Latest / The ReForge Podcast / We Could Open Two Factories a Year (with Caleb Chamberlain) | ReForge 028

Transcript

We Could Open Two Factories a Year (with Caleb Chamberlain) | ReForge 028 · The ReForge Podcast

• 0:05
  Welcome to the Reforge podcast.
• 0:07
  I'm Matthew Chang, and we're here to explore all things AI, automation, and the American worker, proving that the American reindustrialization is possible, pivotal, and pressing.
• 0:19
  Thanks for joining us today.
• 0:21
  Let's get into the episode.
• 0:25
  Welcome back everyone to episode 28 of the Reforged Podcast.
• 0:29
  We are joined today by a gentleman that I'm sure most of you will recognize, Mr.
• 0:35
  Caleb Chamberlain of Oshkut.
• 0:38
  Am I saying it right?
• 0:39
  Oshkut or is it O-S-H Kut?
• 0:41
  Uh, Oshkut is right.
• 0:43
  Okay, good.
• 0:43
  I've, I've...
• 0:44
  I, I sometimes question myself.
• 0:46
  Uh, he is CEO of Oshkut, has an actually a very interesting background, uh, that might surprise some of you.
• 0:53
  Um, so just wanna welcome him to the show.
• 0:54
  Caleb, thank you so much for joining us.
• 0:57
  Yeah, happy to be here.
• 0:59
  So o- obviously, you know, I didn't do you justice.
• 1:01
  Give us a little bit of your background.
• 1:03
  Tell us a bit more about who you are, you know, and, and kind of what led to, what led to Oshkut?
• 1:10
  Yeah.
• 1:11
  So originally I, uh, was an electrical engineer, have a master's in electrical engineering with a control theory, uh, emphasis.
• 1:18
  Um, I did, uh, actually, uh, a lot of research around drones.
• 1:23
  Uh, we didn't call them drones back then, we called them quadrotors but, uh- or multirotors.
• 1:28
  Uh, but now they're drones.
• 1:29
  Um, but I did system identification, state estimation, and control of, uh, unmanned aerial vehicles.
• 1:35
  That was my master's thesis.
• 1:37
  Um, I went from there- Seems pretty relevant right now, actually.
• 1:40
  Right?
• 1:41
  Yeah.
• 1:41
  I, I opined that it's kinda funny that I'm actually not doing anything in that space now.
• 1:46
  Um, but my, my first company I started as a grad student and we made, uh, inertial sensors, think, uh, accelerometers, rate gyros, magnetic sensors, uh, GPS, and we did, uh, little micro sensors that, that, that provide state.
• 2:01
  It was pretty novel back then, uh, less novel now.
• 2:04
  Uh, and we also did some Air Force SBIR contracting.
• 2:08
  We did a covert precision aerial delivery system where we had, um, uh, built an autopilot that steered a silent, uh, winged vehicle, uh, over a combat zone to deliver payloads, uh, autonomously.
• 2:22
  Um, I, uh, I had a stint doing some medical device stuff with Owlet.
• 2:28
  We made, um, uh, pulse oximeters that went on a, on a infant's foot that they could sleep with comfortably and provide, uh, blood oxygen readings.
• 2:37
  Th- those little sock ones?
• 2:39
  Uh-huh.
• 2:39
  Yeah.
• 2:40
  Yeah, I remember that.
• 2:40
  I was looking at those when, you know, my, my youngest, uh, was born.
• 2:44
  I, I geek out on that stuff.
• 2:45
  But wow, that's pretty cool.
• 2:47
  Yeah, yeah.
• 2:47
  The, it's a load off your mind to know that your, that your newborn's safe.
• 2:51
  So, uh, uh, I was se- a senior electrical engineer on their V3 Smart Sock project.
• 2:57
  Um, and I also spent some time at a synthetic aperture radar company called IMSAR.
• 3:03
  Um, and we did, um...
• 3:05
  We- I, I worked, actually worked on sort of a side project where we did, um, a time of flight, uh, based localization of active RFID tags.
• 3:14
  Um, in any event, uh, eventually, um, my brother and I, uh, who started Oshkot, we started building things together just for fun, and one of those projects was a, this VR motion rig, virtual reality motion rig.
• 3:30
  Nice.
• 3:31
  So you're, like, flying an aircraft, uh, or a spaceship or whatever, uh, and, uh, wearing a VR headset.
• 3:39
  And the problem with, with VR is that if you see motion that you don't experience it makes you sick, um, it makes you really nauseous.
• 3:46
  It's not fun.
• 3:47
  Um, so we thought, well, what if the, the chair you're sitting in moves in a way that accurately reflects the motion you're seeing?
• 3:54
  Uh, and so we built this thing.
• 3:56
  But we just found that it was really, really hard to buy fabricated parts.
• 4:00
  We did a lot of tube, we did a lot of sheet metal, and it was just like you would spin up a design in a couple of days and then spend, you know, three weeks to a month to get it made.
• 4:10
  And, and we thought,"Hey, there's gotta be a better way to do this," and that led us to start OSHCut.
• 4:14
  So here we are eight years later.
• 4:16
  Well, I think Colin, after he g- after the CNC machine that he's trying to buy at Costco, I think his second CapEx purchase is gonna be for the VR rig that you build.
• 4:29
  I want one so bad.
• 4:29
  We're building another vision.
• 4:31
  It took us, it took, it took us seven years to finally build that machine, um, using OSHCut, uh, parts.
• 4:38
  Um, but we're working on another version now that's, that's going to be much better, so, um, we'll be- Well, I, I saw the video you posted of, like, the w- with the RC car doing- Yeah like, the, the off ro- Like, that is just was too cool.
• 4:52
  I mean, I'm, I'm sorry, that stuff just, I, I geek out on it, but that was just too fun.
• 4:56
  Yeah.
• 4:56
  One of our, uh, one of our senior software engineers also does a lot of, uh, like RC airplanes and FPV stuff, and we're gonna hook telemetry from an FPV drone up to that chair, so you'll be wearing it and, like, flying around.
• 5:10
  We'll see how it is.
• 5:11
  The mo- the machine can't move as fast as, like, a, uh, an RC, a little RC drone.
• 5:16
  Um, so there might be some kinda smoothing there that makes it not great, but it's easy to try, so.
• 5:22
  Colin, I'm feeling a lot like OSHCut as we know it is just generation one of the company, and there's all this sensing and computation that's gonna be going into the fabricated metal parts, you know, in the next...
• 5:35
  That's just what I'm sensing, so.
• 5:37
  I, I mean, g- I, that might be a good little segue for some of the, you know, recent exciting news, you know, out, out of OSHCut.
• 5:44
  Obviously, you know, w- we talked about the second facility, um, that, that y'all recently opened up.
• 5:50
  But you also opened up, or, or announced recently, an R&D center, right?
• 5:55
  Uh, that's right.
• 5:56
  Uh, we're opening up that space in September.
• 5:59
  Uh, it's not a massive space.
• 6:01
  It's 20,000 square feet, but it's dedicated to, um, supporting projects like the motion rig and piloting new services.
• 6:10
  Um, so the rationale will be, hey, we'll...
• 6:12
  Anytime we need something...
• 6:14
  The way we look at OSHCut and, um, building and design is that it's kind of like industrial Legos.
• 6:19
  Um, so we wanna set it up so that, um, you can build pretty much anything just by buying stuff online, whether it's a commercial off-the-shelf product, like something you can buy on McMaster-Carr, or whether it's custom metal you can get from OSHCut.
• 6:36
  And we wanna do it all the OSHCut way, meaning, meaning, uh, manufacturability is exposed from first principles right in your web browser.
• 6:44
  So by the time you get around to placing an order, you know it's going to arrive on schedule and everything's gonna work just, just fine.
• 6:51
  Um, y- it...
• 6:52
  I mean, it's the most disruptive thing ever when you place an order and, um, somebody reaches out two days later and says,"Actually, we can't make this." So we wanna get as far away from that as possible and just say,"Look, design, build your custom parts.
• 7:05
  If it's made to spec, you'll get it, and, um, it'll be easier than ever." It ch- kind of changes the way you can design and make things.
• 7:13
  And so, so on the R&D space, it's like, okay, well, if we can't...
• 7:17
  If that experience doesn't exist and we need something, um, then it's a candidate for a new service.
• 7:24
  We get the equipment, put it in the R&D space, uh, figure it out, learn it, you know, launch it behind a feature flag for select customers.
• 7:31
  Um, and then when it's ready for prime time, we just go all in and launch it as a new service.
• 7:37
  Well, and one of those new services that y'all added not too long ago was obviously the inclusion of tube.
• 7:42
  Um- Yeah you know, a- adding tube options, which is obviously huge, um, just because that's one of those areas where you don't see a lot of options out there.
• 7:52
  You don't have a lot of the instant quote and such a supply.
• 7:56
  What was it that made you guys choose to go with tube versus, I don't know, any other service?
• 8:02
  Was it just purely demand?
• 8:03
  I mean, I see you guys do a lot of off-road stuff too as well.
• 8:06
  Um, it wasn't demand per se.
• 8:09
  It was mostly, it was a service that we wanted and needed.
• 8:13
  Um, so, um, you know, uh, it's probably could be argued that we should have done a bunch more market research and, uh, the truth is the tube is pretty niche.
• 8:22
  Um, we just launched tube bending as well, uh, CNC tube bending on a draw bender.
• 8:27
  Um, and that's very, very niche, and I don't think a lot of people design for it.
• 8:32
  But man, once you realize that that's a tool in your tube tool bag, it changes how you design.
• 8:37
  Um- Mm.
• 8:38
  So that's kind of...
• 8:39
  We have a long view on that.
• 8:40
  It's like, you know, we launched tube in 2024.
• 8:43
  Um, it's still a very, very small percentage of our business, and it's expensive.
• 8:48
  You know, a, a state-of-the-art tube laser is$2 million.
• 8:51
  Um, a tube bender could range from, you know, quarter million to half a million to a million dollars, depending on the size, and it's a very small revenue center for us.
• 9:02
  Um, so I claim that as kind of a pet project, um, uh, just knowing that, hey, in 10 years, a lot more people are gonna be using this because we've made it accessible.
• 9:12
  Well, I like the idea of your select customers filling up your R&D space with their projects, right?
• 9:19
  So they come to you with a thing, and they say,"It would be amazing if the tube bent itself, if the tube cut itself, uh, if the tube welded itself." And then that gives you the...
• 9:28
  You know, you can make a decision obviously, but, uh, the, it fills up your idea space without the folks at OSH Cut having to think of everything the industry might buy, build it and they will come.
• 9:39
  It's like with every project, you have customer number one, at least, in the bag, and you can kind of learn with them.
• 9:46
  Do they, do your best customers ca- come to you with ideas and then, and then talk it through with you?
• 9:52
  Is that how this comes about?
• 9:55
  Yep, they do.
• 9:56
  And, uh, you know, sometimes customer number one is just us.
• 10:01
  I actually think that's important.
• 10:02
  Um, I think that, um, so much of the, the like, the, the thing that makes big companies bad is when you get like, uh, not to knock MBAs, but like professional MBAs and finance guys, uh- in the executive seats, and they're manipulating spreadsheets, and they're, um, they're financializing everything.
• 10:25
  Uh, you know, you've got a, a business that produces cash, and they're all about like,"Well, how, how do we invest this cash?" And, you know, putting it in equities or whatever.
• 10:33
  We're all about,"Look, we're, we're here to build, and, um, we want to make this as good as possible for people like us." And if that makes us a billion-dollar company in the future, awesome.
• 10:43
  If it makes us$100 million company in a couple years, uh, awesome.
• 10:48
  Um, if we stay at, at about, you know, 60 million a year, um, that's awesome, too.
• 10:53
  Uh, you know, it actually doesn't matter to me as much as, as, uh, creating this awesome way to build.
• 11:01
  Um, and the revenue, I think, will fall naturally from that, depending on how it works, so.
• 11:07
  So a question, after you started as, as this kind of online, you know, designed order company, deliver on time, AI happened, what, since you built the company.
• 11:19
  How much is the vibe coding and, and being able to use agentic AI to streamline the workflow for customers, how much has that been a factor?
• 11:27
  So for us, it's mostly about software development, and we do use it very heavily actually.
• 11:32
  But it's transformed the way we develop.
• 11:34
  I mean, we've got, uh, I think eight full-time developers, um, and, um, and we're growing that team.
• 11:43
  We wanna double that team within the next six to 12 months.
• 11:46
  Um, but we have...
• 11:48
  We, we embed the AI-driven workflows right into how we, how we test, how we deploy, how we write code, and it does accelerate a lot.
• 11:58
  Uh, it's very significant actually.
• 12:00
  Awesome.
