not like a traditional relationship database where you have to scan from a linear standpoint, right?

You got to scan every line from a graph standpoint. You can insert yourself anywhere in the graph at any point in time in the node. Okay. Find a node and insert into that. And then we can just turn around and say, yes, this has relationship or this is how closely it's related to it. And if it is related, can we fit that distribution to it?

So this is continually scanning. It's not something that we just say, Hey, this is what we got to do today. No, we continually scan for that stuff. 

[00:24:19] Ryan Connell: Yeah, so it's so interesting, right? So, so you were supporting the Air Force very easily a cog in the wheel within DoD, right? But there's Ryan's opinion like miles between what the average DoD acquisition professional knows and is doing and what you're supporting.

Yeah. And how do we make that not miles and make that inches so that we're equipping the 180, 000 professionals with, I'll say like all of the knowledge and research that you have. And like, are we years away from that? Or like, what does that look like in your mind? 

[00:24:51] Keith Scheffler: Yeah. You know, it's just like anything else.

Like you can think of the services, different companies, right. But we're owned by the same umbrella, which is DOD. Right. But we all have a different budget and we all have different needs. Yeah. We think we all have different needs. We think we all have different, you know, that things are far apart from one another.

So we start creating these, you know, these spirit projects or efforts and then we're like, Oh yeah, I've been working on mine for three years. Like, Oh, I just started and I, you know, I've got 3 million and I'm spending on it. So like we, Across the services, this is an issue, right? And that's one thing that we're doing a really good job of right now is at the OSD level and OUSD.

We've got some great folks like it, you know, at the industrial based policies division, you know, we've got some folks up there. General Howard's very, very forward thinking and how she's looking at these things and the policies that they're putting in place for it. So we're able to partner with folks like that all the way down to, you know, elements like the four, four eighth that are actually, you know, Ordering the parts.

There are the folks that are looking at it and saying, you know, this is part of sustainment So we're able to look at the full gamut of what does it look like across the board? And then you've got other other departments other entities that are trying to help scale this across services, right? So yeah, we're looking at these things, but we're trying to apply more Analytics to something that isn't really known for you know, the application of analytics 

[00:26:14] Ryan Connell: Yeah, maybe one kind of last question for you.

I'm curious. So, so as you kind of outlined, you know, all of the elements from the security and being able to pass information from beginning to end in a secure domain all the way to the commodities and finding what else exists in the market, you know, the shrinking defense industrial base. What do you want to kind of offer the listeners in terms of Okay.

I'll say main takeaways or, or advice or how to kind of reach out and potentially even like, learn more about what you're doing so that we can kind of spread this type of thinking. I'll kind of turn it over to you to hit any of those. 

[00:26:48] Keith Scheffler: Yeah, I would tell you that for my, the main effort that I'm working on is, you know, it's one of those that, that we're, it's continually evolving, right?

Blockchain is one of the pieces to it. And now granted, it's a very foundational, very essential piece to it. Everything is built off of that and it gives us the autonomy to be able to add so many things to this from a linear standpoint, things are coming out of the woodwork at us right now of like, Hey, we need to do this, or, Hey, can you do this securely?

You're like, yeah. No, no problem. We can do that. You know, ingesting data from literally anywhere and then applying artificial intelligence to it. That's kind of what we do. You know, the artificial intelligence back end that we built is, is very Lego esque. If you will, you know, you can, it's a brick, you can literally pull this brick out this module out and plug it into whatever application you Think that it might be, it might fit for.

Right? So that's part of that umbrella technology principle that we work off of the day or FRO. So we're building these, these technologies that are transferable. So it's not that you have to use them on my platform. They're transferable to your platform, whatever it may be. So I would encourage folks that to think outside the box, right?

A lot of times we're constrained, but. I have the liberty to say that because, you know, that's, this is all I do for a living, you know, I research, right? So I'm able to think outside the box and my leadership is unbelievable at that. You know, they empower us to be, you know, they empower us to be able to do these things and within reason, obviously, but they empower us and they, they know that we're subject matter experts in these areas and they trust us to go do these things.

And they work out right. They take big gambles and they work out to the point of where we're at today. You know, we're able to take these technologies and apply them in such a way that we wouldn't have been able to do 567 years ago. Right? So that's kind of where we're at today. And I would encourage folks to think outside of that box.

If you have a problem that you know that you're thinking like, yeah, we've been doing it like this forever. Should you still be doing it that way? Like, is that like, that's one of the main things that I talk to people when I start gathering requirements. Like, is that really what we should still be doing?

How would you like to see it done? Right. Ask yourself that question. 

[00:28:55] Ryan Connell: Incredible insights, Keith. Appreciate you being on today. Would invite anyone. I hope that's okay to connect with you. Yeah, sure. And I mean, just like you and I did when we first met, you know, I was like, wait, do you know about this project?

Do you know about this? Like, and that's kind of how you kind of can continue research and spread the ideas across the department. So, hey, appreciate you being on today. 

[00:29:13] Keith Scheffler: Yeah. Appreciate the time. Thank you so much for the time that you've given to it and just appreciate everything that you're doing as well.

Awesome.