SPEAKER_00 0:00

You think your welding job is safe from AI? Overseas RD labs are already training AI robots to read weld puddles, program cobots, and wipe out the traditional shop. Adam Hefner from Hero Machine Company went overseas to source machinery for American fabricators, but what he saw genuinely left him terrified for anyone who refuses to adapt. The future of metalworking isn't years away. It's here. Right now. If you aren't using this tech to scale, someone else is using it to replace you. In this episode, we break down the brutal math between lasers versus plasma cutting, how AI can replace three estimators, and how to stay ahead of the curve before it runs your shop over. Hit subscribe, but ditch your old mindset and open your mind because the rules of the fabrication game have just changed. Well, Adam, dude, it's great to catch up with you again, man. Um, I've been following everything you're doing with um more so on the hero machine side now. And um welcome back to the podcast, man.

SPEAKER_02 1:03

I appreciate it, and I had a blast at the Fabricator Olympics. That was such a cool event that we went to last year. The practical fabrication stuff, the gamification, so to speak, of I won't give away any of the events, but the events you designed were so fun and engaging that I was spending my time, because you know, we had the booth back there. I was spending my time in the booth like trying to think, and we had Steven competing from the maker table shop. So me and him were like strategizing, like, okay, like you know, where to start, how to do this, how to set up the torch right for this type of thing. And it was just uh really a total brain bender, and I think a a well-designed competition. And then you also had the art side, which I didn't even know you guys had the art competition there, and then you had the whole different side of fabrication. So you guys really straddled both worlds of metal fabrication in one competition, which I I thought was really cool. So I had a lot of fun there.

SPEAKER_00 2:07

Well, thank you, man. That's what we were trying to do is just uh make it uh just showcase the fabrication side of the welding and just make it cool and just show people what you can do and have some friendly competition because I feel like the guys who guys and girls who get out there and get it done don't get recognized enough and they don't have a place to as an outlet in order to like go out and show their skills and have fun, like the students do. You know what I mean? You've got all that kind of stuff for students, but uh yeah, we're doing it again this year. Um September the Fabricator Olympics uh Fabricator Olympics is September twelfth, and we've got the welding business summit on uh September eleventh. So we're uh we're pretty excited for that, man.

SPEAKER_02 2:53

That's awesome. I think that's actually really close to the Penn State Ag fair. So Marty in Pennsylvania and I've got the booth. I'll see if I can make both of those work because that would be awesome. Yeah, man.

SPEAKER_00 3:08

Um so you were on the podcast what like four years ago, something like that? Uh when you were it was right in the throes of COVID, because I think you had COVID at the time. And Hero Machine you had you were running Maker Table and you started Hero Machine Company, but you were having problems with getting things delivered over here and tech support and stuff. So tell me what you've been doing in the last four years.

SPEAKER_02 3:40

Well, Hero really took a back seat because I tried to launch it in 2020 during the COVID thing, and I couldn't get engineers out. The original deal I had was I was gonna buy machines from them and they were gonna send engineers out here for every new machine that I sold to train us and train the customers at the same time. And my project manager actually got locked in his apartment. They welded his gate shut because like six people in his apartment complex got COVID and they wouldn't let him out. So he totally got stuck and literally couldn't even get to the factory. No one was at the factories for a long time, they were shut down for weeks over there. So I had to put that first machine together without any help, and that was like really frustrating. And I thought it was the worst decision I made in my life at that point. Just, you know, wasted all this money. We disrupted our whole production. We moved our plasma table way to the back of the back building where what didn't make sense for production flow. We're literally walking parts back and forth through the mud every day to get back to the plasma table. And I had this big equipment payment, and I wasn't making any money with this new piece of equipment. So after all that went down, I also brought in the first wave of plasma tables because that was my big focus, and thought I was gonna sell more plasma tables than anything else. And the plasma table we bought actually ran our shop for like two years after we bought it. But those first six months, we actually had two plasma tables. We had our old shopsaber and then the new hero one that we designed, and that like kept us going and paid the bills. But at that point, I was so shredded, like emotionally, basically, to be honest with you, that I was like, I'm not gonna sell these machines. This is too much for me. There's way more than I thought, and I'm not getting the support that was promised. So I totally couched Hero Machine Co. in 2020 and got my contractor's license, and that's when we got into architectural metals in Maker Table. So I threw myself really hard at architectural metals and got that business developed, got some really good talent here that we, you know, promoted from within and moved them up into the management positions and hired some new people, and then that business started getting kind of boring. You know, I was just sitting here doing estimates all day. That was my whole job was dig through you know blueprints and put a number to it. And you know, super important job. Like I was so scared to give that away. But eventually I started training one of my guys as an estimator and started handing him, you know, we consider small packages. I was like, hey man, just take this package and then you know send it to me. I'll basically grade your homework and do a quick you know appraisal. Because after you do it for years, you get a pretty good feel for if a number's good or bad, you know. Like you're like, this would be a good number if it wasn't stainless. Did you not notice it's freaking stainless steel, the whole thing? You know? So there's a few things like that, but after about a year, he really started running with it. And I've slowly been giving him bigger jobs, and he's got like a bonus system now built in where you know, if he makes a certain percentage of profit on the job that he bids, he gets a percentage of that back. So he's really motivated to push that price up as high as the market will support it, and then to drive the team to deliver on all the hour estimates that we we put into the the quote. So yeah, that's where I got was really about two years ago, I was just kind of designed myself out of work, if that makes sense, which I think is the dream for a lot of entrepreneurs.

SPEAKER_00 7:26

If you spend your day pulling a trigger, burning ride, or rating prints, listen up. On September 12th, here in Kempton, Pennsylvania, we're hosting the ultimate trade gauntlet, the Fabricator Olympics. This isn't your sterile corporate trade show where people hand you brochures. It's built by fabricators for fabricators. It's part community meetup, part friendly competition, and a hundred percent about the shop culture that we live every day. We've got 50 fabricators from across the country testing their speed and precision in real-world challenges, torching, drilling, cutting, attacking, fit up, and a host of other things. Plus, we're hosting a welding rig truck show. So whether you drive a decked out custom rig or a beat-up daily workhorse that gets the job done, bring it out, open up the boxes, and talk shop with guys who actually do what you do. At the end of the day, it's about good food, cold drinks, meeting people in the trade, and having a blast, watching some serious skill on the floor. If you want to jump into one of those 50 competitor spots or just bring your rake to the truck show, head on over to Fabricator Olympics.com and get your tickets today. Because I promise you won't find anything like this anywhere else. Come hang out, let's make a day of it. We'll see you in September.

SPEAKER_02 8:37

Yeah. But then I I don't know, my identity is work. You know, my identity is what I produce. So if I'm not producing anything, I go into depression and this feeling of melees, you know, where I'm just in the doldrums and the wind's not blowing, and I'm just out here floating and nothing. And so I really thought, man, I think it's time to go back into Hero. I think I've got the time, the bandwidth, I've got more financing, I've got uh better foundation now, I've got more power in my building, I have space. Lights.

SPEAKER_00 9:12

Well, you built your shop too, right? Or you you moved out of that that little shop in Nashville and you built your like how big is your shop now? It's uh about 13,000 square feet.

SPEAKER_02 9:21

Okay, and we're on six acres, so I have room in the back. The goal, I'm a big goal guy. I'm just throwing it out there on the podcast within three years. I'm gonna have a 50,000 square foot showroom, training center, and loading dock here on site behind my building right now. So heck yeah. We uh we just rented 13,000 square feet with a loading dock. So I was actually that's why I'm late on the call. I'm so sorry, dude. I apologize. All right.

SPEAKER_00 9:51

Um good, man.