• 12:01
  Well, one of the cool things that you guys recently added was kind of that, uh, that for your excess capacity, where people, if they- they've got the flexibility on it, uh, on when they get something, you know, they can get parts and pieces made for much cheaper just because, you know, it's an opportunity for you guys to fill your capacity.
• 12:20
  In kind of that arena, one of the things that I've loved, uh, seeing you talk about a little bit is your use of the Monte Carlo sim- simulation models.
• 12:30
  You don't have a lot of people on X or general discourse who talk about some of that more technical analysis, quote unquote, if you will, to look at all the different possibilities of here's, you know, the likelihood of us being able to hit all of our work within this time period.
• 12:49
  Here...
• 12:49
  Rather than saying,"Our capacity is X," you know, y- you guys are taking a much more probabilistic approach to it, and I, I think it's really, really great to see and really cool, but not a lot of people talk about it.
• 13:02
  Um, obviously, your engineering background, um, probably contributes a bit to some of that.
• 13:08
  So maybe can you speak a bit to how you guys approach the capacity piece?
• 13:14
  Because you did just add a whole second facility as well, that obviously increases that, too.
• 13:19
  Yeah, there's a couple pieces to that.
• 13:21
  So we're really focused right now, on the capacity front, we're really focused right now on, uh, resilience, um, and proximity.
• 13:29
  So, um, it's really, really important that as we scale, that our customers have a great experience and they know that OSH 切 is the quality option, and we're leaning heavily into that.
• 13:39
  Um- And, uh, you know, a big piece of that is, look, y- we just have to optimize around excess capacity because you can't predict the future.
• 13:46
  So we need to kind of gear our CapEx plan and financial modeling around always having those extra machines and people.
• 13:54
  Uh, and it, it is expensive, but I think in the end it, it really pays for itself because, um, you know, equipment, uh, depreciation actually works out to be a fairly small percentage of our cost of goods sold.
• 14:05
  So we have a little bit of wiggle room there.
• 14:08
  Um, meanwhile, I think that, uh, uh, you know, when, when people understand that there's a reliable option, I think they'll just tend to kinda go there.
• 14:16
  You get, you get people in rebuy mode, and it, it's an investment in the future.
• 14:21
  Um, what was the second piece of that question?
• 14:23
  So we talked about capacity.
• 14:24
  Uh- Oh, Monte Carlo.
• 14:25
  Um, Mon- Yeah, I mean, I, I geek out on some of that just'cause I think it's great from...
• 14:30
  You know, in a, in a high mix, low volume shop, capacity can be a little bit difficult to accurately represent as a true single number because that mix of work.
• 14:40
  So, you know, with Monte Carlo, it, it, it's would seem useful since you can model so many different possible combinations of incoming work rather than focusing on, you know, some utilization number that is kind of not reflective of reality perhaps.
• 14:57
  Yeah.
• 14:58
  Monte Carlo can allow you to predict bottlenecks.
• 15:00
  It can help you understand, you know...
• 15:03
  We actually have code that allows us to simulate incoming orders sta- that are statistically indistinguishable from real orders and that use real parts.
• 15:11
  Uh, so effectively we say,"Hey, okay, you know, here's the, here's the distribution on arrivals.
• 15:16
  Um, here's the distribution on, uh, parts per order, on likely materials, on quantities." Um, and then you can just say,"Okay, starting from today, um, give me a bunch of synthetic orders that are believable based on our current volume.
• 15:32
  Okay, now do the same thing with 20% more.
• 15:35
  Now do the same thing with 100% more.
• 15:37
  What breaks?
• 15:37
  Where does the bottleneck arrive?" And then you can proactively plan and say,"This is the machine I need to handle this problem." Um- Love it the, the way the simulation works, just if people aren't familiar with Monte Carlo, um, it's basically you take something that's very complicated and impossible to simulate from first principles, uh, accurately, and you say,"Okay, well, we're gonna model it statistically." So, um, the way that works in practice is you've got, like, how long is it gonna take to make this part?
• 16:06
  Um, well, a person can say,"You know, I, I'm not sure, but I think it's gonna take, you know, five to 10 minutes, um, to weld that together." It's like, okay, well, we could either, either produce a distribution on that style of part analytically and build a statistical model, or we can say,"All right, five to 10 minutes, so let's put a uniform distribution on that." And then what happens is you, you build a model, um, that reflects every operation in production, uh, and then you simulate it end to end And say,"Okay, well this time it took five minutes, this time it took 10 minutes." You simulate 1,000 times, and then at the end of the day you have an outcome that says,"Hey, this part shipped on time 85% of the time." Um, what can we do?
• 16:49
  How can we change the scheduling heuristics to nudge that number up, you know, to 90% or 95 or, or 100%?
• 16:56
  Um, there's always, like, late scrap and stuff, and you model that too.
• 17:00
  Um, so there's, uh, there's the statis- statistical part, um, and then there's, um, what I call, you know, part selection heuristics, and that is what do I work on next?
• 17:12
  And that question is asked 1,000 times at every manufacturing operation every day.
• 17:17
  So that's an impossible thing to, to answer optimally actually.
• 17:21
  Yeah.
• 17:22
  It would take more than the age of the universe to simulate all the possible outcomes.
• 17:26
  Yeah.
• 17:26
  Uh, so what you do is you say,"Well, I'm gonna build a sim- a s- a simple rule of thumb," it's called a heuristic,"and I'm gonna run that, run the simulation on that, on projected incoming orders and existing.
• 17:38
  Um, what's my on time shipment percentage?" And then you design a new heuristic and you say,"All right, well, what if I do the same...
• 17:45
  What if I run the same simulation but I change how I choose what parts I work on next?" You run that and it's like,"Hey, I just got 5% better on time shipment with my existing capacity." Um, you can use it for labor allocation, like where should people be working and how?
• 18:00
  Um, and there's a lot of nuts and bolts that actually make this work, but, um, but it's pretty cool technology actually.
• 18:07
  I, yeah, I mean, I, I think one of the things that's really interesting is that a lot of manufacturing management approaches are based on averages.
• 18:16
  Um, you know, looking at your average cycle time, lead time, scrap, all of those pieces of it.
• 18:21
  But averages, you know, ha- have a tendency to hide some of that tail risk that causes issues down the line, but where Monte Carlo focuses more on that probability piece of it.
• 18:35
  What is the, what is the probability that I can achieve X, Y, or Z?
• 18:40
  Has using, or rather, does using the Monte Carlo approach, running those simulations, change the way that you think about managing the business, kind of ma- making some of those decisions and managing, um, so- some of the different levers you have to kind of pull on?
• 19:00
  Yeah, I mean, I can give you an example that's actually quite intuitive.
• 19:04
  Um, if you're using just averages, um, then effectively you're saying, all right, well, it's gonna take this long on average to make this part, but, but there's...
• 19:14
  Let's, let's consider scrap rate, for example.
• 19:17
  Um, let's say usually you don't scrap a part, but there's kind of, uh, this blip at the end of the probability distribution that says, you know, they get scrapped this percentage of the time.
• 19:27
  So Monte Carlo will sample that periodically and say,"This got scrapped," and okay, it gets scrapped, now it has to be re-released into production.
• 19:35
  Um, the simulation captures that.
• 19:38
  Um, and so if, if you're waiting till the very last minute to release something to production, Monte Carlo notices that, you know, maybe 2% of the time there's a scrap problem, and that makes the order ship late, and that affects how you need to schedule.
• 19:52
  Says,"Well, to account for this likelihood of scrap, I actually have to release early enough that I can push this through production twice and still, um, get it on time." Um, so I, I will say that it's debatable whether all of this is, is what everybody should do.
• 20:09
  I mean, it is really cool and the engineer in me gets really excited about it.
• 20:12
  Um, but what I just described is also very intuitive and something that you could just kind of design a process around without having to build and maintain a big Monte Carlo simulator.
• 20:24
  Um, but it is still- But where's the fun in that if you don't get to- There's the fun factor, for sure.
• 20:32
  Um, but you know, for cap- capacity planning, bottleneck prediction, um- W-, you know, being proactive about...
• 20:39
  This is something we're working on right now, is being proactive about, um, notifications.
• 20:45
  Like, if something's likely to ship late, it's very good to reach out to the customer ASAP.
• 20:50
  Um, but when you've got, you know, 5,000 orders in queue and, you know, 30 th- 32,000 SKUs and it's all in the production mix, and the customer's like,"Is it gonna ship on time?" And you look at that and you're like,"Oh, man, I, I don't know.
• 21:04
  Your guess is as good as mine, honestly.
• 21:05
  I have no clue.
• 21:06
  I, I know that we're busy." Um- but the Monte Carlo actually puts a number on that and says,"Yeah, 50% of the time this ships late, so, um, let's see what we can do about that." But...
• 21:18
  That's crazy.
• 21:19
  I mean, I...
• 21:19
  The way I look at it is all the, the statistical analysis and then just kind of being fanatical about your equipment selection and labor allocation, to me that all just builds to customer service.
• 21:31
  So it's like it, it plays into...
• 21:33
  I like the way you framed it, Oshkosh, as the, it's kind of the premium solution, it's the high quality solution.
• 21:38
  Not saying it's the most expensive, s- you probably beat, win on price quite a bit, but it is, it is predictable.
• 21:45
  And I'm, I'm the tail end, right?
• 21:48
  So I'm aggregating, in the factory business, a lot of supply chains and when we have risk it sure would be nice to know about it.
• 21:56
  We can make alternative plan and my...
• 21:58
  Everything I'm doing is going in the field for a factory.
• 22:02
  It would be nice to know, because on the, on the receiving end, we're planning labor and we're planning specialty, uh, you know, commissioning agents and engineers and pro- probably controls and software all to come together at one time.
• 22:15
  If we have something that's gonna hang it up, I probably wanna let the folks that I'm traveling in know about it so we can make other arrangements and shift our focus to maybe another machine center, for example.
• 22:26
  Um, so I love that and I wish everybody did it.
• 22:29
  You should Well, we don't quite do it yet.
• 22:32
  I...
• 22:32
  Right now I think in general, if there's a problem, people find out about it on the day it's due, and I don't like that at all.
• 22:39
  So that, that's an active area, um, to take those simulation results and turn them into actionable insights the customer can get early.
• 22:48
  Um- Um, yeah.
• 22:51
  The challenge is it's gotta be pretty accurate.
• 22:54
  You're only as good as your models, and the last thing you wanna do is send emails out to dozens of customers and say,"Hey, this might ship late," and then ship it on time, or, you know, or the reverse, right?
• 23:04
  So we've gotta make sure everything's just dialed in, um- Yeah and accurate before we close the loop like that.
• 23:12
  Yeah, that's amazing.
• 23:13
  Absolutely.
• 23:14
  Well, so, you know, I, I can geek out on this stuff for a bit l- for as long as, you know, we could go- Yeah but we'll, we'll shift a little bit to it.
• 23:22
  Um, obviously big news with expanding to Cincinnati.
• 23:25
  Um, I would just like to say for the record, I did guess correctly that it was Cincinnati.
• 23:30
  I'm gonna pat myself- on the back there.
• 23:32
  Um, I don't know why, but whatever.
• 23:35
  Um, no, talk to us about, you know, that selection process.
• 23:39
  You know, we- we've talked a bit, um, on previous episodes about what it takes to kind of get a new factory going.
• 23:46
  You, you guys are getting a new facility, you know, expanding.
• 23:49
  You're in Utah, adding Cincinnati.
• 23:51
  Ta- tell us about that, the drive to it, and kind of what that's been like.
• 23:56
  So it's actually gone way smoother than I expected, which isn't something you might say when opening a new factory.
• 24:04
  Um, when we moved in 2024, we moved from a, like a 20,000 square foot facility into 55,000 square feet, and we moved while keeping production simultaneously running.
• 24:16
  Ooh.
• 24:16
  So we would, like- Wow shut down one machine and move it, bring it up at the new site, and then we're doing production in two spaces and...
• 24:23
  Anyway, that was a real challenge.
• 24:25
  Do you have a lot of WIP from- Yeah site to site B?
• 24:29
  Yeah, we tried to avoid it as much as possible, but it happened, and it was very disruptive.
• 24:33
  It was, it was very stressful.
• 24:37
  If you can avoid it, I would recommend, uh, avoiding it.
• 24:40
  But anyway, we wrote all the software to support that.
• 24:43
  Um, multi-facility, you know, transfers, all that kind of stuff, tracking inventory per facility.
• 24:50
  Um, so when we added Cincinnati as a site, um, it just kind of flowed, and we had to tweak software a little bit, but for the most part it just kinda happened.
• 24:58
  Uh, meanwhile, all our production operations are very, um, software managed and software driven, so, um, it's, it's gone very smoothly.
• 25:08
  Um, I actually think, uh, my operations team is gonna maybe kill me for saying this, but I think we could probably open, um, two or three factories a year, um, if we had a team dedicated to it, and it wouldn't be, uh, a big deal operationally.