SPEAKER_02 9:52

Yeah, uh, but we unloaded our first container with a loading dock today, instead of in the middle of the yard with three forklifts and rigging and you know a 35-foot-long machine hanging out the end of the container by just holding on by two inches of bearing, all that yeah, crazy machinery moving stuff we've we've done because we had to is over, and I feel like we're in the next chapter. So um nice, you know, that that's I you know, the long short story, whatever, is basically the laser thing's complicated. You've got to be ready for it, you've got to really, I think, technically have a deep understanding, and you've also got to be in a position where you can take some losses because I've been scammed a few times overseas, had bad brokers that have taken advantage of me. Um, the tariff situation screwed me a few times when everything was going up, down, sideways, in between, changing every week. So you've got to, yeah, it's been a lot of lessons to learn, but we're in a really good spot right now. I'm super happy with the manufacturing, the quality, the designs that we've come out with in the last year are fitting all the needs that I'm finding out in the market. And I think we've got an aggressive pricing that is accessible to a lot more people. Like, you know, a FabTech last year, we had a $69,000 special on a two kilowatt up on the booth, and people were coming in like, wait, is that a plasma table or a laser? I'm like, no, that's a laser. Like, that's a laser laser. Uh laser precision, laser speed, laser quality, like full blown. And I we're really starting to see people's gears click, and I'm getting calls from people that don't necessarily run their laser every day or their plasma table. They're just like, hey, we want to bring stuff in-house. We're tired of subbing stuff out and we can't control the lead time, we can't control the quality, and we're disappointing our customers. So with the prices now, man, it makes sense to bring in a laser, even though we're only going to use it five or six days a month, because we can control all these other things that are costing us money. And when you put that into the mix, it's a slam dunk for a lot of these people.

SPEAKER_00 12:12

Yeah, absolutely. So you uh said that uh for you started importing plasma and and your laser, but um so like is talk to me about the is plasma still a thing now that like the lasers are so so cheap? Because like, you know, uh a good plasma table, like I bought a JD Squared and the last thing I heard it was like probably forty thousand dollars, thirty-five, forty thousand dollars now. And like you know, that's not it's not nothing, but uh for a little bit more you have so much less cleanup. So like where does where does plasma like where do plasma tables fit in with the laser scene now, and what's that gonna look like in a couple years?

SPEAKER_02 12:56

I think if you're cutting half inch and below, plasma's dead. It's out of the conversation because like you just touched on, as soon as you start calculating your consumable cost for the plasma and your post-process cleanup labor for every single part that comes off of it, you're already paying what it costs for the payment on a plot on a laser. If you have a plasma and you're running it more than four hours a day, you already have enough money to have a laser. You just don't have it. Like literally. That's interesting. The math just flat out works. Really?

unknown 13:34

Yeah.

SPEAKER_02 13:34

Four hours a day, how many days a week? Uh if you're cutting four hours every day, it's a slam dunk. Like if half the day your plasma table's running every day, it's like no questions asked. Stop what you're doing right now and find a laser. Because wow, like, yeah, once you take what are the hypertherm consumables now? They just came out with a new set. I think they're over 45 bucks. Yeah. So, and I think there might be as high as 60. And what we figured out was once we were also doing artwork, so a lot of pierces. This isn't like base plates where there's 250 pierces in a sheet, these are thousand-pierce sheets. And piercing is where you get the most wear and tear on your consumables. So if you're doing artwork or architectural stuff or intricate detailed parts on your plasma, which you're probably not doing because it's a freaking plasma table, let's be honest. But if you've got a lot of pierces, that consumable is going to wear down after your first sheet, and then it's going to have a hard time restarting. If you sit there and restart your plasma table, you can milk those consumables probably for a week. But you have the cost of somebody standing there restart, restart, restart. So we figured that it was actually cheaper to buy a new set of consumables for every sheet. And we also got some knockoff Italian ones that I think Hypertherm sued them for patent infringement, but we weren't involved in the lawsuit. We were just trying to make a buck and you know, find some profit somewhere. And we figured that with those cheap consumables, we could um have that person that was running the plasma table keep up with grinding. So we didn't have to pay two people to run the table, it got down to one, and then we had the consumable cost. So we were cutting up to seven sheets a day on the plasma table and replacing the consumables every set because it made more sense than the labor cost. And then you've got the grinder, which we could never keep a grinder, it's the worst job in the shop. You you can't keep anybody in that job for more than six months. We bought them uh an air conditioner, they were able to listen to music. We let them smoke cigarettes at that station. We uh also bought them that the weightless arm that takes all the vibration, all the weight off the grinder. They hated that. They were like, that's stupid, that's for girls. Like, you know, I spent like 3,500 bucks on this thing, thought it was awesome, and they never used it. But we did everything to make that station like the nicest, most organized, sweet job, even though you're grinding all day. They were the only people in our shop with an air conditioner, and you could not keep somebody there for more than six months. So, you know, the turnover, the change, uh, the training, the consumable cost, and your cleanup cost, all those expenses together, plus the cost of the plasma table, that's what I'm talking about. Is you if you do the math for you and your shop and you take an honest look at those, you know, four drivers, you're gonna realize that the I don't even know what it is, the twelve hundred dollar a month laser payment, you're already spending the money.

SPEAKER_00 16:42

Yeah, you're already doing it.

SPEAKER_02 16:43

Yeah, and consumables on a laser are next to nothing. Like you can run a lower lens for a week and that costs you ten bucks. Your nozzles cost you twenty-five bucks and they'll last three or four months. Holy cow. Like other than that, you've got air, so you gotta buy a badass air compressor with an air dryer and a filter system on it to get down to five microns for the head. But there's more upfront cost, but once it's in there and you've turned it into a payment, yeah, like it just the math doesn't sell.

SPEAKER_00 17:16

Wow. So I I've been watching a a few of your videos on YouTube. What like what is it like to go from I want to import lasers to like actually being able to import lasers? Give us like a you know, a medium view of like, can you just go to China and like go to these places and say, I want to buy this, and they ship it to you and you're in business?

SPEAKER_02 17:40

Uh in simple terms, yeah. I mean, and maybe I shouldn't be sharing all this, but literally anybody you could go on Alibaba, find a supplier, and order their lasers and start installing them for people. But the problem is the tech support, and we're not just ordering lasers off the shelf. That first machine that we had, I've actually hired an electrical engineer and a mechanical engineer to help me get it running and to build a schematic for it just so we could troubleshoot it because there was no schematic with the machine. And we ended up rebuilding and replacing and upgrading almost everything on that machine except for the frame and the linear rails. So, like we replaced the servo packs, the relays, the transformers, the motors, the software, the head, the resonator, like literally every single major component on that machine we have upgraded over you know that five year span of us owning it before we started Hero. So I'm not just ordering machines from any supplier and just give me whatever you have. Um, and that's why I travel over there so much because I'm driving for my spec. I have created a build sheet of what I'm looking for and specifications for how I want the frame built and what size I want it to be, because a lot of their stuff. Well, I mean, everything over there is metric. So if you order a 5x10 laser table off of Alibaba, you're quickly going to realize it's actually 58 by 119. It's like two inches short one way and one inch short the other way, which is not a huge deal, but it's super frustrating if you're trying to nest your sheets and get the most yield out of your machine. That's like a very simple problem that a lot of people aren't solving. So, you know, these Chinese machines, they will literally make they make machines for the whole world, right? So if they're selling a laser to India, right? And they go, hey, we've got this solenoid here. If you want the good solenoid, it's gonna add $400 to the price of this machine. And a customer in India is gonna be like, $400? What the hell are you talking about? There's no way $400. There's no, the sale is gone. I'm gonna find somebody else. So they'll go, okay, we're gonna put the cheap solenoid on here to make the sale because we told them what it costs, and they told us they don't want the good one. So they've been continually beat down on price by all these developing countries because they're selling to all them too. So, and the same factories will make a developing country model, and there's some brands out there that are selling really cheap lasers that like will get people into the laser market. And I'm I'm glad that they they're filling that gap there and kind of becoming the bridge machine, but they're not production animals, they're gonna have tons of problems, and their tech support is gonna have to be really active. And for a company, it's gonna be really expensive. So you have to come up with I mean, in my opinion, you've got to build a better mousetrap. You've got to figure out what components and what makes a good laser and what makes a bad laser. And then you also have to start learning which components are easily knocked off or commonly copied. Because, like on our press breaks, there's we use Rexroth proportioning valves. That's you know one of the top-tier brands, and they have a metal enclosure around them. So if it's the metal enclosure and says Rexroth, you know it's the real genuine one. But if it's a plastic enclosure and says Rexroth, it's a Chinese knockoff. It's a Chinese company, stole their patented information, made their version of it, whatever is inside is different, and then they put Rexroth's name on top. So I've also had to learn which brand name components are actually Fugazi and not the real thing. So there's a there's several traps along the way in designing and reinforcing a spec on these machines. And then the other side of that is in my opinion, since we don't own the manufacturing, the only real thing that we're offering is service. Right? I mean, if anybody can go over and buy lasers and slap their name on it and sell them, what's gonna set us apart and what's gonna make us last a decade or more in this space? It's going to be service. It's how well you back up your machines. So, yeah, how are you going to train your support staff? You know, how are you going to set up a tech support hotline and document the entire sales process of what the what the customer even bought, what their configuration is on their machine, so you know even what you're fixing when they call without asking them, uh, what machine do you have? What what head is that again? What software? You know, asking all those stupid questions you should already know. Yeah. And a lot of these companies are trying to hire out the tech support and sub it out to other companies. So they're buying the machines, selling them on, and they're saying they have tech support, but then they're just shuttling you off to this other guy that doesn't work for the company, has no vested interest or loyalty or education on their specific machines. It's just, excuse me, a generic LaserTech that knows maybe IPG machines, uh, resonators. He might know Trump and Bistronic, but now you've got this cheap Chinese knockoff version of these really high-speed machines that these guys know how to work on. And they're starting from square one trying to figure out where the button is in the software because they've never used this software before. So they know the principles, but you know, then you're kind of stuck in this loop of now who pays for this tech support guy? Are they pawning that off on you as the customer? So that's that's what I think our real killer edge is, is that we have trained techs that have run these machines in production. Because I'm I'm promoting my guys from MakerTable over into this business. The guys that have been running production lasers and troubleshooting them and keeping them going for minimum eight hours a day. And there's times of the year where this machine runs 22 hours a day. We've split into two shifts, and for about three months out of the year, we run the hell out of our laser table just to keep up with orders. So you want that kind of guy, in my opinion, that is gonna pick up the phone when you have a problem and be like, oh yeah, man, I had that problem, you know, last year or two months ago. It's you know, you got to go and take take an Allen wrench and it got hot, so you need to go to your condenser, twist that open that valve open a little bit, wait for that gauge to go back to two degrees, and then everything's gonna start running again. Instead of that's a ghost problem. I don't know what's going on. I've never heard of that. You know, like who do you want to call?

SPEAKER_00 24:31

Yeah, yeah. Wow. That's uh yeah, so like it's funny because I've looked into other um other Chinese when I was looking for a press break and I was like, yeah, this guy imports he imports them, but like does he have tech support? And I heard from a couple other people who were also looking at buying press breaks from him. Um you know, I I I I I want to say this isn't a quote, but he's like, yeah, most people just don't like don't call and ask that many questions because they're so cheap. They just buy them. And I'm like, dang, if that's the if that's the kind of support I'm gonna get, I don't know if I want to buy one, even if the price is really good. You know what I mean?

SPEAKER_02 25:22

I know exactly what you mean, and I think I know who we're talking about here because I've had at least two of his customers reach out to me right after they got delivery, and I've helped them calibrate their machines and teach them how to get them rolling. So we're on the same page then, yeah. I'm I'm not in the habit of doing free tech support, but if somebody reaches out to me and I just can tell they're a real deal, just like I am, like they're trying to make a buck and support their family and keep the lights on and support payroll and all that stuff, like I'll at least give them the breadcrumb trail. Like, this is what you're looking for. You know, uh one of the guys I he couldn't figure out how to zero out the um guillotine. It was it was coming down too far, it wasn't zeroed out, so all of his bends were over like eight degrees. And there's a correction in there where you can go minus eight, you know, and it'll it'll go back eight degrees and give you your 90. But for a brand new machine straight out of the box, it should be within, like we try to set ours up for plus or minus three degrees because you're gonna have different material thicknesses, the you know, the heat might be a little bit different. You're not gonna get much better than that right out of the box. You you still need an experienced operator. But I was like, hey dude, you know how you do this test? Put a piece of quarter-inch metal in there, tell it it's a piece of quarter-inch metal, and then set the reference point so it just kisses the top of that without bending it, where you can't slide it in and out. You know, I'm like, it's a really simple thing. And then all you're doing is I showed them where to go, and there's a reference point where the machine basically knows how many steps from its limit switch is zero. And I was like, you just keep playing with that number until that plate doesn't bend and you can't pull it out, and then you're calibrated, you know, like it was uh probably took me five minutes to type it up and take some pictures on Instagram and send it to him. And two days later, he calls me and he's like, dude, that was awesome! Thank you so much. The press break is running great. Like, super simple thing to do, and only takes a five or 10-minute phone call with somebody. Yeah, but he wasn't getting that support, so that's that's what we're building. That's why we're also not the cheapest machine. I'm not I'm not chasing the cheapest price. I'm literally overseas every single time I go. I'm like, what's the best? No, no, no. Give me the best. What's the most, whatever you guys think is the most expensive and does it make sense? I want to see that. That's what I want. I was like, the American customer will pay more if you give them more. You know, I'm like, I'm not doing a bait and switch, I'm not gonna be swapping out that solenoid for the cheap one just so I can save 400 bucks on the machine. I'm like, I will charge an extra $400 for the machine and I'll tell you straight to your face like this is why, is because this machine is better. You know, yeah, if you want less problems, call me. If you don't worry about problems and you're just gonna roll with them and be a firefighter and turn into a laser technician like I had to, and not focus on your business for six months, yeah, then go that direction.

SPEAKER_00 28:31

I feel like there's not many guys who have been in a laser like and have the thorough workings of it, other than the tax, like like you do, be because because you had to, you know what I mean? Yeah, totally.

SPEAKER_02 28:47

I mean, that's that's what I'm selling is expertise. You know? It's like and not just laser expertise, like some of these guys, I've got a customer, he's just starting his business, he got a loan, and I'm giving him like business help. I'm sending him my you know, like that YouTube video I made about um how to do pricing, that pricing model spreadsheet. I like sent that to him and sent him the video, and I'm like, this is how you need to start doing your your estimates and your quotes is use this worksheet and you're gonna make money. But yeah, you know, so I I'm even down to I mean, uh actually, I don't want to throw I don't want to name any names, but I've got a uh another customer that he literally wrote me an email last week and he's like, hey, I've got a lot of customers. He uh we haven't delivered his machine yet. We actually just unloaded it today, that's why I was late. And he goes, Hey, I've got a lot of customers already know that I'm bringing the laser in and they want me to quote parts. And I know I didn't pay you for business advice, but if there's anything you can share with me, and I wrote him back, I was like, dude, I'm here for you. Like, I want your business to be successful and I want you to make a lot of money with this laser so that you buy another one from me. You know, like I want you to grow and need two lasers or three, or you know, like so it only helps me. So I I sent him a bunch of time tests and stuff that we had from our shop that we've collected. I'm like, for quarter inch, you're looking at this, for eighth inch, you're looking at that, for half inch, you're looking at at these kind of sheet times, and I don't charge less than $200 an hour for the laser. And you know, most of the time we charge $220 for laser cut parts, you know, machine time. So I gave him all that info because I think that's just gonna make uh a better customer in the long run.

SPEAKER_00 30:37

Yeah.

SPEAKER_02 30:38

So yeah, I'm trying to be full stack, like you buy a machine and you've got me in your corner and and the whole team.