• 25:23
  That's good.
• 25:23
  Uh, the talent portion would matter.
• 25:25
  We'd need good GMs and shift leads in every facility.
• 25:28
  That's probably the hardest part.
• 25:30
  Um, but we could, we could launch dozens of factories, um, and it would just flow.
• 25:36
  Yeah.
• 25:36
  How about on the state side?
• 25:37
  Did you use the, like the, um, Ohio Economic Development?
• 25:43
  JobsOhio or whatever?
• 25:44
  Is the drive- Yes, we're actually, we're actually...
• 25:46
  We're five minutes from Cincinnati, but we're actually just south of the border.
• 25:49
  Cincinnati's right on the border with Kentucky.
• 25:50
  So technically Kentucky, yeah.
• 25:52
  Yeah.
• 25:53
  Oh, does it cover- We're in Kentucky, but we're, we're right next to the Cincinnati Airport.
• 25:57
  Yep.
• 25:57
  Nice.
• 25:58
  Yeah, there's a- So, so- there's a ton of manufacturers in that region.
• 26:02
  I mean, it sounds like you'll have to really start serving the aerospace community a lot,'cause I know they make just tons of, uh, tier two aerospace parts in that region.
• 26:12
  Uh, we already do, uh, a lot of, uh, aero and, uh, uh, am- among other things.
• 26:20
  Um- Yeah, it's, it's interesting.
• 26:23
  We've actually kind of hypothesized that the best model for Oshkot going forward is to find, um, areas where there are as many shops as possible, like as many traditional shops, and drop shops there.
• 26:35
  Um, the, the presence of traditional shops signal plenty of demand, and our model is to capture the long tail of work that those shops don't want.
• 26:45
  Uh, my thesis is that, um, there's plenty of that long tail demand even in these heavy manufacturing areas, and perhaps especially in these heavy manufacturing areas, so, uh, that we can scale the best and the fastest by, by dropping shops where there's the most manufacturing presence already.
• 27:04
  I mean, it, it, it makes sense.
• 27:06
  It's one of those things where you hear somebody say it and it's like, well, duh, obviously, if you're, you're going for that concentration of it, like there's...
• 27:12
  Somebody's proven it a bunch of times with all those.
• 27:16
  Um, but I, I don't think I've heard anybody a- actually say it quite that plainly of we're looking for that concentration because it's clear that that's established.
• 27:24
  So that's, that's certainly interesting.
• 27:26
  But, um, you know, w- was that really just that, that...
• 27:31
  Was it as simple as that, saying,"Hey, that's the concentration, we'll go there," or was it kind of strategic where you guys...
• 27:37
  You know, being in Utah, you've got great proximity to kind of that western half of the United States.
• 27:41
  Was this as much to try and open up the eastern, uh, portion of, of the, uh, country as well?
• 27:48
  Yeah, you wouldn't believe it, but actually, you know, New York is one of our biggest states, and it's mostly freight.
• 27:54
  So somehow we've managed to collect a bunch of work in New York, and we ship it freight from, from Utah, which is like five-plus days freight, um, al- almost always on a rail, uh, rail car.
• 28:06
  Um, and it's always been kind of like, you know, we scratch our heads at that.
• 28:09
  We ship a lot to Boston.
• 28:11
  Um- Hmm and it's just like, why are we, why are we shipping all the way across the country?
• 28:16
  W- Isn't there a better option out there?
• 28:18
  Um, and the answer is no, and not in, not the way we do business.
• 28:22
  So, um, really we...
• 28:24
  It, it's kind of funny because we offer same-day turnaround.
• 28:27
  Like, you can place an order today and it'll ship today.
• 28:29
  Um, but if it's going freight, then, like, your lead time jumped from a few hours to, you know, five or six days because it's gotta, it's gotta get on a train and go all the way to Boston.
• 28:41
  Um, so we just realized, hey, we've got a lot of customers that we're already serving east of the Mississippi.
• 28:47
  Uh, shipping is 50% or more of the lead time almost every time, so let's, uh, let's put a shop in a place that has great, uh, proximity to existing customers, and that was the real selection criteria.
• 29:00
  So since that exists- Well, Colin and I want to impose on you, Caleb, and, uh, sorry to cut you off on that thought.
• 29:05
  That's okay.
• 29:05
  But, uh, since site two is in the books, we want site three and four to be Erie and Jacksonville.
• 29:13
  C- Colin- Com- completely, you know, no self-interested aspects of it here at all, obviously.
• 29:19
  Not whatsoever.
• 29:20
  We'll just keep expanding the country for you.
• 29:23
  No, it's interesting.
• 29:24
  Uh, I, I might expose some of our, um, um, some of our thinking here.
• 29:30
  We haven't made solid decisions about the direction we wanna take the business, but there exists a model where we have dozens of, dozens of shops that do 80% of our services, like the real, like the real popular ones, and just, like, put a bunch of hyper-efficient sheet metal shops all over the country.
• 29:48
  Uh, you get one-day ground proximity with, I think, seven shops if they're all optimally placed.
• 29:52
  Um, and, uh, so think seven Oshkoshs all over the country with one-day ground proximity.
• 29:59
  Um- That's a- an option.
• 30:02
  It's actually very different from the play that is like, we're gonna do every service.
• 30:06
  If you wanna do every service, you really need some concentration.
• 30:10
  That's right.
• 30:10
  Yep.
• 30:10
  Um, if you wanna do...
• 30:12
  If you wanna be in everybody's backyard, you need hyper-efficient, like these things just crank, they're, they're OPEX efficient, uh, we know how to spin these out, we could make a dozen a year, that kind of a thing.
• 30:25
  Um, that's a different model, and there's maybe some overlap where we have like concentrated regional centers that have tons of capacity and capability, and then smaller shops that offer a sub- a popular subset of services everywhere else.
• 30:38
  Um, we'll see where that lands.
• 30:40
  Uh, it might depend a lot on how Cincinnati scales now that we have a regional presence there.
• 30:44
  Um, but we'll see.
• 30:47
  Well, I...
• 30:47
  That- that's one of those things you- you've talked about, uh, and written about a bit, is kind of that the- the future of manufacturing is not necessarily a bunch of tiny mom-and-pop job shops, but rather some level of concentration of more capable and more effective, I guess, operationally and execution-wise.
• 31:06
  So, I, I mean, I, I, we'll have to get into that another time and kind of that model.
• 31:11
  Um, but, you know, as we kind of get to top of the hour here, um, what's, what's next?
• 31:17
  What's the big stuff coming for Oshcut that you could share with us?
• 31:21
  Maybe some things you obviously can't.
• 31:23
  Um, but I guess, what are you most excited about, um, you know, in, in this next chapter as you guys continue to scale and grow?
• 31:31
  So I'm really personally excited about maturing the service.
• 31:34
  Um, what we found is that a lot of our customers trust Oshcut to become this, like, um, critical component of their supply chain.
• 31:44
  Uh, like, we bolt onto their production line, um, and then order parts kind of on demand, uh, and then those parts go in and they make their product.
• 31:52
  Uh, a very high percentage of our customers do that, and when there's an interruption and you're feeding a, a, uh, manufacturing, uh, line, it's actually extremely disruptive if there's ever an error of any kind.
• 32:06
  So what we're really focused on right now is just absolutely nailing that reliability portion even more than we already have, um, so that we can offer guarantees.
• 32:16
  Like, this part will show up- Love it it will be on time, and it will be perfect, and, um, and here's our proof.
• 32:24
  Um, so we're going there, uh, and we've got some pretty, uh, interesting technologies we're developing that will enable that without ramping costs way up.
• 32:34
  Um, uh, something that basically does not exist right now, so, I'm pretty excited about that.
• 32:40
  Exci- exciting stuff ahead.
• 32:42
  I got a, I got a final question, Colin.
• 32:44
  So last, last week with Will Alverson, we developed a new index.
• 32:49
  It's our first national, uh, growth index on Reforge.
• 32:54
  Turns out he's fabricating those hooks on dock locks that hook the, the trailer to the building, and so he talked about how he's just seen explosion in dock lock hooks, which Colin and I immediately interpreted the number of dock doors- You've got the growth divided by the average dock doors.
• 33:11
  We had the number of distribution logistics facilities.
• 33:14
  We're just, uh, you know- Boom we had CapEx number.
• 33:17
  Do you, have you seen something like that?
• 33:19
  Is there just something that, that most people have never heard of, like a dock lock hook, that Oshcut is excelling in and you think,"Wow, this is an indication of where things are going"?
• 33:29
  An obscure market or economic indicator, if you will.
• 33:36
  Um, boy, that's a good question.
• 33:37
  To be honest, that's probably a little outside my pay grade.
• 33:40
  We're, we're a lot like a service center in that we serve hundreds of industries.
• 33:44
  Um, so what we tend to see, like if we, we have a, a report...
• 33:49
  This is, l- like we build a report that says, that categorizes all of our revenue, and there's like, you know, 100 different, uh, rows in this.
• 33:57
  Wow.
• 33:58
  Um, but there's always like a few top industries, um, and it could be aero and it could be automotive and it could be, uh, energy.
• 34:06
  Um, could be like industrial OEM.
• 34:09
  Um, and they kind of move up and down, and one like drops and then the other accelerates.
• 34:15
  But what I use and have used in the past for my modeling is just the, the, uh, uh, manufa- purchasing managers index for manufacturing.
• 34:25
  Um, it, it has co- historically correlated very tightly with the number of new accounts we're onboarding every single month, um, very, very predictably.
• 34:35
  Um, that all broke when we figured out how to, like, mature our marketing function and get the brand out there more.
• 34:41
  Um, all my old models broke, and I just stopped using them.
• 34:43
  But- um, but purchasing managers index is actually a pretty good one, I think.
• 34:48
  Um, just as- All right.
• 34:49
  We'll add that to the basket, Colin, as part of- I, I think we've got a, a new question we have to ask everyone- is what's their kind of, what, what's their view on economic indicator?
• 34:57
  Um- Mine is, mine, just for the record, I, I was the first.
• 35:00
  Mine is switchgear, so Interesting.
• 35:03
  Like, uh, switchgears for, like, backup, uh, generators and, uh- No.
• 35:09
  We're, uh, we, we got started on this a couple months ago, and, um, we're right now, in order to power automation, and all...
• 35:18
  We believe all new factories should be automated.
• 35:20
  So in order to power automation, the switchgear order profile is now competing with the data centers for your, your internal kind of medium voltage switchgear and your sh- and all the step-down functions, motor control centers, all that type of stuff.
• 35:36
  So it's really exploding because, uh, automation is now the base case, which is the greatest thing in my career, that that happened, is now if you're planning a new factory, it's not like,"Oh, we might automate it someday." It's like,"You gotta automate on day one," right?
• 35:53
  So your, your, uh, process electrical demand is dramatically more intense and complex than it was just five years ago when people planned a basic facility and they left the empty transformer pad outside and they didn't really think about the automation.
• 36:08
  Now you've gotta have high-quality power everywhere for the, the process equipment, the robotics, the, uh, you know, automated warehouse, whatever it is.
• 36:17
  Yeah, that could be one.
• 36:18
  I would probably, um...
• 36:20
  This is maybe driving on...
• 36:21
  I've got time, by the way.
• 36:22
  We can keep chatting if you want.
• 36:24
  Um- Uh, absolutely.
• 36:25
  Think in progress.
• 36:27
  The, uh, I would hesitate- Solving the world's problems to use anything as a, uh, future predictor/indicator if AI is influencing it, and that's not because I'm bearish on AI.
• 36:40
  I think AI is in the process of and will continue to change everything about computing and how we interact with, with technology, and it's really exciting.
• 36:50
  But I would be surprised if there's not some kind of, um- correction, uh, on, uh, that, that affects maybe data center build-outs.
• 37:00
  I could be totally wrong about that.
• 37:02
  I mean, who knows?
• 37:03
  Um, but AI is driving...
• 37:05
  Like, we're, we're seeing a ton of demand right now, and I think, and we don't serve data centers per se.
• 37:13
  Um, but data centers displace a lot of, uh, capacity.
• 37:16
  They take up a lot of capacity and displace customers who aren't data centers, for example.
• 37:21
  Like, somebody is making all those server racks and stuff, and they happen in the US, um, for a variety of reasons.
• 37:27
  Uh, so those someones are dedicating capacity to data centers that didn't need to be built 10 years ago.
• 37:33
  Um, so all the other customers are looking around, and I think, I think that is driving our growth right now.
• 37:40
  Um, just as a caution, for anybody who's experiencing a growth boom right now like we are, like we're 100% ahead year over year right now- Woo.
• 37:48
  Um- Congrats.
• 37:50
  Thank you.
• 37:51
  Um, and I, I- Cautiously think that our, our systems are...
• 37:54
  The fact that we can do that without everything breaking, I think says good things about our systems.