SPEAKER_00 30:46

Yeah, absolutely. Um so in your video that I watched that you posted the other day, um, it's funny because I was talking with somebody um about AI and you know, I I feel like a lot of guys are I feel like everybody's I don't know, there's a lot of guys who hate it, there's a lot of guys who love it, and there's a lot of people who are like, oh yeah, I've talked to Chat GPT once or twice, you know? Um and I was I I I was doing a pick podcast with Dan from DC Bill, and we were just talking about how AI, like what we're doing with AI right now in our businesses. And I saw on the horizon, I'm like, look at it, like I I I'm I'm not super in-depth like as other some other people, but I've made my own apps. I play with cloud code, like I've got I've done I I dug into this l last winter pretty hard. Um and with the understanding of it, I was like, you know what, there's gonna be a time where this is I and I I think I was talking with Donna from Yeti a couple years ago. I'm like, w there's gonna be a time when you can just put you can make a drawing of a railing and tell it you need welds here, here, here, and here. It's gonna remember that, and then it's gonna be able to locate itself and scan that, and it's gonna happen within the next like three to five years. And then I watched your video and you showed them doing it right now. So that technology is available, um, which kind of just blew my mind. So just give me a little bit of a rundown of like what that was like, and um just you know, you know, maybe just like give a visual of like the technology that's out there that goes hand in hand with AI right now that we probably haven't seen.

SPEAKER_02 32:44

Yeah, I came back from China on this last trip scared. Like I'm to be honest with you, it was it was a uh I felt scared. I'm like, these guys over here are so far ahead of us, and 50% of the world's robots are working right now in China. You know, I mean they probably do 50% of the world's manufacturing, maybe that's that's why it happens, and it's not a great statistic if you want to pick it apart, but that doesn't change the fact that half the world's robots are being used in China right now, which means they have more robot experience and more robot expertise and more robot engineers. Like China is built by engineers. 80% of their politicians are engineers, 80% of our politicians are lawyers, you know? So we all, you know, we all we all talk trash about engineers when they make stupid decisions and draw stuff that can't be built, and it gives us uh an air of superiority for a minute that we didn't even go to college and we know more than this dummy behind the desk. But if you just look at how engineers think, is engineers look at everything as a how problem, right? How do we fix this? How do we pump the concrete up there? How do we get the water out of this? How do we make this motor run faster, right? That's an engineer's brain, is how and our politicians' brain is why. Why would we do that? Why why do we need to protect the you know Australian dingo in South Africa or whatever? Like they're they're distracted. So sorry, I'm getting off track here, but AI is like electricity. I just got to start with the fundamentals here. When electricity first came out, people were terrified of it. It was burning down blocks of buildings, it was electrocuting people to death. Like it was sparky, it was scary, it was hot, it was unknown. And I feel like AI needs to be understood like electricity. It's gonna be dangerous in the beginning. It's gonna change a lot of things. You know, all of a sudden, everybody can have an electric motor in their house, a light bulb. You don't have to go to bed when the sun goes down. Like it is foundational type of changes here that we're talking uh with AI. And AI is gonna change the way we work. I don't believe it's gonna replace all of us. I think it's gonna be our favorite sidekick, just like electricity is. We couldn't run our welding shops, we didn't have electricity. But what we're waiting for is the code. You know, we have an electrical code book. We know you got to be double insulated. Anytime you have a connection, it's got to be in this fireproof box. Uh, you've got to oversize your breakers 20% to prevent a fire or an overload. You have to put grounding on everything. So I think we're waiting for the rails to be built for AI. And nobody's gonna build the rails until we start having the problems with it. And that's the unsettling phase that we're in right now, is we're waiting to find the problems. Like there's an AI that they won't even release right now because it can crack every single internet security protocol on Earth. And hackers would have a field day and none of our stuff would be safe. Like credit card transactions, can't do them online anymore. It would blow up everything. So I will say AI is not going to replace you. Somebody that can use AI better than you is who's gonna replace you first. That's that's the first wave. So the second wave is what I felt like I saw overseas in China, and that is the combination of the soft robotics, they call them cobots, and the cobalt is a motor system that can sense voltage overload. And if it comes into contact with something, the motor knows that the voltage spiked and it's not doing its normal move, and that you need to pull back and you can't continue moving in that direction. So you no longer need a laser safety fence, you don't need any fences around the cobot. You can work collaboratively. That's the name of the cobalt is a collaborative robot. You can be standing in the work cell. So the safety part of the robots has been solved. Okay, so that was like milestone one. Then the second milestone is how do we teach this robot to weld apart? And that's what I demoed in that video, too, is I think the coolest type of teaching pendant is where there's a clutch and you put the clutch in neutral and you can physically move that arm to the start and the end of the weld and tell it start, stop, and then you go in and say, Hey, I need a fillet weld three-eighths play. And it's gonna go and lay down that that fillet weld. Where it gets difficult is when you have like we make these garden columns, is what we call them. This is a garden decor thing, and they are big square tubes that we make on the press break out of two two L pieces, and then the robot would we have to reach our arm inside of that tube and then weld all the way back from the inside. So you could show the robot where the weld is, but it might not take that specific path inside the tube, it might try and take the shortest distance and go through the tube, and that's impossible. So that's where the kind of teaching robots have a bit of a weakness is it's like you have to tell it, hey, this point over here in space is how you're gonna enter the tube, and there's no welding commands yet. And then you're gonna have to get in here this way, and now this is position two. You're still not welding yet, you just got in there, and then position three, now you're at the start of your weld. So it's a little, a little cumbersome. Um, and now, you know, that's what I call wave two. Wave three is where you're introducing computer vision. And I've been inside the RD labs of a lot of these places for the last two years, and I've known that they've been working on this stuff for two years already. I just haven't seen it out in the public, so I haven't been able to say anything. And there's other stuff I've seen in the RD labs that I'm still not allowed to talk about of new technology coming down the pipeline in lasers that can basically set their own parameters and dynamically change your settings based off of how the computer is reading the material. Like in incredible. So they've now added computer vision with these easily trainable robots that are now soft that you can be inside the work cell with. So now the next hurdle is going to be training data. So that uh Rob AI company that I showed you, where you can load either your CAD model in and it'll scan. And I think when it does that scan, it will tell you if it's out of square, if it's out of plum, if it's over or under dimensions. I think I didn't get that far with them in the booth, but I think if you have the model, that scanner is actually your QC step. It's confirming that that part is what you want it to be. And then you're gonna assign hey, I want it welded here, here, and here with this, this this is a fillet, this is a butt joint, this is an outside corner. Change your parameters and your settings and your travel speed and wire speed and heat for all these three different situations. And then you press go. So they also have one if you don't have a model, and this is the really scary one, in no model, you just tack something up that didn't have a drawing, and you're gonna lay it on the table where that robot arm can bring the camera, scan it, and it's gonna come up with its own STL file, its own 3D model, and then ask you where you want it welded after it detected all the joints. Like Gray Matter Robotics is already doing this, and they've been doing it with Sanders uh at least for two years. So Gray Matter Robotics has a it's a traditional robot, it's not a it's not a cobalt, it's a Kookka or Yaskawa, you know, big heavy industrial robot, because the big heavy industrial robots have a pressure sensor in them. So it knows how hard that sandpaper is pushing. And they've also figured out how to use that. I believe they're using the pressure sensor. I don't know this for a fact. This is just from what I know of robotics. I believe they're using the pressure sensor along with the voltage information or the PSI from the sander to figure out when the paper is worn. So this thing, you put a planter underneath this robot arm, not in a jig, not in a jig at all. You're just like, hey, this thing needs to get sanded. You dump it on the table, tell it, go do a scan. It scans it, creates a 3D model, and then generates its own G-code path of how to sand it before you even say you need to start with 80, then go to 120, then go to 140, do three passes on this thing, and we're looking for this type of finish. And I don't want you to go diagonal, I need you to go straight up and down, I need you to change the rasterization of your step over, or I need extra step over to make sure this thing's really sanded on your first rough, rough cut. And then when you get to the finer ones, you can do a half-inch step over, and so you can fine-tune it, but that technology has already been running and proving itself for two years. But like now, you add a welder to that, and then what I saw at the booths is they have all these welding inspection cameras. Like, have you ever tried to film a weld puddle? I have, and it just doesn't work.

SPEAKER_00 42:08

No, like I tried one time, it was terrible.