• 37:58
  But I, I can't take victory laps because we're benefiting from these macro tailwinds that we don't control.
• 38:04
  And, and I, for, for any listeners who are running, uh, manufacturing operations, I would be very cautious about, um- Over levering to capture this boom, um, because, uh, manufacturing is very cyclical and if you're leveraged and the leverage looks fine at your current operating rate and then suddenly it plummets by 50%, then, then you're in trouble.
• 38:29
  So, you know, optimize with- It's, it's two episodes in a row where- Yeah where we've gotten that message now.
• 38:34
  So it's a, that reinforced from two brilliant men, you know, say- saying the same thing in different ways.
• 38:40
  So I, I think you're spot on.
• 38:42
  Yeah, and if you, if you and Will A- Alverson aren't personal friends, you should be,'cause you're both beyond smart, uh, for Colin and I.
• 38:49
  But Colin, the, the trick is you have to get 25 pods in before you get the heavy hitters at Steel on, and then you learn- Yeah what's really going on.
• 38:56
  But the, just to close off on the AI, uh, I watched one of our good friends, uh, Michael Gayed, who's a, you know, loud voice on X.
• 39:04
  He was on Squawk Box, I think yesterday, and they gave him a long segment, uh, to try to unpack this.
• 39:12
  And he basically said that there's, there's two issues with AI compute.
• 39:18
  Um, number one is, um, that the new models coming out are much more...
• 39:23
  that are open source, are more efficient on tokens, so that's gonna decrease demand overall.
• 39:29
  And then number two, that he believes, um, the blowback that we see, all the discourse about data centers is just gonna lead to regulations.
• 39:37
  And that's just gonna kinda, as it always does, kinda parachute behind it and slow it down.
• 39:42
  And that the funding was gonna get precarious based on, this is my favorite topic I don't understand, is the reverse carry trade.
• 39:52
  I was like, only Michael could take those three things and just kinda mash them into one coherent ball.
• 39:57
  But I do agree, I think it's clearly a boom, right?
• 40:01
  And we know what happens after every boom.
• 40:03
  So, um, I think you guys are amazing that you're in 100 industries, so you're well protected, you know, and you're conservative in your financial expansions.
• 40:11
  That's, that's absolutely brilliant.
• 40:13
  Oh, f- Caleb, I had a final question since you had a few extra minutes.
• 40:16
  Yeah.
• 40:16
  Can you talk to us about your early stage funding, how you put the company together?
• 40:22
  Uh, you know, we're, uh...
• 40:24
  It was, it's bootstrapped, so, um- Yes so I know some other, uh, people in the industry have said,"Oh, we're bootstrapped," but it works out that they raised$6 million, which is a funny way to say,"I'm bootstrapped." Um- They bootstrapped their capital raising.
• 40:40
  I suppose.
• 40:42
  I mean, when you're an early stage company, 6 million of seed makes a difference, especially when you're, um, uh, you know, especially when you're, when you're- Yeah, when you couldn't get a loan for 6 million and then somebody hands you 6 million, yeah, it, it makes a huge difference.
• 40:55
  Yeah, so when I say we're bootstrapped, we're, we're really bootstrapped.
• 40:57
  So I, I took a HELOC on my house, which was paid off.
• 41:00
  My man.
• 41:01
  Wow.
• 41:01
  Previous company, uh, had pay, was able to pay off my house.
• 41:06
  Um, and, uh, and that's, that's the primary source of funding for our business.
• 41:12
  Also borrowed, um, not borrowed, but, uh, took about$180,000 from my dad out of his, out of his 401.
• 41:19
  Uh, was lucky that he had that available, um, so he's a part owner in the business because he did that.
• 41:25
  Um, but that was it.
• 41:26
  After that we, uh, um, it's been, uh, reinvesting cashflow to grow the business, so.
• 41:33
  Yeah.
• 41:33
  Well, you would've had- That's awesome a tough time at Reindustrialize, Caleb, because- Oh, yeah?
• 41:39
  a couple of us there that when the V- There's, it's about a three to one ratio VC to founder there.
• 41:46
  Oh, wow.
• 41:46
  So the VCs are kind of walking around and they're looking, they're,"Hey, you got any founders?
• 41:49
  You got any founders?
• 41:50
  We're looking for founders.
• 41:51
  Um, try under the sofa.
• 41:53
  There might be a founder under there." And, uh, and our, our shtick in the happy hours was,"Oh, hey, nice to meet you.
• 42:02
  We're bootstrapped and EBITDA positive." And then we'd just watch as they scurry away.
• 42:07
  And they disappear.
• 42:08
  Oh, really?
• 42:10
  Yeah, believe it or not, they, they, the majority of them, I, I, I get to say this'cause I own a venture capital company, so that's what gives me the license to dunk.
• 42:18
  But they don't wanna talk to you unless you're just burning cash to, for the sake of growth with no profit.
• 42:26
  And so I can tell you're an engineer by trade, I'm an engineer by trade.
• 42:29
  It does not compute to just run a business at negative forever and just assume the next funding round is, is coming, you know, to me.
• 42:37
  Yeah.
• 42:38
  Well, I have opinions on that.
• 42:39
  Um, we can- Yeah.
• 42:39
  We can stop, we can wrap up anytime if you want.
• 42:42
  Uh, but I love this stuff, so.
• 42:43
  Oh, let, let's hear'em.
• 42:44
  Let's hear'em.
• 42:45
  Get yourself in trouble.
• 42:46
  Let's go.
• 42:47
  See what platform we can get banned on next.
• 42:50
  Uh, are you guys familiar with...
• 42:52
  Oh, geez.
• 42:53
  Are you guys familiar with, uh, Hamilton Helmer, uh, Seven Powers, that book?
• 42:58
  Yep.
• 43:00
  Sounds like college.
• 43:00
  Yeah somewhere.
• 43:02
  Um, look, uh, I think that, um, bur- burning cash, like burning in the red to capture market share makes sense if you're creating a business that has power.
• 43:15
  And by power, in the Helmer sense, it's, um, you have the ability to earn differentiated margin, differentiated returns compared to competitors, and there's a moat.
• 43:26
  You have those two things.
• 43:27
  If you can create that, it makes total sense to spend as much money as needed, um, to create the power, and those powers can be scale economies.
• 43:38
  Um, it can be like, um- Network effects.
• 43:42
  Yeah, yeah.
• 43:42
  Yeah.
• 43:42
  There's, there's a bunch of them.
• 43:43
  There's some- Switching costs.
• 43:44
  Yeah.
• 43:44
  Facebook, Google, SpaceX, right?
• 43:46
  Those would be examples.
• 43:48
  Yeah, yeah.
• 43:49
  Like Faceba- Facebook's a classic network economy.
• 43:51
  Like it, it benefits the user, you, if there are other users on it, and it doesn't benefit you to go out and do another thing.
• 43:58
  LinkedIn is another one, um, uh, in a different space.
• 44:03
  Um, you know, you, you look at ridesharing, uh, platforms like Uber a- and Lyft.
• 44:09
  There basically can't be another Uber and Lyft, uh, that work in that way, like because the networks already exist and it's a, it's a two-sided network.
• 44:18
  It's the network of drivers and it's the network of, uh, users, um, uh, you know, riders.
• 44:24
  And, uh, so yeah, it made sense for them to hemorrhage money for as long as it took, and now they, now they can be cash flow positive because they have those network economies that protect the business.
• 44:34
  Um, so, uh, when you look at manufacturing, uh, I think there exists powers that can actually be worth spending to acquire.
• 44:45
  Um, but I think it's very hard.
• 44:47
  Um, and, uh, anyway, so when I see like a company like Xometry that's, that's been in the red for the last 13 years but is approaching a billion dollars in revenue this year, um, I think their thesis is that they're creating a two-sided network, and once they have it, now they can be cash flow positive.
• 45:04
  My thesis has been on that, on the Xometry front, is that, uh, direct manufacturing- Aren't they spending like 30% of their, their cost is just to acquire customers, though?
• 45:14
  Yeah, I ran the numbers.
• 45:15
  Isn't that like that?
• 45:15
  They, they actually sh- say,"Here's their, um, uh, their, uh, h- here's how much we're spending on sales and marketing." And, and then they say,"Here's our active customer count." I think they were spending, last I checked, like$15,000 per every net new account, uh, for customer accounts and, and obviously that's not- Um, in general going to be sustainable.
• 45:39
  No.
• 45:40
  But they've got this fortress balance sheet, they have so much cash.
• 45:43
  Um, and maybe they're right.
• 45:45
  Maybe once they've created the network and soaked up all the demand, it's just th- n- now they can scale and monetize.
• 45:52
  Uh, but I think that, like, a subset of direct manufacturers like Oshcut are just going to be better priced and more performant as a rule.
• 46:01
  Um, uh, supplier differentiation is the Achilles heel of, uh, broker models like Xometry.
• 46:08
  Um, so anyway, we'll, we'll see.
• 46:11
  We'll see how it happens and how it goes together, but, uh, very interesting.
• 46:15
  Well, I have a bias towards founders that own the majority of the company equity.
• 46:19
  That's one.
• 46:20
  And a, and a, uh, go ahead and, and cube that bias towards founders that took a HELOC, so Oh, man.
• 46:29
  There might be a few on this podcast right now.
• 46:33
  You have to really believe in something to, to put your house on the line- That's right to fund it.
• 46:40
  Well, clearly you guys are doing something right.
• 46:43
  I mean, Oshcut has, you know...
• 46:45
  O- obviously I'm a fan, I love, I mean, just the entire customer experience, the simulations, being able to, you know, see the pricing update live by, you know, changing quantities.
• 46:55
  I mean, I- you guys have done an amazing job.
• 46:57
  I'm obviously a huge fan.
• 46:58
  Um, and you know, appreciate all the work that you're doing.
• 47:03
  For those of you who perhaps, and somehow they've been living under a rock and have not heard of you before, um, how can they find out more?
• 47:11
  Where can they, you know, get in touch, learn about Oshcut, all that good stuff?
• 47:16
  Yeah, oshcut.com.
• 47:18
  We've got a page there.
• 47:19
  Our app is accessible from there, where you can get instant prices.
• 47:23
  Um, if you have any questions, you can email our support team or give us a phone call.
• 47:26
  Um, you can, you can actually talk to a person at Oshcut.
• 47:30
  Um, so, um, yeah, happy to, happy to help.
• 47:35
  And, uh, yeah, connect with me on X and I'm, I'm open to, uh, to DMs, so- And if you follow him on X, you will see some very cool projects.
• 47:45
  You can even harass him about trying to get Star Fox hooked up to- you know, the, the motion rig simulator, which I'm still holding out on that, and, you know, a life-size, uh, you know, was it Lawless Spider from Wild Wild West featuring Will Smith?
• 48:02
  Uh, that's for another day, obviously.
• 48:04
  Um, so anyways, Caleb, thank you so much for joining us.
• 48:09
  This has been a great, great conversation.
• 48:12
  You have given us plenty to think about, um, and it is really exciting to see all the moves you guys are ma- making.
• 48:19
  Best of luck to you, um, and the rest of the Oshkot team.
• 48:23
  Uh, keep up the great work, plain and simple.
• 48:26
  Yeah, thanks for the great conversation.
• 48:27
  And of course.
• 48:29
  And to all of you listening, thank you for tuning in to the Reforge podcast, whether we're talking robotics on the factory floor, policy that actually moves the needle, or the next generation of American makers, our goal is always the same, to prove that re-industrializing America isn't just possible, it's happening right now, and it's pivotal.
• 48:44
  If you've enjoyed this episode, make sure you subscribe and follow@thereforgepod everywhere, and check out thereforgepodcast.com for more, including our seven steps to re-industrialization.
• 48:55
  Until next time, keep building.
• 49:02
  Thanks for tuning in to The Reforged Podcast.
• 49:05
  Make sure to like, subscribe and follow us on all of our social media@thereforgedpod.
• 49:11
  We will see you next time Welcome to the Reforge podcast.
• 49:25
  I'm Matthew Chang, and we're here to explore all things AI, automation, and the American worker, proving that the American reindustrialization is possible, pivotal, and pressing.
• 49:37
  Thanks for joining us today.
• 49:38
  Let's get into the episode.
• 49:42
  Welcome back everyone to episode 28 of the Reforged Podcast.
• 49:47
  We are joined today by a gentleman that I'm sure most of you will recognize, Mr.
• 49:53
  Caleb Chamberlain of Oshkut.
• 49:56
  Am I saying it right?
• 49:57
  Oshkut or is it O-S-H Kut?
• 49:59
  Uh, Oshkut is right.
• 50:00
  Okay, good.
• 50:01
  I've, I've...
• 50:02
  I, I sometimes question myself.
• 50:04
  Uh, he is CEO of Oshkut, has an actually a very interesting background, uh, that might surprise some of you.