SPEAKER_02 42:11

I mean, Jody Collier at welding tips and tricks, he helped me pass my my structural certs because he had figured out the camera setup of how to actually capture the puddle. And then I've tried it. I've bought different cameras, I've bought filters, I've like jumped into YouTube and taken tutorials. I still cannot get good arc shots. Now they make a camera that's only for checking the arc, right? I mean, it's a little five-inch by two-inch tall box that you can stick on absolutely anything. It's light enough, you could put it on the end of your welding gun, and it would sit there and read all your welds. So they've added computer vision to these cameras that can see through the arc flash. And now the camera is detecting the weld puddle and whether it's connecting both sides of the metal together. And then it's also picking up your slag shelf if you run a flux core or stick, and it's showing you where your slag coverage is and which part of the molten pool is not covered by the slag, and then you can add a temperature sensor to that. And so now you know your inner pass temperature, your apparent metal temperature, and then you know the voltage of the machine, the wire speed. So now you have a completely documented QC weld that if that weld ever fails, you're like, what day was that? Oh, that's this project. You look up in your computer, you throw you send them all the info, like, hey, this was in spec inner pass temperatures were correct, apparent metal temperatures were correct. The preheat was already done before we started welding. We were running book parameters at 27 volts, 150 you know, inches per second on the wire feed or inch per minute. Like, what else do you guys want? Like, yeah, obviously, you abused this this, or it wasn't engineered correctly because we we put this in. So, what I think. The AI welding is lacking right now is training data. It's like we need to get thousands of these cameras out watching people weld, and then the computer's gonna learn, just like Tesla self-driving cars. The more data you feed into it, the more video you feed into it, the better and better it's gonna get. You know, the accident rate of Teslas is way lower than individual drivers now. Like oh man. So welding is still gonna be a job. You know, maintenance is gonna be a thing, but I have seen robots that can climb up columns. So you put a welder on a column climbing robot, and when I used to weld moments when I welded high rises together in the union, like everybody was like, Oh, I'd like to see a robot get up here. Because we're thinking about, oh, you gotta like walk up the stairs, then you gotta jump on the ladder, then you have to walk across this, you know, three-foot-wide stack of decking that's not down yet, and then climb your way up there, and then pull all your leads. Yeah, this is like we're not even taking the stairs, we're just gonna hook this thing to the column on the ground, it's gonna scramble its way up there and come back down. Yeah, weld out both sides of the moment, and then come back down, and somebody's gonna go walk it over to the next spot and go, okay, go to floor eight. Weld that moment.

SPEAKER_00 45:20

It's gonna have video of what it did. It's gonna QC itself. Yeah. And yeah, like you're gonna un you you're just gonna unplug your your your your whip to it, it's gonna walk on over, you plug the whip back in, and yeah.

SPEAKER_02 45:36

This is how the the railroads are welded together. I almost got hired by a company that welds and repairs rails, and they have a big I can't even remember the type of welder it is, but it's a polar welder. It can stretch a piece of solid rail one mile long, it can stretch it 18 inches to bring it into the welder from both sides. So they start they lay these tracks down like three feet apart, and they have to stretch them before they weld them together.

SPEAKER_00 46:08

And this thing would okay, yeah, yeah.

SPEAKER_02 46:10

And it it because yeah, the heat, if you have um the heat expands to metal and it'll start widening the track, and the train will fall right through the middle of the track if they don't do this. So a lot of the times when it gets really hot, that's when they have rail failures, is because that that rail noodled out. So they'll they'll heat them up. They have big weed burners that go and preheat the whole section of track, and then they also stretch it before they weld it, and then they just bang it together with uh electrical current on both sides until it starts to melt on the faces, and then as soon as it starts to melt, it stops and starts to do a little vibration to agitate the puddle in between both of them and then cools itself off, then it grinds all the excess, shears it, and the whole thing is recorded. Every single part of that weld is documented and recorded because they're hired by the rail companies, they aren't the rail company. So when they have another failure, they go back and say, Hey, here's our QC data on that joint. See, everything was in the code, everything was in parameters, it's not our fault. This is just Mother Nature, Mother Nature in, you know. So I think for anybody, production welding, like we have a shop here called Anarco that makes all the pallet racks, and they soak up tons of welders from all the trade schools. It's just like you go there to die. You know, it's just this you're not getting a raise. This it's decent money, you know, you can totally make a living there, but it's like there's nowhere to go, there's no there's no upward momentum. They just need you to fill that gap on this part that never changes. If you've got a job like that, you're the first wave that's gonna get taken over by these machines. Yep. So if you are out in the field, you you've got 10, maybe even 20 years. Like those are harder problems. If you're doing maintenance, like uh we all know Icy Weld, the you know Isaac. Yeah, Isaac the torch master, that guy out there, like line boring, knocking pins out, cutting buckets apart, and making a uh repair for it to get people back on the road, that guy's not getting replaced by AI. Okay, like not anytime soon, anyways. So the the next wave we really got to watch out for is the humanoid robots. And that's where you have a bipedal, two legs, two arms, one head, robot walking, talking, manipulating parts. And that's where that kind of a job gets I don't know, kind of risky, you know.

SPEAKER_00 48:46

So and like risky as in losing it.

SPEAKER_02 48:50

Yeah, you know, yeah. So like and the price on these cobalt arms that I'm seeing is some of them are as cheap as like 10 or 12 grand. So you think about the math. You're paying minimum fifty thousand dollars a year to a welder in your shop, right? And in many places, uh up to a hundred thousand a year pretty easily, especially California, New York along the coast. Like that's a welder. So if you're paying somebody fifty thousand a year on the low end and you find a machine that can replace 20% of their labor, that machine's paid for in the first year if it costs 10 grand.

SPEAKER_00 49:29

Yep, absolutely.

SPEAKER_02 49:30

And it's gonna last more than one year. So year two, you've now made 20% more margin on that welder by having that robot arm just take some of his work. It's not even like a total replacement thing. Yeah. So, like, what's the skill there? Is you gotta be the guy that already knows welding because the the biggest gap I've seen is no robot is doing fit up. No robot is doing fabrication. Once the fabrication is done and it's fit and tact and stable, then the robot can take over and fill all that stuff up, right? So if you are not a fabricator, like Isaac is a fabricator, that dude is out there solving problems. He's doing geometry, whether he likes it or not. Like he is literally doing specialty rigging to hold this thing up to make it safe before he can disconnect the hydraulics so the whole thing doesn't collapse on top of him. You know, that's that's not a spot where a robot's gonna walk into that job any day soon. So I'm telling all my guys, like, dude, you need to learn fabrication skills. And I also came back and I said I was scared. And the first thing I told my general manager said, look, within the next year, automate something. This is one of your new projects. I don't care what it is or how it happens, it could be a simple little pick and place robot that just grabs stuff off the table and puts it on a cart. It could be a welding robot, it could be an automated sprayer for the powder coat gun. It could be something that closes the boxes for us or tapes them up. I'm like, I don't care what it is, but we need to automate something down here so we just understand what the complexities are, what the challenges are, how expensive, how hard, how useful it is.

SPEAKER_00 51:22

Yeah.

SPEAKER_02 51:23

Because I'm I'm a hands-on guy. Like I can read a book, I have great recall for reading, but I don't really learn until I've freaking done it.

SPEAKER_00 51:32

You know, it's like it's like driving a car. Like you have to get in behind the car and go a few feet and just just before you go 60 miles an hour, you know what I mean? Yeah, and and and and this is the time to just learn and implement, even if it doesn't save any much time, or it's probably even if it's a net negative, you're still learning um to how to do it so that when it comes down the pike and you can do things more efficiently and implement it in a better way that actually makes sense, you already have your feet wet and you've got the basics of it.

SPEAKER_02 52:10

Yeah, so I'm not trying to take on a fifty thousand or a hundred thousand dollar automation project. I'm just like, we need to we need to find something. We need to exactly like you're saying, we need to get our feet wet and learn a little bit more. You know, it's like when you you know use the driving analogy. Whenever it snows, I make my wife go out and just spin Brodies in the school parking lot. I'm like, you need to learn counter steering. You need to know how hard you can break when you get up to a stop sign. You need to, because we grew up in California, you know, we didn't we're not grown driving in the snow or seeing people drive in the snow. So every time the first snow hits, I'm like, hey, let's go to the school parking lot, let's go spin some brodies, let's practice counter steering, let's practice hard stops, you know, and get your feel for the snow. So that's where I'm at at the automation process in our shop is we need to go spin some brodies in the school parking lot because it's snowing, guys.

SPEAKER_00 53:07

So what are you um what are you using for AI right now? Like what are you how are you implementing AI in your in Hero and Maker Table? And like what does that look like for you right now?

SPEAKER_02 53:24

Yeah, functionally today, what's actually very actionable and doable in Chat GPT, we have trained it as one of our estimators. So the same training we gave an estimator, we spent about four months training uh Chat GPT, I forget what they call it. It's like one of the notebooks where you can create like your own like kind of workspace inside of there and save your data. And so now it can we load in a set of architectural drawings to it and tell it because everything we do is is coded by its finish. It's like we need MT04, MTO6, and MT12, and that's what we're bidding. So, because that's the that's the stainless, that's the powder coated metal, and that's the brass stuff. Go find it. And then we taught it how we write out our quotes because we're very specific about how many linear feet, how many square feet, what material, what finish, the overall dimensions, what's included, what's excluded. We have all these little caveats that we put because we're basically writing our contract functionally when we're estimating. So anything not included in that, you don't owe. But anything that you specifically, if you're good at writing that out, and that was taking sometimes half the time, was just literally okay, how long is that? How deep is it? How wide is it? What's is it two-inch tube, or is it one and a half inch tube? Is it aluminum? Is it steel? Is it stainless? Is it bronze? Is it brass? Like all that was taking a ton of time. So then we taught it how we do the pricing because we already had it built out into a spreadsheet where everything is activity-based costed. Every activity has its own hour and dollar value. So we break down the whole job before we even start it in the estimating section. So we budget every activity by how many hours it has. So we had that data to already feed it. And from there, we got to the point where it was within 7% of us manually quoting stuff. So now we just automatically increase the price by 10% and we put more bids out. Like we don't we don't sit there and and pinch the pennies on everything. We still review it, we still grade the job and look for any outliers or anything wrong. And there are mistakes. Like, this isn't a fully automated thing yet, guys. It still takes a little bit of work, but it took minimum 50% of the work out of our bid process and probably close to 80. So, how many more bids can we put out? You know, it was everything I could do to bid a million dollars worth of work in a month if I was totally focused and hammering down and not distracted by anything else going on, no machines break down, you know, no people don't show up. And now we can bid $3 million a month worth of work. So our bids don't have to be as competitive because we're putting so many more out. We're just like like we look like we have an estimating team, but really it's like one and a half guys that can do it, and then also has time to do other activities and other roles and projects down here because he's not he's not spending all day doing estimating. So that's number one. Uh, number two, I'm definitely using some image AI to make like the marketing materials. Um, some of the machines I start selling before we have the completed machine here in-house. I've just like seen it in parts or I've spec'd it out and I know the design because I design a lot of the stuff in CAD and then send it over to them to have them build my design. So I'm using AI to build basically the whole hero machine website. There's like a video that plays at the beginning of the right when you get on the homepage. That whole video literally I fed in a rendering of one of my laser machines, and I said, make me a pan shot in a superhero's layer, and when you get around from the side of the machine to the front of the machine, when you get to the front, I want the doors to pop open and there to be some kind of sci-fi alien thing going on inside.

SPEAKER_00 57:27

I'm watching it right now.

SPEAKER_02 57:29

Like, is that not what that video is?

SPEAKER_00 57:32

What did you use to do to do this?

SPEAKER_02 57:35

That was cling AI for video. Cling? Okay. Yeah. So I cling with a Kevin K-L-I-N-G.

SPEAKER_00 57:43

I'm gonna I'm gonna go right there.

SPEAKER_02 57:45

Yeah, I use it in the last video. If you saw where I'm talking about the in India, they're they're putting GoPros on the textile workers to collect the training data. That video of the the Indian workers inside the factory, that was from Cling AI.

SPEAKER_00 57:59

I've been using Chat GBT for my just promotional stuff. And bro, it like because I use it and I see it, it all looks the same. I can spot it from like a mile away. Billboards, people's signatures and stuff. I'm like, you you use chat GPT for that. And I've been searching out like something better that just it doesn't look like I mean, I'm you know, like you can like I can see that this looks a little AI, this video, because you can't read the screen and things like that. So like I can pick that out, but like I'm just trying to find something that works better for me for the image generation that doesn't look like everybody else's AI.

SPEAKER_02 58:44

Yeah, the the image generation, I haven't found a great one because um it can't keep characters, so like you'll prompt it and you'll be like, okay, you come up with this good character A, character B, and now you're like, okay, character A walks into the room and surprises character B, and then it generates it again, and character A and B are completely different, and you already got the first shot of the frame. You're like, I can't have the people change, the story doesn't work anymore. Like, people are wondering what happened to character A? They're character W now. Like, how do they even fit into the storyline? So I mean, I don't I don't know if Cling can hold characters, I haven't done a super deep dive on it, but I will say one of my killer hacks right now is using Gemini for anything Google SEO. So anything on your website, who's the biggest search engine? It's Google. Google, right? So whose data is Gemini training off of? Uh everything on Google. Google. So if you want your stuff to show up on Google, you're stupid to be using any other AI than Gemini. So everything for my website, all the SEO, all the descriptions down at the bottom of the product pages, those are out of Gemini. Because Gemini feeds Google, it feeds off of everything that's going on. Google is feeding Gemini. It's this it's this symbiotic relationship. So if you want stuff to perform well on Google and not pay a bunch of ad dollars and do it with SEO, don't hire an SEO person. Just like literally load your page. Like my wife started doing it uh for MakerTable first, and uh she was hunting around all the different AIs to redo all the SEO on everything, and um you know she's got all the the statistics and the analytics and stuff that she's she's better digging into than I am, but she's like I don't know, three months into this project, and she comes home one day and she's like Adam, Gemini, Gemini, and I was like, What? She's like, dude, I just made my first listing with Gemini, and it's already got a thousand views.

SPEAKER_00 1:00:55

Really?

SPEAKER_02 1:00:56

Like, and I was like, no way. I was like, try it again, you know. So the next day she tries like three more listings, and she's like, dude, and then she's pulling up her analytics, she showed me that night, she's like, look, these this right here, that's all Gemini listings, and she'd been using chat for like the last three months and getting limited success. So then I was like, Oh perfect sense, uh why did like why didn't that like hit me over the head like a ton of bricks? Right, you know, so I think really we have to start looking at what is the AI model trained on, and what are we using it for? You know, so like if I need to code, I'm gonna go to Claude. Yep. Like, so I'll use Yeah, I build my websites on Shopify, so I'll just tell Claude, I'm like, hey, I have Shopify store, honey theme 1.056, and I need to add a press break tonnage calculator into one of the pages. Can you do that? And then I'll I might even feed it like links to other ones that I've seen. I'm like, I want it to kind of look like this and work like this. And dude, I don't even think it was 45 minutes. The whole project was done. It was branded, it was color schemed, copy and pasted it in, made that page visible, and 45 minutes. I'm I made a press break tonnage calculator for people. So you tell it, I got 48 inches, I got quarter-inch metal, I got a two-inch V opening, how many tons is that gonna take? Could I have done that? No, you know, could I have even found some Indian coder on Fiverr or Upwork to do that for me? Probably. But how many months would that have taken? How many dollars would that have taken? Absolutely. Because I've I've gone that road, like I've tried to develop apps and stuff, and like it's thousands of dollars and months of your time. Yep. So we've even made a custom plugin for Fusion 360 that does automatic tab and slot.

SPEAKER_00 1:03:00

Interesting.

SPEAKER_02 1:03:01

Yeah, we're just like, hey, we want tab and slot just like SolidWorks, but in Fusion. Can you make me a plugin? Then it's like, oh yeah, here's some code. Paste that in. Hey, this doesn't work. What should what should we change? Okay, try this, paste this in. Okay, that fixed that, but now we're trying to do this part and that's not working. Okay, paste this one in. So I think within two hours we had a plugin made for tab and slot automatic in Fusion 360.

SPEAKER_00 1:03:28

Wow. I didn't know Fusion 360 let you put your own plugins in.