• 50:10
  Um, so just wanna welcome him to the show.
• 50:12
  Caleb, thank you so much for joining us.
• 50:15
  Yeah, happy to be here.
• 50:16
  So o- obviously, you know, I didn't do you justice.
• 50:19
  Give us a little bit of your background.
• 50:21
  Tell us a bit more about who you are, you know, and, and kind of what led to, what led to Oshkut?
• 50:28
  Yeah.
• 50:28
  So originally I, uh, was an electrical engineer, have a master's in electrical engineering with a control theory, uh, emphasis.
• 50:35
  Um, I did, uh, actually, uh, a lot of research around drones.
• 50:41
  Uh, we didn't call them drones back then, we called them quadrotors but, uh- or multirotors.
• 50:45
  Uh, but now they're drones.
• 50:47
  Um, but I did system identification, state estimation, and control of, uh, unmanned aerial vehicles.
• 50:53
  That was my master's thesis.
• 50:55
  Um, I went from there- Seems pretty relevant right now, actually.
• 50:58
  Right?
• 50:58
  Yeah.
• 50:59
  I, I opined that it's kinda funny that I'm actually not doing anything in that space now.
• 51:04
  Um, but my, my first company I started as a grad student and we made, uh, inertial sensors, think, uh, accelerometers, rate gyros, magnetic sensors, uh, GPS, and we did, uh, little micro sensors that, that, that provide state.
• 51:19
  It was pretty novel back then, uh, less novel now.
• 51:22
  Uh, and we also did some Air Force SBIR contracting.
• 51:25
  We did a covert precision aerial delivery system where we had, um, uh, built an autopilot that steered a silent, uh, winged vehicle, uh, over a combat zone to deliver payloads, uh, autonomously.
• 51:40
  Um, I, uh, I had a stint doing some medical device stuff with Owlet.
• 51:45
  We made, um, uh, pulse oximeters that went on a, on a infant's foot that they could sleep with comfortably and provide, uh, blood oxygen readings.
• 51:55
  Th- those little sock ones?
• 51:57
  Uh-huh.
• 51:57
  Yeah.
• 51:57
  Yeah, I remember that.
• 51:58
  I was looking at those when, you know, my, my youngest, uh, was born.
• 52:01
  I, I geek out on that stuff.
• 52:03
  But wow, that's pretty cool.
• 52:05
  Yeah, yeah.
• 52:05
  The, it's a load off your mind to know that your, that your newborn's safe.
• 52:09
  So, uh, uh, I was se- a senior electrical engineer on their V3 Smart Sock project.
• 52:15
  Um, and I also spent some time at a synthetic aperture radar company called IMSAR.
• 52:20
  Um, and we did, um...
• 52:23
  We- I, I worked, actually worked on sort of a side project where we did, um, a time of flight, uh, based localization of active RFID tags.
• 52:32
  Um, in any event, uh, eventually, um, my brother and I, uh, who started Oshkot, we started building things together just for fun, and one of those projects was a, this VR motion rig, virtual reality motion rig.
• 52:48
  Nice.
• 52:48
  So you're, like, flying an aircraft, uh, or a spaceship or whatever, uh, and, uh, wearing a VR headset.
• 52:56
  And the problem with, with VR is that if you see motion that you don't experience it makes you sick, um, it makes you really nauseous.
• 53:03
  It's not fun.
• 53:04
  Um, so we thought, well, what if the, the chair you're sitting in moves in a way that accurately reflects the motion you're seeing?
• 53:12
  Uh, and so we built this thing.
• 53:14
  But we just found that it was really, really hard to buy fabricated parts.
• 53:18
  We did a lot of tube, we did a lot of sheet metal, and it was just like you would spin up a design in a couple of days and then spend, you know, three weeks to a month to get it made.
• 53:27
  And, and we thought,"Hey, there's gotta be a better way to do this," and that led us to start OSHCut.
• 53:32
  So here we are eight years later.
• 53:34
  Well, I think Colin, after he g- after the CNC machine that he's trying to buy at Costco, I think his second CapEx purchase is gonna be for the VR rig that you build.
• 53:46
  I want one so bad.
• 53:47
  We're building another vision.
• 53:48
  It took us, it took, it took us seven years to finally build that machine, um, using OSHCut, uh, parts.
• 53:55
  Um, but we're working on another version now that's, that's going to be much better, so, um, we'll be- Well, I, I saw the video you posted of, like, the w- with the RC car doing- Yeah like, the, the off ro- Like, that is just was too cool.
• 54:10
  I mean, I'm, I'm sorry, that stuff just, I, I geek out on it, but that was just too fun.
• 54:14
  Yeah.
• 54:14
  One of our, uh, one of our senior software engineers also does a lot of, uh, like RC airplanes and FPV stuff, and we're gonna hook telemetry from an FPV drone up to that chair, so you'll be wearing it and, like, flying around.
• 54:28
  We'll see how it is.
• 54:29
  The mo- the machine can't move as fast as, like, a, uh, an RC, a little RC drone.
• 54:34
  Um, so there might be some kinda smoothing there that makes it not great, but it's easy to try, so.
• 54:39
  Colin, I'm feeling a lot like OSHCut as we know it is just generation one of the company, and there's all this sensing and computation that's gonna be going into the fabricated metal parts, you know, in the next...
• 54:52
  That's just what I'm sensing, so.
• 54:55
  I, I mean, g- I, that might be a good little segue for some of the, you know, recent exciting news, you know, out, out of OSHCut.
• 55:02
  Obviously, you know, w- we talked about the second facility, um, that, that y'all recently opened up.
• 55:08
  But you also opened up, or, or announced recently, an R&D center, right?
• 55:13
  Uh, that's right.
• 55:14
  Uh, we're opening up that space in September.
• 55:17
  Uh, it's not a massive space.
• 55:18
  It's 20,000 square feet, but it's dedicated to, um, supporting projects like the motion rig and piloting new services.
• 55:28
  Um, so the rationale will be, hey, we'll...
• 55:30
  Anytime we need something...
• 55:31
  The way we look at OSHCut and, um, building and design is that it's kind of like industrial Legos.
• 55:37
  Um, so we wanna set it up so that, um, you can build pretty much anything just by buying stuff online, whether it's a commercial off-the-shelf product, like something you can buy on McMaster-Carr, or whether it's custom metal you can get from OSHCut.
• 55:53
  And we wanna do it all the OSHCut way, meaning, meaning, uh, manufacturability is exposed from first principles right in your web browser.
• 56:01
  So by the time you get around to placing an order, you know it's going to arrive on schedule and everything's gonna work just, just fine.
• 56:09
  Um, y- it...
• 56:10
  I mean, it's the most disruptive thing ever when you place an order and, um, somebody reaches out two days later and says,"Actually, we can't make this." So we wanna get as far away from that as possible and just say,"Look, design, build your custom parts.
• 56:22
  If it's made to spec, you'll get it, and, um, it'll be easier than ever." It ch- kind of changes the way you can design and make things.
• 56:31
  And so, so on the R&D space, it's like, okay, well, if we can't...
• 56:35
  If that experience doesn't exist and we need something, um, then it's a candidate for a new service.
• 56:42
  We get the equipment, put it in the R&D space, uh, figure it out, learn it, you know, launch it behind a feature flag for select customers.
• 56:49
  Um, and then when it's ready for prime time, we just go all in and launch it as a new service.
• 56:54
  Well, and one of those new services that y'all added not too long ago was obviously the inclusion of tube.
• 57:00
  Um- Yeah you know, a- adding tube options, which is obviously huge, um, just because that's one of those areas where you don't see a lot of options out there.
• 57:10
  You don't have a lot of the instant quote and such a supply.
• 57:13
  What was it that made you guys choose to go with tube versus, I don't know, any other service?
• 57:19
  Was it just purely demand?
• 57:21
  I mean, I see you guys do a lot of off-road stuff too as well.
• 57:24
  Um, it wasn't demand per se.
• 57:27
  It was mostly, it was a service that we wanted and needed.
• 57:31
  Um, so, um, you know, uh, it's probably could be argued that we should have done a bunch more market research and, uh, the truth is the tube is pretty niche.
• 57:40
  Um, we just launched tube bending as well, uh, CNC tube bending on a draw bender.
• 57:45
  Um, and that's very, very niche, and I don't think a lot of people design for it.
• 57:50
  But man, once you realize that that's a tool in your tube tool bag, it changes how you design.
• 57:55
  Um- Mm.
• 57:56
  So that's kind of...
• 57:56
  We have a long view on that.
• 57:58
  It's like, you know, we launched tube in 2024.
• 58:01
  Um, it's still a very, very small percentage of our business, and it's expensive.
• 58:05
  You know, a, a state-of-the-art tube laser is$2 million.
• 58:09
  Um, a tube bender could range from, you know, quarter million to half a million to a million dollars, depending on the size, and it's a very small revenue center for us.
• 58:19
  Um, so I claim that as kind of a pet project, um, uh, just knowing that, hey, in 10 years, a lot more people are gonna be using this because we've made it accessible.
• 58:30
  Well, I like the idea of your select customers filling up your R&D space with their projects, right?
• 58:36
  So they come to you with a thing, and they say,"It would be amazing if the tube bent itself, if the tube cut itself, uh, if the tube welded itself." And then that gives you the...
• 58:45
  You know, you can make a decision obviously, but, uh, the, it fills up your idea space without the folks at OSH Cut having to think of everything the industry might buy, build it and they will come.
• 58:57
  It's like with every project, you have customer number one, at least, in the bag, and you can kind of learn with them.
• 59:04
  Do they, do your best customers ca- come to you with ideas and then, and then talk it through with you?
• 59:10
  Is that how this comes about?
• 59:13
  Yep, they do.
• 59:14
  And, uh, you know, sometimes customer number one is just us.
• 59:19
  I actually think that's important.
• 59:20
  Um, I think that, um, so much of the, the like, the, the thing that makes big companies bad is when you get like, uh, not to knock MBAs, but like professional MBAs and finance guys, uh- in the executive seats, and they're manipulating spreadsheets, and they're, um, they're financializing everything.
• 59:43
  Uh, you know, you've got a, a business that produces cash, and they're all about like,"Well, how, how do we invest this cash?" And, you know, putting it in equities or whatever.
• 59:51
  We're all about,"Look, we're, we're here to build, and, um, we want to make this as good as possible for people like us." And if that makes us a billion-dollar company in the future, awesome.
• 1:00:01
  If it makes us$100 million company in a couple years, uh, awesome.
• 1:00:06
  Um, if we stay at, at about, you know, 60 million a year, um, that's awesome, too.
• 1:00:11
  Uh, you know, it actually doesn't matter to me as much as, as, uh, creating this awesome way to build.
• 1:00:19
  Um, and the revenue, I think, will fall naturally from that, depending on how it works, so.
• 1:00:25
  So a question, after you started as, as this kind of online, you know, designed order company, deliver on time, AI happened, what, since you built the company.
• 1:00:36
  How much is the vibe coding and, and being able to use agentic AI to streamline the workflow for customers, how much has that been a factor?
• 1:00:45
  So for us, it's mostly about software development, and we do use it very heavily actually.
• 1:00:50
  But it's transformed the way we develop.
• 1:00:52
  I mean, we've got, uh, I think eight full-time developers, um, and, um, and we're growing that team.
• 1:01:00
  We wanna double that team within the next six to 12 months.
• 1:01:03
  Um, but we have...
• 1:01:05
  We, we embed the AI-driven workflows right into how we, how we test, how we deploy, how we write code, and it does accelerate a lot.
• 1:01:15
  Uh, it's very significant actually.
• 1:01:17
  Awesome.
• 1:01:18
  Well, one of the cool things that you guys recently added was kind of that, uh, that for your excess capacity, where people, if they- they've got the flexibility on it, uh, on when they get something, you know, they can get parts and pieces made for much cheaper just because, you know, it's an opportunity for you guys to fill your capacity.
• 1:01:38
  In kind of that arena, one of the things that I've loved, uh, seeing you talk about a little bit is your use of the Monte Carlo sim- simulation models.
• 1:01:48
  You don't have a lot of people on X or general discourse who talk about some of that more technical analysis, quote unquote, if you will, to look at all the different possibilities of here's, you know, the likelihood of us being able to hit all of our work within this time period.
• 1:02:06
  Here...
• 1:02:07
  Rather than saying,"Our capacity is X," you know, y- you guys are taking a much more probabilistic approach to it, and I, I think it's really, really great to see and really cool, but not a lot of people talk about it.
• 1:02:19
  Um, obviously, your engineering background, um, probably contributes a bit to some of that.
• 1:02:25
  So maybe can you speak a bit to how you guys approach the capacity piece?
• 1:02:31
  Because you did just add a whole second facility as well, that obviously increases that, too.
• 1:02:37
  Yeah, there's a couple pieces to that.