SPEAKER_02 1:03:32

Dude, there's totally, yeah. So, like whatever you wanted to do, tell AI to make you a plugin.

SPEAKER_00 1:03:39

I love it.

SPEAKER_02 1:03:41

So that's the that's the AI stuff that I'm using today and have firsthand experience with. I'm I'm sure there's more you can do, but um, that's we found so far, and I'm I'm not stopping. Oh, I also have last thing here with AI. I made a Gemini notebook, and every single owner's manual for every single component in my machines I loaded into this AI. So now when we're like, hey, where is the curve offset button on the six kilowatt Bochu C8000 driver card? And it's like it searches all the manuals for me. So we don't we don't search uh owner's manuals anymore. I tell all my guys, like as soon as they pull out a manual, I slap it out of their hand. I'm like, look up the PDF of that online, load it into the AI, tell it what you want, and have it read all 75 pages of that in 10 seconds. Yeah. Like, and if you need to know more info, ask the AI, what page is that on? Show me that page.

SPEAKER_00 1:04:44

It's funny, uh, so I I did something pretty similar. Um, I was looking at buying buildings around here, and uh there is um a building that I was looking at, and and I want to change the use on it. And I went to the township's website, and everything on the town, it was like 180 something pages of like zoning and this and that. I downloaded that PDF, I put it into Claude. Claude went and it read every because it was like it was all scanned in. So it was like blurry and dark, and it went through, it scanned it all in, it double-checked itself and it categorized it. So like I can cross-reference uh, you know, it gave me a PDF of it, and it also gave me the sections of where it's at and what the original page was back to it. And I just sat there and said, I want to do this with this and put this on this building, like tell me what the parameters are for this. And it would tell me, like, oh, you can't do that at this building, uh, your square footage for your uh for the footprint of the lot size, this and that. I'm like, okay, well, how about if I do this? Like, tell me what that square footage is. And it went back and it told me what it was, and it said, if you want to double check me. You can go look at this original page over here. I'm like, you know how long that would have taken me to page through? Like, I would have been there for a day or two, just looking to get my questions answered in that PDF from the township. And I had a conversation with it while we were, you know, sitting there, sitting down watching TV, and the kids were getting ready for bed.

SPEAKER_02 1:06:22

You know, it's amazing. I would have procrastinated that hard job too. You know what I mean? Like two weeks before it's really a problem. Been like get halfway started on them, like, this is too much. I'll I'll come back to this later. You know, I'm out of energy. So yeah. Um, I I think there's lots of use cases. I have it do my uh contract reviews. So I'll like when I get a new subcontract, I'll put it in there, and I'm like, hey, I'm concerned about pay of paid clauses. I'm concerned about the British rule of tort law, which is where if you sue somebody else and you lose, then you have to pay their court fees, you know, and just tell it everything I'm worried about, and then it'll it'll highlight and tell me where all those things are. I still read the full contract, but now I can like, okay, this section is like indemnification. I'm like, okay, insurance requirements, I already got those. So then I can kind of skim the spots that I know are already a lock and nothing's hiding, and then just hammer in on the sections of where I'm at. Yeah. So I love it. Lots lots of uses. I think AI is super useful. And if you're not using it, you're left in the dark. You're you're basically ignoring electricity right now. Like you're choosing to live in a hut on the side of a mountain with a wood-burning stove and burn candles for light.

SPEAKER_00 1:07:42

Yeah.

SPEAKER_02 1:07:43

It's that's literally the camp you're picking. If you are scared of AI or and it's not gonna have any less progress without your participation. Right. You boycotting it has zero influence on how fast it's gonna move. So, yeah, I mean, you gotta choose how productive you want to be. I'm even trying to figure out ways to have my kids use AI. So last month I was like, hey, you guys want to have AI cook us a dessert? And they were like, Yes. So I was like, okay. They're like, how do we do that? I said, well, we need to give the AI an inventory of everything like we have to make dessert in the house. So we like went in there, I was like, tell it, we have brown sugar, we've got milk, we've got oatmeal, we've got cocoa powder, we've got baking soda. So we made a list of everything, and we're like, hey, we want to make a dessert in 45 minutes or less. Give me three options. Popped up three different dessert options. I was like, which one do you guys want to make? And sure enough, we were done in like 42 minutes, start to finish. Wow. And the dessert, it was something I never would have cooked. I can't remember what it was, but it was tasty and uh it used only what we had in the house, we didn't have to go shopping. Like, so AI made us dessert. So I'm I'm trying to like figure out ways to get my kids to use it in good, friendly, positive ways and just get exposure to it because like it's not going away. No, in in some form or another. I I mean, really, my next goal is there's a a PC that actually somebody I met, um, he told me how to build the PC. I don't have it in front of me, but basically you can build your own AI data center. And I think that's the next step is to take this like owner's manual thing out of Gemini, take this estimating thing out of Chat GPT, and take my contract review out of Gemini, my coding for my custom plugins that I want out of Claude and put them on my own AI server so that nobody else gets my data. Is that a um open claw? Uh I I don't, it's not, it wasn't a ready-built machine. There probably are ready-built machines. He just told me like which GPU I needed and hard drive and what processor to build one. Okay. But I think that's gonna be in the next year, we're gonna host our own AI. And because right now, whatever I'm giving all these other models, everybody else is getting the benefit of that. Everything I'm figuring out in lasers is as soon as you jump into Gemini and ask it a question about a laser, you're gonna get a great answer. Because I already did all the work over here. You know?

SPEAKER_00 1:10:30

I I actually played around with that a little bit. Um, I had an old laptop laying around, and there's there's a program called Open Claw, and it it kind of interfaces and piggybacks off of one of the other. You know, you can choose whatever model you want, but basically it hosts your own um it hosts all the data on the computer. So like you know, when you get deep into like a conversation or like a like a chat, it like the memory creep starts to like you know go away and you for you're like we already talked about this because it forgot because it was so far away. Well, this it doesn't forget because it's all up and ready on this computer. Because like w when you're interfacing online, it needs to clear out some of that data so that it can run fast and it be because and just so that it's more, you know, it basically it runs fast and it doesn't use as much computing power reading all the other conversation that you had for the last two hours with it. But this actually will remember all that data so that like it's it's almost like an infinite memory, and it is all hosted on this computer that stays on all the time. And mine actually interfaced with um telegram. So I can so I had a bot that I texted in Telegram. I actually like uh you know, I do speak text all the time. So I would talk to this bot and I would tell it to do things. So basically it gave it arms, so it would actually go and search for things every morning at 6 a.m. on the internet, and I gave it parameters of like, you know, I won this, it's a hundred dollars. If you find it for 50% off, message me and let me know. Give me the links, and I want to go buy it. And you can actually give it permission to go and actually buy stuff for you, but it's kind of like the wild west right now, like uh it's easy to get hacked, and I actually just shut it down because it didn't really work that great for me. But um, yeah, it's crazy what you can do out there, man.

SPEAKER_02 1:12:41

Yeah, I mean it's it's coming for all of us. My buddy uses uh he buys sneakers, like high dollar, you know, rare sneakers, and he has a bot that just buys sneakers for him while he's at work. Wow, yeah, like he's like, Oh yeah, these are I don't know anything about shoes. He's like, these are panda dunks, these, you know, anytime they drop below $90, buy them. Wow.

SPEAKER_00 1:13:06

I love it. So a question for you since you've seen what's going on over in China, um give me your opinion on what a fab shop will look like implementing different components of this AI. And get like, you know, we've got this this uh this cobot, this arm that comes and let's say I'm uh I'm a guy in my shop, I got two, three guys working for me. I want to start implementing this AI to get ahead of the curve uh and really just kind of punch outside of my weight class, you know what I mean? I can get more stuff done with less guys. What is that gonna look like as far as machinery and technology? And what is the time frame in your from your eyes?