• 1:02:39
  So we're really focused right now, on the capacity front, we're really focused right now on, uh, resilience, um, and proximity.
• 1:02:47
  So, um, it's really, really important that as we scale, that our customers have a great experience and they know that OSH 切 is the quality option, and we're leaning heavily into that.
• 1:02:57
  Um- And, uh, you know, a big piece of that is, look, y- we just have to optimize around excess capacity because you can't predict the future.
• 1:03:04
  So we need to kind of gear our CapEx plan and financial modeling around always having those extra machines and people.
• 1:03:12
  Uh, and it, it is expensive, but I think in the end it, it really pays for itself because, um, you know, equipment, uh, depreciation actually works out to be a fairly small percentage of our cost of goods sold.
• 1:03:23
  So we have a little bit of wiggle room there.
• 1:03:26
  Um, meanwhile, I think that, uh, uh, you know, when, when people understand that there's a reliable option, I think they'll just tend to kinda go there.
• 1:03:34
  You get, you get people in rebuy mode, and it, it's an investment in the future.
• 1:03:38
  Um, what was the second piece of that question?
• 1:03:41
  So we talked about capacity.
• 1:03:42
  Uh- Oh, Monte Carlo.
• 1:03:43
  Um, Mon- Yeah, I mean, I, I geek out on some of that just'cause I think it's great from...
• 1:03:48
  You know, in a, in a high mix, low volume shop, capacity can be a little bit difficult to accurately represent as a true single number because that mix of work.
• 1:03:57
  So, you know, with Monte Carlo, it, it, it's would seem useful since you can model so many different possible combinations of incoming work rather than focusing on, you know, some utilization number that is kind of not reflective of reality perhaps.
• 1:04:15
  Yeah.
• 1:04:15
  Monte Carlo can allow you to predict bottlenecks.
• 1:04:18
  It can help you understand, you know...
• 1:04:21
  We actually have code that allows us to simulate incoming orders sta- that are statistically indistinguishable from real orders and that use real parts.
• 1:04:29
  Uh, so effectively we say,"Hey, okay, you know, here's the, here's the distribution on arrivals.
• 1:04:34
  Um, here's the distribution on, uh, parts per order, on likely materials, on quantities." Um, and then you can just say,"Okay, starting from today, um, give me a bunch of synthetic orders that are believable based on our current volume.
• 1:04:50
  Okay, now do the same thing with 20% more.
• 1:04:52
  Now do the same thing with 100% more.
• 1:04:54
  What breaks?
• 1:04:55
  Where does the bottleneck arrive?" And then you can proactively plan and say,"This is the machine I need to handle this problem." Um- Love it the, the way the simulation works, just if people aren't familiar with Monte Carlo, um, it's basically you take something that's very complicated and impossible to simulate from first principles, uh, accurately, and you say,"Okay, well, we're gonna model it statistically." So, um, the way that works in practice is you've got, like, how long is it gonna take to make this part?
• 1:05:23
  Um, well, a person can say,"You know, I, I'm not sure, but I think it's gonna take, you know, five to 10 minutes, um, to weld that together." It's like, okay, well, we could either, either produce a distribution on that style of part analytically and build a statistical model, or we can say,"All right, five to 10 minutes, so let's put a uniform distribution on that." And then what happens is you, you build a model, um, that reflects every operation in production, uh, and then you simulate it end to end And say,"Okay, well this time it took five minutes, this time it took 10 minutes." You simulate 1,000 times, and then at the end of the day you have an outcome that says,"Hey, this part shipped on time 85% of the time." Um, what can we do?
• 1:06:06
  How can we change the scheduling heuristics to nudge that number up, you know, to 90% or 95 or, or 100%?
• 1:06:14
  Um, there's always, like, late scrap and stuff, and you model that too.
• 1:06:18
  Um, so there's, uh, there's the statis- statistical part, um, and then there's, um, what I call, you know, part selection heuristics, and that is what do I work on next?
• 1:06:30
  And that question is asked 1,000 times at every manufacturing operation every day.
• 1:06:34
  So that's an impossible thing to, to answer optimally actually.
• 1:06:39
  Yeah.
• 1:06:40
  It would take more than the age of the universe to simulate all the possible outcomes.
• 1:06:43
  Yeah.
• 1:06:43
  Uh, so what you do is you say,"Well, I'm gonna build a sim- a s- a simple rule of thumb," it's called a heuristic,"and I'm gonna run that, run the simulation on that, on projected incoming orders and existing.
• 1:06:55
  Um, what's my on time shipment percentage?" And then you design a new heuristic and you say,"All right, well, what if I do the same...
• 1:07:03
  What if I run the same simulation but I change how I choose what parts I work on next?" You run that and it's like,"Hey, I just got 5% better on time shipment with my existing capacity." Um, you can use it for labor allocation, like where should people be working and how?
• 1:07:17
  Um, and there's a lot of nuts and bolts that actually make this work, but, um, but it's pretty cool technology actually.
• 1:07:25
  I, yeah, I mean, I, I think one of the things that's really interesting is that a lot of manufacturing management approaches are based on averages.
• 1:07:34
  Um, you know, looking at your average cycle time, lead time, scrap, all of those pieces of it.
• 1:07:39
  But averages, you know, ha- have a tendency to hide some of that tail risk that causes issues down the line, but where Monte Carlo focuses more on that probability piece of it.
• 1:07:53
  What is the, what is the probability that I can achieve X, Y, or Z?
• 1:07:58
  Has using, or rather, does using the Monte Carlo approach, running those simulations, change the way that you think about managing the business, kind of ma- making some of those decisions and managing, um, so- some of the different levers you have to kind of pull on?
• 1:08:18
  Yeah, I mean, I can give you an example that's actually quite intuitive.
• 1:08:22
  Um, if you're using just averages, um, then effectively you're saying, all right, well, it's gonna take this long on average to make this part, but, but there's...
• 1:08:32
  Let's, let's consider scrap rate, for example.
• 1:08:34
  Um, let's say usually you don't scrap a part, but there's kind of, uh, this blip at the end of the probability distribution that says, you know, they get scrapped this percentage of the time.
• 1:08:45
  So Monte Carlo will sample that periodically and say,"This got scrapped," and okay, it gets scrapped, now it has to be re-released into production.
• 1:08:53
  Um, the simulation captures that.
• 1:08:56
  Um, and so if, if you're waiting till the very last minute to release something to production, Monte Carlo notices that, you know, maybe 2% of the time there's a scrap problem, and that makes the order ship late, and that affects how you need to schedule.
• 1:09:09
  Says,"Well, to account for this likelihood of scrap, I actually have to release early enough that I can push this through production twice and still, um, get it on time." Um, so I, I will say that it's debatable whether all of this is, is what everybody should do.
• 1:09:26
  I mean, it is really cool and the engineer in me gets really excited about it.
• 1:09:30
  Um, but what I just described is also very intuitive and something that you could just kind of design a process around without having to build and maintain a big Monte Carlo simulator.
• 1:09:42
  Um, but it is still- But where's the fun in that if you don't get to- There's the fun factor, for sure.
• 1:09:49
  Um, but you know, for cap- capacity planning, bottleneck prediction, um- W-, you know, being proactive about...
• 1:09:57
  This is something we're working on right now, is being proactive about, um, notifications.
• 1:10:02
  Like, if something's likely to ship late, it's very good to reach out to the customer ASAP.
• 1:10:08
  Um, but when you've got, you know, 5,000 orders in queue and, you know, 30 th- 32,000 SKUs and it's all in the production mix, and the customer's like,"Is it gonna ship on time?" And you look at that and you're like,"Oh, man, I, I don't know.
• 1:10:21
  Your guess is as good as mine, honestly.
• 1:10:23
  I have no clue.
• 1:10:24
  I, I know that we're busy." Um- but the Monte Carlo actually puts a number on that and says,"Yeah, 50% of the time this ships late, so, um, let's see what we can do about that." But...
• 1:10:35
  That's crazy.
• 1:10:36
  I mean, I...
• 1:10:37
  The way I look at it is all the, the statistical analysis and then just kind of being fanatical about your equipment selection and labor allocation, to me that all just builds to customer service.
• 1:10:48
  So it's like it, it plays into...
• 1:10:50
  I like the way you framed it, Oshkosh, as the, it's kind of the premium solution, it's the high quality solution.
• 1:10:55
  Not saying it's the most expensive, s- you probably beat, win on price quite a bit, but it is, it is predictable.
• 1:11:03
  And I'm, I'm the tail end, right?
• 1:11:05
  So I'm aggregating, in the factory business, a lot of supply chains and when we have risk it sure would be nice to know about it.
• 1:11:14
  We can make alternative plan and my...
• 1:11:16
  Everything I'm doing is going in the field for a factory.
• 1:11:19
  It would be nice to know, because on the, on the receiving end, we're planning labor and we're planning specialty, uh, you know, commissioning agents and engineers and pro- probably controls and software all to come together at one time.
• 1:11:33
  If we have something that's gonna hang it up, I probably wanna let the folks that I'm traveling in know about it so we can make other arrangements and shift our focus to maybe another machine center, for example.
• 1:11:44
  Um, so I love that and I wish everybody did it.
• 1:11:47
  You should Well, we don't quite do it yet.
• 1:11:49
  I...
• 1:11:50
  Right now I think in general, if there's a problem, people find out about it on the day it's due, and I don't like that at all.
• 1:11:57
  So that, that's an active area, um, to take those simulation results and turn them into actionable insights the customer can get early.
• 1:12:06
  Um- Um, yeah.
• 1:12:09
  The challenge is it's gotta be pretty accurate.
• 1:12:12
  You're only as good as your models, and the last thing you wanna do is send emails out to dozens of customers and say,"Hey, this might ship late," and then ship it on time, or, you know, or the reverse, right?
• 1:12:22
  So we've gotta make sure everything's just dialed in, um- Yeah and accurate before we close the loop like that.
• 1:12:29
  Yeah, that's amazing.
• 1:12:31
  Absolutely.
• 1:12:32
  Well, so, you know, I, I can geek out on this stuff for a bit l- for as long as, you know, we could go- Yeah but we'll, we'll shift a little bit to it.
• 1:12:39
  Um, obviously big news with expanding to Cincinnati.
• 1:12:43
  Um, I would just like to say for the record, I did guess correctly that it was Cincinnati.
• 1:12:47
  I'm gonna pat myself- on the back there.
• 1:12:49
  Um, I don't know why, but whatever.
• 1:12:52
  Um, no, talk to us about, you know, that selection process.
• 1:12:57
  You know, we- we've talked a bit, um, on previous episodes about what it takes to kind of get a new factory going.
• 1:13:03
  You, you guys are getting a new facility, you know, expanding.
• 1:13:06
  You're in Utah, adding Cincinnati.
• 1:13:09
  Ta- tell us about that, the drive to it, and kind of what that's been like.
• 1:13:14
  So it's actually gone way smoother than I expected, which isn't something you might say when opening a new factory.
• 1:13:21
  Um, when we moved in 2024, we moved from a, like a 20,000 square foot facility into 55,000 square feet, and we moved while keeping production simultaneously running.
• 1:13:34
  Ooh.
• 1:13:34
  So we would, like- Wow shut down one machine and move it, bring it up at the new site, and then we're doing production in two spaces and...
• 1:13:41
  Anyway, that was a real challenge.
• 1:13:43
  Do you have a lot of WIP from- Yeah site to site B?
• 1:13:47
  Yeah, we tried to avoid it as much as possible, but it happened, and it was very disruptive.
• 1:13:51
  It was, it was very stressful.
• 1:13:54
  If you can avoid it, I would recommend, uh, avoiding it.
• 1:13:57
  But anyway, we wrote all the software to support that.
• 1:14:01
  Um, multi-facility, you know, transfers, all that kind of stuff, tracking inventory per facility.
• 1:14:08
  Um, so when we added Cincinnati as a site, um, it just kind of flowed, and we had to tweak software a little bit, but for the most part it just kinda happened.
• 1:14:16
  Uh, meanwhile, all our production operations are very, um, software managed and software driven, so, um, it's, it's gone very smoothly.
• 1:14:26
  Um, I actually think, uh, my operations team is gonna maybe kill me for saying this, but I think we could probably open, um, two or three factories a year, um, if we had a team dedicated to it, and it wouldn't be, uh, a big deal operationally.
• 1:14:40
  That's good.
• 1:14:41
  Uh, the talent portion would matter.
• 1:14:42
  We'd need good GMs and shift leads in every facility.
• 1:14:46
  That's probably the hardest part.
• 1:14:48
  Um, but we could, we could launch dozens of factories, um, and it would just flow.
• 1:14:53
  Yeah.
• 1:14:54
  How about on the state side?
• 1:14:55
  Did you use the, like the, um, Ohio Economic Development?
• 1:15:01
  JobsOhio or whatever?
• 1:15:01
  Is the drive- Yes, we're actually, we're actually...
• 1:15:04
  We're five minutes from Cincinnati, but we're actually just south of the border.
• 1:15:07
  Cincinnati's right on the border with Kentucky.
• 1:15:08
  So technically Kentucky, yeah.
• 1:15:10
  Yeah.
• 1:15:10
  Oh, does it cover- We're in Kentucky, but we're, we're right next to the Cincinnati Airport.
• 1:15:14
  Yep.
• 1:15:15
  Nice.
• 1:15:16
  Yeah, there's a- So, so- there's a ton of manufacturers in that region.
• 1:15:19
  I mean, it sounds like you'll have to really start serving the aerospace community a lot,'cause I know they make just tons of, uh, tier two aerospace parts in that region.
• 1:15:30
  Uh, we already do, uh, a lot of, uh, aero and, uh, uh, am- among other things.
• 1:15:37
  Um- Yeah, it's, it's interesting.
• 1:15:40
  We've actually kind of hypothesized that the best model for Oshkot going forward is to find, um, areas where there are as many shops as possible, like as many traditional shops, and drop shops there.
• 1:15:53
  Um, the, the presence of traditional shops signal plenty of demand, and our model is to capture the long tail of work that those shops don't want.
• 1:16:03
  Uh, my thesis is that, um, there's plenty of that long tail demand even in these heavy manufacturing areas, and perhaps especially in these heavy manufacturing areas, so, uh, that we can scale the best and the fastest by, by dropping shops where there's the most manufacturing presence already.
• 1:16:22
  I mean, it, it, it makes sense.
• 1:16:24
  It's one of those things where you hear somebody say it and it's like, well, duh, obviously, if you're, you're going for that concentration of it, like there's...
• 1:16:30
  Somebody's proven it a bunch of times with all those.
• 1:16:33
  Um, but I, I don't think I've heard anybody a- actually say it quite that plainly of we're looking for that concentration because it's clear that that's established.
• 1:16:42
  So that's, that's certainly interesting.
• 1:16:44
  But, um, you know, w- was that really just that, that...
• 1:16:48
  Was it as simple as that, saying,"Hey, that's the concentration, we'll go there," or was it kind of strategic where you guys...
• 1:16:54
  You know, being in Utah, you've got great proximity to kind of that western half of the United States.
• 1:16:59
  Was this as much to try and open up the eastern, uh, portion of, of the, uh, country as well?
• 1:17:06
  Yeah, you wouldn't believe it, but actually, you know, New York is one of our biggest states, and it's mostly freight.
• 1:17:11
  So somehow we've managed to collect a bunch of work in New York, and we ship it freight from, from Utah, which is like five-plus days freight, um, al- almost always on a rail, uh, rail car.
• 1:17:23
  Um, and it's always been kind of like, you know, we scratch our heads at that.
• 1:17:27
  We ship a lot to Boston.
• 1:17:29
  Um- Hmm and it's just like, why are we, why are we shipping all the way across the country?
• 1:17:33
  W- Isn't there a better option out there?
• 1:17:36
  Um, and the answer is no, and not in, not the way we do business.
• 1:17:39
  So, um, really we...
• 1:17:41
  It, it's kind of funny because we offer same-day turnaround.
• 1:17:44
  Like, you can place an order today and it'll ship today.
• 1:17:47
  Um, but if it's going freight, then, like, your lead time jumped from a few hours to, you know, five or six days because it's gotta, it's gotta get on a train and go all the way to Boston.
• 1:17:59
  Um, so we just realized, hey, we've got a lot of customers that we're already serving east of the Mississippi.
• 1:18:05
  Uh, shipping is 50% or more of the lead time almost every time, so let's, uh, let's put a shop in a place that has great, uh, proximity to existing customers, and that was the real selection criteria.
• 1:18:18
  So since that exists- Well, Colin and I want to impose on you, Caleb, and, uh, sorry to cut you off on that thought.
• 1:18:22
  That's okay.
• 1:18:22
  But, uh, since site two is in the books, we want site three and four to be Erie and Jacksonville.
• 1:18:30
  C- Colin- Com- completely, you know, no self-interested aspects of it here at all, obviously.
• 1:18:37
  Not whatsoever.
• 1:18:38
  We'll just keep expanding the country for you.
• 1:18:41
  No, it's interesting.
• 1:18:42
  Uh, I, I might expose some of our, um, um, some of our thinking here.
• 1:18:48
  We haven't made solid decisions about the direction we wanna take the business, but there exists a model where we have dozens of, dozens of shops that do 80% of our services, like the real, like the real popular ones, and just, like, put a bunch of hyper-efficient sheet metal shops all over the country.
• 1:19:05
  Uh, you get one-day ground proximity with, I think, seven shops if they're all optimally placed.
• 1:19:10
  Um, and, uh, so think seven Oshkoshs all over the country with one-day ground proximity.
• 1:19:17
  Um- That's a- an option.
• 1:19:20
  It's actually very different from the play that is like, we're gonna do every service.
• 1:19:24
  If you wanna do every service, you really need some concentration.
• 1:19:27
  That's right.
• 1:19:28
  Yep.
• 1:19:28
  Um, if you wanna do...
• 1:19:29
  If you wanna be in everybody's backyard, you need hyper-efficient, like these things just crank, they're, they're OPEX efficient, uh, we know how to spin these out, we could make a dozen a year, that kind of a thing.
• 1:19:42
  Um, that's a different model, and there's maybe some overlap where we have like concentrated regional centers that have tons of capacity and capability, and then smaller shops that offer a sub- a popular subset of services everywhere else.
• 1:19:56
  Um, we'll see where that lands.
• 1:19:58
  Uh, it might depend a lot on how Cincinnati scales now that we have a regional presence there.
• 1:20:02
  Um, but we'll see.
• 1:20:04
  Well, I...
• 1:20:05
  That- that's one of those things you- you've talked about, uh, and written about a bit, is kind of that the- the future of manufacturing is not necessarily a bunch of tiny mom-and-pop job shops, but rather some level of concentration of more capable and more effective, I guess, operationally and execution-wise.
• 1:20:24
  So, I, I mean, I, I, we'll have to get into that another time and kind of that model.
• 1:20:29
  Um, but, you know, as we kind of get to top of the hour here, um, what's, what's next?
• 1:20:35
  What's the big stuff coming for Oshcut that you could share with us?
• 1:20:39
  Maybe some things you obviously can't.
• 1:20:40
  Um, but I guess, what are you most excited about, um, you know, in, in this next chapter as you guys continue to scale and grow?
• 1:20:48
  So I'm really personally excited about maturing the service.
• 1:20:52
  Um, what we found is that a lot of our customers trust Oshcut to become this, like, um, critical component of their supply chain.
• 1:21:02
  Uh, like, we bolt onto their production line, um, and then order parts kind of on demand, uh, and then those parts go in and they make their product.
• 1:21:10
  Uh, a very high percentage of our customers do that, and when there's an interruption and you're feeding a, a, uh, manufacturing, uh, line, it's actually extremely disruptive if there's ever an error of any kind.
• 1:21:24
  So what we're really focused on right now is just absolutely nailing that reliability portion even more than we already have, um, so that we can offer guarantees.
• 1:21:34
  Like, this part will show up- Love it it will be on time, and it will be perfect, and, um, and here's our proof.
• 1:21:41
  Um, so we're going there, uh, and we've got some pretty, uh, interesting technologies we're developing that will enable that without ramping costs way up.
• 1:21:52
  Um, uh, something that basically does not exist right now, so, I'm pretty excited about that.
• 1:21:58
  Exci- exciting stuff ahead.
• 1:22:00
  I got a, I got a final question, Colin.
• 1:22:02
  So last, last week with Will Alverson, we developed a new index.
• 1:22:07
  It's our first national, uh, growth index on Reforge.
• 1:22:11
  Turns out he's fabricating those hooks on dock locks that hook the, the trailer to the building, and so he talked about how he's just seen explosion in dock lock hooks, which Colin and I immediately interpreted the number of dock doors- You've got the growth divided by the average dock doors.
• 1:22:29
  We had the number of distribution logistics facilities.
• 1:22:32
  We're just, uh, you know- Boom we had CapEx number.
• 1:22:35
  Do you, have you seen something like that?
• 1:22:36
  Is there just something that, that most people have never heard of, like a dock lock hook, that Oshcut is excelling in and you think,"Wow, this is an indication of where things are going"?
• 1:22:47
  An obscure market or economic indicator, if you will.
• 1:22:53
  Um, boy, that's a good question.
• 1:22:55
  To be honest, that's probably a little outside my pay grade.
• 1:22:58
  We're, we're a lot like a service center in that we serve hundreds of industries.
• 1:23:02
  Um, so what we tend to see, like if we, we have a, a report...
• 1:23:07
  This is, l- like we build a report that says, that categorizes all of our revenue, and there's like, you know, 100 different, uh, rows in this.
• 1:23:15
  Wow.
• 1:23:15
  Um, but there's always like a few top industries, um, and it could be aero and it could be automotive and it could be, uh, energy.
• 1:23:24
  Um, could be like industrial OEM.
• 1:23:27
  Um, and they kind of move up and down, and one like drops and then the other accelerates.
• 1:23:33
  But what I use and have used in the past for my modeling is just the, the, uh, uh, manufa- purchasing managers index for manufacturing.
• 1:23:42
  Um, it, it has co- historically correlated very tightly with the number of new accounts we're onboarding every single month, um, very, very predictably.
• 1:23:53
  Um, that all broke when we figured out how to, like, mature our marketing function and get the brand out there more.
• 1:23:59
  Um, all my old models broke, and I just stopped using them.
• 1:24:01
  But- um, but purchasing managers index is actually a pretty good one, I think.
• 1:24:05
  Um, just as- All right.
• 1:24:06
  We'll add that to the basket, Colin, as part of- I, I think we've got a, a new question we have to ask everyone- is what's their kind of, what, what's their view on economic indicator?
• 1:24:15
  Um- Mine is, mine, just for the record, I, I was the first.
• 1:24:18
  Mine is switchgear, so Interesting.
• 1:24:21
  Like, uh, switchgears for, like, backup, uh, generators and, uh- No.
• 1:24:26
  We're, uh, we, we got started on this a couple months ago, and, um, we're right now, in order to power automation, and all...
• 1:24:36
  We believe all new factories should be automated.
• 1:24:38
  So in order to power automation, the switchgear order profile is now competing with the data centers for your, your internal kind of medium voltage switchgear and your sh- and all the step-down functions, motor control centers, all that type of stuff.
• 1:24:54
  So it's really exploding because, uh, automation is now the base case, which is the greatest thing in my career, that that happened, is now if you're planning a new factory, it's not like,"Oh, we might automate it someday." It's like,"You gotta automate on day one," right?
• 1:25:11
  So your, your, uh, process electrical demand is dramatically more intense and complex than it was just five years ago when people planned a basic facility and they left the empty transformer pad outside and they didn't really think about the automation.
• 1:25:26
  Now you've gotta have high-quality power everywhere for the, the process equipment, the robotics, the, uh, you know, automated warehouse, whatever it is.
• 1:25:35
  Yeah, that could be one.
• 1:25:35
  I would probably, um...
• 1:25:37
  This is maybe driving on...
• 1:25:39
  I've got time, by the way.
• 1:25:40
  We can keep chatting if you want.
• 1:25:42
  Um- Uh, absolutely.
• 1:25:43
  Think in progress.
• 1:25:44
  The, uh, I would hesitate- Solving the world's problems to use anything as a, uh, future predictor/indicator if AI is influencing it, and that's not because I'm bearish on AI.
• 1:25:58
  I think AI is in the process of and will continue to change everything about computing and how we interact with, with technology, and it's really exciting.
• 1:26:08
  But I would be surprised if there's not some kind of, um- correction, uh, on, uh, that, that affects maybe data center build-outs.
• 1:26:18
  I could be totally wrong about that.
• 1:26:19
  I mean, who knows?
• 1:26:21
  Um, but AI is driving...
• 1:26:23
  Like, we're, we're seeing a ton of demand right now, and I think, and we don't serve data centers per se.
• 1:26:31
  Um, but data centers displace a lot of, uh, capacity.
• 1:26:34
  They take up a lot of capacity and displace customers who aren't data centers, for example.
• 1:26:38
  Like, somebody is making all those server racks and stuff, and they happen in the US, um, for a variety of reasons.
• 1:26:45
  Uh, so those someones are dedicating capacity to data centers that didn't need to be built 10 years ago.
• 1:26:51
  Um, so all the other customers are looking around, and I think, I think that is driving our growth right now.
• 1:26:57
  Um, just as a caution, for anybody who's experiencing a growth boom right now like we are, like we're 100% ahead year over year right now- Woo.
• 1:27:06
  Um- Congrats.