SPEAKER_02 1:13:59

I already know an engineer that owns a welding shop that doesn't know how to weld because he has two cobots. Like that has existed. So yeah, I think honestly, finding a good cobalt and putting a MIG welder on it and finding that 20% for it to do, which for like my product mix with Maker Table, I have a whole retail business that all the fabricated parts are repeatable. It's that planner is the same planner. That geo planner never changes size, it's the same thing, you know. So I think I would introduce a jig that way don't have to keep retraining. So my my idea is make a three-sided welding table that pivots, and one side is flat for manual fabrication, and the other two sides are full of jigs for my products. So, and this table rotates basically like a triangle table, right? And so I'm like, hey, I'm doing bar stools. You rotate the table to the bar stool jig, and that bar stool jig never moves, it's in a you know, the 5/8 hole two-inch grid tables, it's bolted down there. You load your parts in, and then you tell the robot, hey, we're doing a bar stool. Go do the bar stool. And since that's in a jig, it never moves. You don't have to retrain it every position now. You just pull up that file and it goes, oh, there's a bar stool over here. Bar stool. Then you flip it over, you're like, hey, do the other side. So then what is your guy doing while the welder's welding? Now he's fitting up the next part already. You just got him an assistant, you just got him the best welder's helper he's never had. Right? And that is ready to go out of the box right now. You know, and I know you can buy systems for like 50 grand that'll do that. I'm I'm not trying to bring in the cobots and robots yet. I just went over there to that show to see what's coming and what what new has come out because I think we do need to add cobots to our product mix in the next year or two. I think if we're gonna be a metal equipment shop, we need to be doing that. So, first phase for us is bringing in laser welders. So I think that's a huge opportunity right now, and people are finally getting it and seeing all the use cases, and it's no longer like a sideshow circus act. It's like real application. You're gonna save grinding time, you're not gonna have as much warping and shrink on your stainless steel parts. The heat affected zone is so much smaller, the penetration is deeper, like green light go. The limitation on laser welding is AWS hasn't come out with a stated welding procedure for it for the structural shops. So they have six kilowatt, four wire feeder, automatic nitrogen generator welders over there for structural shops, but I'm not gonna bring any of those in because AWS doesn't say you can laser weld a structural joint yet. And that's I I'm not gonna do that work. I'm I'm not a CWI, I don't have any connections to that. I'm just waiting for Lincoln Electric or Miller or somebody to take that on and start saying, okay, here's your welding procedure. You've got a structural joint. This is what you got to do for a laser welder. And once that happens, I think we bring in the big structural laser welders. And once you've got the laser welders in the lineup, the next step for those is to automate them. So this is phase one is getting a good welding machine. And then phase two will be putting it onto a cobalt and laser welding with a robot, which just sounds like the future. Like, I can't believe I said those words out of my mouth right now. Like, I'm almost 40 years old. Like that that's sci-fi storyline. That's the the but it's real. It's yeah, like it's not there yet. So I'm I'm waiting for the AI to get more training data. Like, I see what they're doing, I don't have confidence that they're ready to go out of the box. All this computer vision stuff that's on the robots, I've seen the two-year development process in the RD facilities and seen the problems they've had and the challenges in place. I don't think they're all the way flushed out yet. You know, but to say that they're not gonna figure it out is foolish.

SPEAKER_00 1:18:29

Yeah.

SPEAKER_02 1:18:30

So they have so many engineers working on it. Like the we're still gonna do a uh a battle between the I've picked two laser welder manufacturers that I really like. And so I bought a machine from both of them. We're gonna bring them back and torture test them versus our Everlast that we've run for over a year now. Um, but one of the companies has 50 engineers full time that their whole job is to make their laser welder teachable in five minutes or less to someone that's never welded before. That's the stated goal of their engineering department. It's on the wall. Everybody talks about it. You ask them what they're working on, they're like, oh, this, so we can get down to five minutes. Oh, I'm working on this, so we can get down to five minutes. They're like focused. 50 engineers. Wow. And that's not even a huge company. Yeah, like they're a big company. Like I'm I'm not saying they're like mom and pop shop, but like they're not the you know, hundred million dollar laser manufacturer companies, but they're gonna succeed.

SPEAKER_00 1:19:33

Like you've got that much brain power going at a problem, they're gonna succeed with that, dude.

SPEAKER_02 1:19:38

And they have all the manufacturing there, so they want to try something. Which of the 70 CNC mills do you want to run it on? You know, like, hey, yeah, we got machine time. We go use that mill down there. Go, go make that part. Like, they're making their own nuts that go into the plastic extrusion, they're like making their own threaded parts, they're making literally everything, the plastic, they have plastic extrusion at this factory. Like, they're yeah, they're making their own silicone uh gaskets. They have a silicone factory inside this thing. So, like, you want to you want to change something? Yeah, just go over there to the silicone thing and make that gasket one mil thicker and then go put it in one of the machines and see if it works. It's not like, well, we got to order one, see if it exists, get one made. Uh no, it's like literally every part of that machine is made there and torture tested. Like they have an oven, they crank up to what is it, 114 degrees Fahrenheit and 85% humidity, and then they run the machines for 500 hours at 100% duty cycle to test them. So when they say this thing works at 104 degrees Fahrenheit at 85% humidity with 100% duty cycle and it's air cooled, they literally mean that. Like they had a uh test unit for bending the whip. So they they take the whip right where there's like that little spring that is like the stress relief, and they bend it 10,000 times to make sure it doesn't fail. That's crazy. Like so the speed at which they can uh innovate and change is yeah is nuts.

SPEAKER_00 1:21:22

Well, Adam, it's uh I it's mind-boggling to think what the uh what the future looks like. So I appreciate you giving us some insight on that. And um yeah, man, thank you for coming on. Thanks for sharing your knowledge, and um I really look forward to seeing what Hero Machine Company turns into, man.

SPEAKER_02 1:21:41

Thanks, man.

SPEAKER_00 1:21:42

I can't wait to get up to your shop and hook you up with a tube laser. Dude, uh like my vision of like uh of just a one-man shop or a three-man shop putting out like tens of thousands of feet of rail, man. It's it it's be it's it's every year it gets closer and closer.

SPEAKER_02 1:22:01

Yeah, I mean, one of my biggest pains is I feel like I can see the future. I know where I need to go. I you know, I know where it's gonna end, but I just I don't have the time, the money, or the talent to get there right now.

SPEAKER_00 1:22:17

You know, no, yeah.

SPEAKER_02 1:22:18

And it's just like just keep digging.

SPEAKER_00 1:22:22

There's gonna be a tipping point where you know our ambitions of what we want to do and the and the and the the the technology is gonna be like, you know, if you try to do it too early, you're gonna spend so much money failing and so much time failing. But if you do it too late, everybody's gonna be doing it. So like there's there's gotta be that intersection of like when the technology is catching up and when the you're and it meets your time and ambition and the feasibility to do it, I think that's getting closer and closer.

SPEAKER_02 1:22:56

Yeah, the the battle's not at our front door yet, but I can hear the drums beating in the next mountain range over.

SPEAKER_00 1:23:02

You know? Exactly.

SPEAKER_02 1:23:03

So that's my that's my prediction, is it's coming, and I don't think it's quite time to jump into it immediately, but just next time you go to a FabTech or uh, what is that, IMTS, one of these big shows, spend an hour or two looking at these cobots seriously, and think about your products and how it could functionally weld out that rail. You know, you'd probably have to have a cobalt on a rail next to the table that could travel the length of the table, but it's it's not crazy. And if you find something at Fab Tech, call me, dude. I'll get you direct. I'll be like, dude, like these are the guys to call. Like, just save yourself, you know, 60% off the retail price and like you know, deal with some of the headaches, but yeah, man, like because that's what I'm gonna do. I'm gonna I'm gonna go buy one of those things direct and bang my head against the wall and save the money and figure it out. So, and if I figure it out, you'll be the first to know, man. I'll I'll give you a call. Be like, dude, I got the handrail robot, man.

SPEAKER_00 1:24:14

Well, man, I look forward to it, buddy. Um, I appreciate your time and uh let's catch up again soon.

SPEAKER_02 1:24:19

Likewise, it was great chatting with you. Thanks for inviting me on the show again. And uh I'm gonna make that Fabricator Olympics work because it was worth it. I love it, man. Thank you. All right, dude. All right, talk to you soon, buddy.

SPEAKER_00 1:24:32

See ya, have a good evening. See ya.