• 1:27:08
  Thank you.
• 1:27:09
  Um, and I, I- Cautiously think that our, our systems are...
• 1:27:12
  The fact that we can do that without everything breaking, I think says good things about our systems.
• 1:27:16
  But I, I can't take victory laps because we're benefiting from these macro tailwinds that we don't control.
• 1:27:22
  And, and I, for, for any listeners who are running, uh, manufacturing operations, I would be very cautious about, um- Over levering to capture this boom, um, because, uh, manufacturing is very cyclical and if you're leveraged and the leverage looks fine at your current operating rate and then suddenly it plummets by 50%, then, then you're in trouble.
• 1:27:47
  So, you know, optimize with- It's, it's two episodes in a row where- Yeah where we've gotten that message now.
• 1:27:52
  So it's a, that reinforced from two brilliant men, you know, say- saying the same thing in different ways.
• 1:27:57
  So I, I think you're spot on.
• 1:28:00
  Yeah, and if you, if you and Will A- Alverson aren't personal friends, you should be,'cause you're both beyond smart, uh, for Colin and I.
• 1:28:06
  But Colin, the, the trick is you have to get 25 pods in before you get the heavy hitters at Steel on, and then you learn- Yeah what's really going on.
• 1:28:13
  But the, just to close off on the AI, uh, I watched one of our good friends, uh, Michael Gayed, who's a, you know, loud voice on X.
• 1:28:22
  He was on Squawk Box, I think yesterday, and they gave him a long segment, uh, to try to unpack this.
• 1:28:30
  And he basically said that there's, there's two issues with AI compute.
• 1:28:36
  Um, number one is, um, that the new models coming out are much more...
• 1:28:41
  that are open source, are more efficient on tokens, so that's gonna decrease demand overall.
• 1:28:47
  And then number two, that he believes, um, the blowback that we see, all the discourse about data centers is just gonna lead to regulations.
• 1:28:54
  And that's just gonna kinda, as it always does, kinda parachute behind it and slow it down.
• 1:29:00
  And that the funding was gonna get precarious based on, this is my favorite topic I don't understand, is the reverse carry trade.
• 1:29:09
  I was like, only Michael could take those three things and just kinda mash them into one coherent ball.
• 1:29:15
  But I do agree, I think it's clearly a boom, right?
• 1:29:19
  And we know what happens after every boom.
• 1:29:21
  So, um, I think you guys are amazing that you're in 100 industries, so you're well protected, you know, and you're conservative in your financial expansions.
• 1:29:29
  That's, that's absolutely brilliant.
• 1:29:31
  Oh, f- Caleb, I had a final question since you had a few extra minutes.
• 1:29:34
  Yeah.
• 1:29:34
  Can you talk to us about your early stage funding, how you put the company together?
• 1:29:39
  Uh, you know, we're, uh...
• 1:29:42
  It was, it's bootstrapped, so, um- Yes so I know some other, uh, people in the industry have said,"Oh, we're bootstrapped," but it works out that they raised$6 million, which is a funny way to say,"I'm bootstrapped." Um- They bootstrapped their capital raising.
• 1:29:58
  I suppose.
• 1:29:59
  I mean, when you're an early stage company, 6 million of seed makes a difference, especially when you're, um, uh, you know, especially when you're, when you're- Yeah, when you couldn't get a loan for 6 million and then somebody hands you 6 million, yeah, it, it makes a huge difference.
• 1:30:12
  Yeah, so when I say we're bootstrapped, we're, we're really bootstrapped.
• 1:30:15
  So I, I took a HELOC on my house, which was paid off.
• 1:30:18
  My man.
• 1:30:18
  Wow.
• 1:30:19
  Previous company, uh, had pay, was able to pay off my house.
• 1:30:23
  Um, and, uh, and that's, that's the primary source of funding for our business.
• 1:30:30
  Also borrowed, um, not borrowed, but, uh, took about$180,000 from my dad out of his, out of his 401.
• 1:30:37
  Uh, was lucky that he had that available, um, so he's a part owner in the business because he did that.
• 1:30:43
  Um, but that was it.
• 1:30:44
  After that we, uh, um, it's been, uh, reinvesting cashflow to grow the business, so.
• 1:30:51
  Yeah.
• 1:30:51
  Well, you would've had- That's awesome a tough time at Reindustrialize, Caleb, because- Oh, yeah?
• 1:30:56
  a couple of us there that when the V- There's, it's about a three to one ratio VC to founder there.
• 1:31:03
  Oh, wow.
• 1:31:03
  So the VCs are kind of walking around and they're looking, they're,"Hey, you got any founders?
• 1:31:07
  You got any founders?
• 1:31:08
  We're looking for founders.
• 1:31:09
  Um, try under the sofa.
• 1:31:11
  There might be a founder under there." And, uh, and our, our shtick in the happy hours was,"Oh, hey, nice to meet you.
• 1:31:19
  We're bootstrapped and EBITDA positive." And then we'd just watch as they scurry away.
• 1:31:25
  And they disappear.
• 1:31:25
  Oh, really?
• 1:31:27
  Yeah, believe it or not, they, they, the majority of them, I, I, I get to say this'cause I own a venture capital company, so that's what gives me the license to dunk.
• 1:31:36
  But they don't wanna talk to you unless you're just burning cash to, for the sake of growth with no profit.
• 1:31:43
  And so I can tell you're an engineer by trade, I'm an engineer by trade.
• 1:31:47
  It does not compute to just run a business at negative forever and just assume the next funding round is, is coming, you know, to me.
• 1:31:55
  Yeah.
• 1:31:55
  Well, I have opinions on that.
• 1:31:56
  Um, we can- Yeah.
• 1:31:57
  We can stop, we can wrap up anytime if you want.
• 1:32:00
  Uh, but I love this stuff, so.
• 1:32:01
  Oh, let, let's hear'em.
• 1:32:02
  Let's hear'em.
• 1:32:03
  Get yourself in trouble.
• 1:32:04
  Let's go.
• 1:32:04
  See what platform we can get banned on next.
• 1:32:08
  Uh, are you guys familiar with...
• 1:32:10
  Oh, geez.
• 1:32:10
  Are you guys familiar with, uh, Hamilton Helmer, uh, Seven Powers, that book?
• 1:32:16
  Yep.
• 1:32:17
  Sounds like college.
• 1:32:18
  Yeah somewhere.
• 1:32:20
  Um, look, uh, I think that, um, bur- burning cash, like burning in the red to capture market share makes sense if you're creating a business that has power.
• 1:32:33
  And by power, in the Helmer sense, it's, um, you have the ability to earn differentiated margin, differentiated returns compared to competitors, and there's a moat.
• 1:32:43
  You have those two things.
• 1:32:45
  If you can create that, it makes total sense to spend as much money as needed, um, to create the power, and those powers can be scale economies.
• 1:32:56
  Um, it can be like, um- Network effects.
• 1:33:00
  Yeah, yeah.
• 1:33:00
  Yeah.
• 1:33:00
  There's, there's a bunch of them.
• 1:33:01
  There's some- Switching costs.
• 1:33:02
  Yeah.
• 1:33:02
  Facebook, Google, SpaceX, right?
• 1:33:04
  Those would be examples.
• 1:33:06
  Yeah, yeah.
• 1:33:06
  Like Faceba- Facebook's a classic network economy.
• 1:33:09
  Like it, it benefits the user, you, if there are other users on it, and it doesn't benefit you to go out and do another thing.
• 1:33:16
  LinkedIn is another one, um, uh, in a different space.
• 1:33:21
  Um, you know, you, you look at ridesharing, uh, platforms like Uber a- and Lyft.
• 1:33:26
  There basically can't be another Uber and Lyft, uh, that work in that way, like because the networks already exist and it's a, it's a two-sided network.
• 1:33:35
  It's the network of drivers and it's the network of, uh, users, um, uh, you know, riders.
• 1:33:42
  And, uh, so yeah, it made sense for them to hemorrhage money for as long as it took, and now they, now they can be cash flow positive because they have those network economies that protect the business.
• 1:33:52
  Um, so, uh, when you look at manufacturing, uh, I think there exists powers that can actually be worth spending to acquire.
• 1:34:02
  Um, but I think it's very hard.
• 1:34:04
  Um, and, uh, anyway, so when I see like a company like Xometry that's, that's been in the red for the last 13 years but is approaching a billion dollars in revenue this year, um, I think their thesis is that they're creating a two-sided network, and once they have it, now they can be cash flow positive.
• 1:34:22
  My thesis has been on that, on the Xometry front, is that, uh, direct manufacturing- Aren't they spending like 30% of their, their cost is just to acquire customers, though?
• 1:34:32
  Yeah, I ran the numbers.
• 1:34:33
  Isn't that like that?
• 1:34:33
  They, they actually sh- say,"Here's their, um, uh, their, uh, h- here's how much we're spending on sales and marketing." And, and then they say,"Here's our active customer count." I think they were spending, last I checked, like$15,000 per every net new account, uh, for customer accounts and, and obviously that's not- Um, in general going to be sustainable.
• 1:34:57
  No.
• 1:34:57
  But they've got this fortress balance sheet, they have so much cash.
• 1:35:01
  Um, and maybe they're right.
• 1:35:02
  Maybe once they've created the network and soaked up all the demand, it's just th- n- now they can scale and monetize.
• 1:35:09
  Uh, but I think that, like, a subset of direct manufacturers like Oshcut are just going to be better priced and more performant as a rule.
• 1:35:18
  Um, uh, supplier differentiation is the Achilles heel of, uh, broker models like Xometry.
• 1:35:26
  Um, so anyway, we'll, we'll see.
• 1:35:28
  We'll see how it happens and how it goes together, but, uh, very interesting.
• 1:35:32
  Well, I have a bias towards founders that own the majority of the company equity.
• 1:35:37
  That's one.
• 1:35:38
  And a, and a, uh, go ahead and, and cube that bias towards founders that took a HELOC, so Oh, man.
• 1:35:46
  There might be a few on this podcast right now.
• 1:35:51
  You have to really believe in something to, to put your house on the line- That's right to fund it.
• 1:35:58
  Well, clearly you guys are doing something right.
• 1:36:00
  I mean, Oshcut has, you know...
• 1:36:02
  O- obviously I'm a fan, I love, I mean, just the entire customer experience, the simulations, being able to, you know, see the pricing update live by, you know, changing quantities.
• 1:36:12
  I mean, I- you guys have done an amazing job.
• 1:36:14
  I'm obviously a huge fan.
• 1:36:16
  Um, and you know, appreciate all the work that you're doing.
• 1:36:21
  For those of you who perhaps, and somehow they've been living under a rock and have not heard of you before, um, how can they find out more?
• 1:36:28
  Where can they, you know, get in touch, learn about Oshcut, all that good stuff?
• 1:36:34
  Yeah, oshcut.com.
• 1:36:36
  We've got a page there.
• 1:36:37
  Our app is accessible from there, where you can get instant prices.
• 1:36:41
  Um, if you have any questions, you can email our support team or give us a phone call.
• 1:36:44
  Um, you can, you can actually talk to a person at Oshcut.
• 1:36:48
  Um, so, um, yeah, happy to, happy to help.
• 1:36:52
  And, uh, yeah, connect with me on X and I'm, I'm open to, uh, to DMs, so- And if you follow him on X, you will see some very cool projects.
• 1:37:03
  You can even harass him about trying to get Star Fox hooked up to- you know, the, the motion rig simulator, which I'm still holding out on that, and, you know, a life-size, uh, you know, was it Lawless Spider from Wild Wild West featuring Will Smith?
• 1:37:19
  Uh, that's for another day, obviously.
• 1:37:22
  Um, so anyways, Caleb, thank you so much for joining us.
• 1:37:27
  This has been a great, great conversation.
• 1:37:30
  You have given us plenty to think about, um, and it is really exciting to see all the moves you guys are ma- making.
• 1:37:37
  Best of luck to you, um, and the rest of the Oshkot team.
• 1:37:40
  Uh, keep up the great work, plain and simple.
• 1:37:44
  Yeah, thanks for the great conversation.
• 1:37:45
  And of course.
• 1:37:46
  And to all of you listening, thank you for tuning in to the Reforge podcast, whether we're talking robotics on the factory floor, policy that actually moves the needle, or the next generation of American makers, our goal is always the same, to prove that re-industrializing America isn't just possible, it's happening right now, and it's pivotal.
• 1:38:02
  If you've enjoyed this episode, make sure you subscribe and follow@thereforgepod everywhere, and check out thereforgepodcast.com for more, including our seven steps to re-industrialization.
• 1:38:13
  Until next time, keep building.
• 1:38:20
  Thanks for tuning in to The Reforged Podcast.
• 1:38:23
  Make sure to like, subscribe and follow us on all of our social media@thereforgedpod.
• 1:38:28
  We will see you next time