By Jennifer Dubose

530. The AI Everyone Will Actually Care About at IMTS 2026

AI is going to be everywhere at IMTS this September, and if you walk the floor without a filter, it’ll make your head spin. So we brought in someone whose whole job is helping manufacturers tell the real thing from the marketing fluff.

Ryan Kelly, VP of Technology at AMT, joins us to break the noise into something usable. He walks through a simple way to sort every AI claim you’ll see into three layers: the hyperscalers and cloud platforms laying the foundation, the AI native tools that couldn’t have existed a few years ago, and the embedded AI features showing up as new buttons in the CAD, CAM, and ERP systems you already run.

Then we get into physical AI, the part that actually matters for a machine shop. It’s the loop that takes information from the physical world, lets an AI reason on it, and sends a decision back to the machine or the robot in close to real time. Nick’s tool load monitoring, Paul’s native machine monitoring in ProShop, Mike’s managing 60 jobs a day by exception, they’re all versions of the same idea. The catch is data. If the grammar of your data isn’t good enough, the AI will hand you an answer you can’t trust.

Ryan also previews what AMT is building at the show: a new Industrial AI Arena with 32 exhibitors, a one day Industrial AI Conference focused on the factory floor instead of TED-talk futurism, and an Emerging Technology Center where, alongside Oak Ridge National Lab, they’ll be making and flying drones right off the line. Three quadcopter models, a fresh one every couple of minutes, with the ability to change the design on a dime and ramp right back into production.

Whether you’re already wired up or still getting your data in order, there’s something at IMTS for wherever you are on the journey. Take stock of your readiness before you go, bring your real problems to the vendors, and you’ll come home with something you can use on Monday. That’s making chips.

What’s Covered in this Episode

  • (0:00) Drones that fly themselves right off the assembly line
  • (2:11) Ryan’s back: the tariff comment that drew hundreds of responses
  • (4:40) Boost spindle uptime without adding headcount with the Hennig Workflow
  • (5:40) Meet Ryan Kelly of AMT, and why this year feels like AI-MTS
  • (7:16) A filter for the floor: the three layers of AI at IMTS
  • (11:42) Turn AI buzzwords into outcomes at the Industrial AI Conference at IMTS
  • (12:30) How to shop for AI: bring your problem, not your intimidation
  • (13:58) Inside the new Industrial AI Arena: 32 exhibitors and the ETC
  • (17:15) Know your AI readiness before you walk the floor
  • (19:41) Why it’s worth seeing tech you’re not ready for yet
  • (21:37) Two frameworks: the data-to-application stack and the part lifecycle
  • (23:13) Automation everywhere, and what physical AI actually means
  • (27:14) The human body metaphor: sensors, feedback loops, and real-time decisions
  • (31:52) The real challenge isn’t robots, it’s the data and its semantics
  • (33:40) ProShop’s machine monitoring: what people say vs what machines say
  • (37:52) Reach true high end machining with DN Solutions
  • (39:03) The two dangerous extremes: blindly trusting AI vs writing it off
  • (40:48) Managing by exception: AI flags the few jobs that need you
  • (42:44) Reflexive factories and capturing shop floor knowledge for AI
  • (47:35) Inside the ETC: making and flying drones live with Oak Ridge National Lab
  • (53:50) Register at IMTS.com and find us live in the Hennig booth (339133)

Resources Mentioned

Read the Full Episode Transcript

[0:00] I think the cool factor is that we’re effectively flying drones right off the assembly line. That is so cool. Which I think will be cool to see. I mean, that is like what you imagine in some futuristic movie about like, you know, in the year whatever, we’re preparing for a war and you just see like this, all these drones flying out of a factory going to work right away. Very cool. I think, yeah, I think, I think the concept that we’re working for now is that they’re going to fly themselves like basically into like, you know, the shipping box or like whatever, like, you know, the last mile delivery, they’re going to sort of fly themselves. So that is so, so, so cool. It’s like a Tesla car coming off the factory line and driving itself straight to a dealership or straight to someone’s house. Yeah, right. Yeah. That’s kind of the thing that we’re, that’s our, that’s our goal. That’s what my wife and I are working on. Like we just keep her pregnant and then she has a baby and they just go right into the family business. If the sound of a machine tool removing metal gets your blood pumping, then you are Metalworking Nation. This is Making Chips, where we talk all things metalworking, engineering and design, production and tooling combined with business best practices, technology, marketing, news and new media for manufacturing professionals.

[1:31] Here are your hosts. Let’s make some chips. Nick, do you have something stupid to say to start this episode? Yes. Good. Are you ready? That was a stupid question. Nick always has something stupid to say. That’s not fair. You guys. Oh, are we recording? We are. We are. Oh, okay. Well, I was going to say, cause that leaves me in the difficult spot of having to say all the smart stuff. So that’s why you’re here, Ryan. Having had conversations with you, I don’t think that’s going to be a big lift. So excited to have you here, Ryan. We’ve had some dumb ones. Well the last time you were on, Ryan, it was an emergency, if I remember correctly. Yeah, we were talking about tariff stuff. I think there was a high impact conversation. I don’t remember exactly what it was about, but I think it was trying to encourage people to respond to the open comments on tariff. And I think that particular comment, or sort of opportunity to comment, received hundreds and hundreds of comments, which is far more than any of these things normally get. So I’m not sure if, you know, the making chips segment was solely responsible for that, but it definitely didn’t hurt. It was. It was definitely us.

[2:49] Well, you asked if we had something stupid to say, and making chips was solely responsible for the massive response to that call to action. And remind me again what it was, Ryan. I think it was in reference to the machine tool importing being tariffed so high. And unfortunately, there’s not that many North American machine tool builders or U.S. machine tool builders. So most of us import our machines. But I would say, you know, whether it’s a result of the comment or just all of the discourse that followed, you know, from that announcement on the government side, and this is something that AMT has been really involved with, is the government is really interested and want to understand more about the whole value chain for machine tools and robotics as well. Not just about the final product itself, like the machine tool, but the things that go on it and things that go into it. Really trying to understand what it’s going to take to, first of all, allow Americans to build more factories. But then also how we can really galvanize a domestic machine tool industry that could be just making those replacement parts made here, or maybe those machine tools made here. So you know, there was a lot of fervor surrounding those events. But I definitely think the government is paying very close attention, wants to understand it more and really put us on better footing in that respect. You know what would be awesome, Ryan? If there was like one multi-generational family who decided to build an automation system in the Chicagoland area.

[4:27] That’s what the world needs. It seems too far-fetched, Nick. I don’t know. That’s a unicorn. We’re trying our best, man. If you’re looking to increase your spindle uptime without adding any headcount, check out the Hennig Workflow Automation System. I’ve seen it. I love it. It’s a compact, automated pallet delivery solution that lets you work ahead of your machine so your shop never stops. You can load up to 40 pallets with material, fixtures, programs, and just let the workflow cook. The Hennig Workflow is designed for high-mix machining, works with a wide range of machines, and even with a 125-pound payload on top of a 300-millimeter pallet, yet still takes up less space than most competitors’ pallet automation systems. That means more production, fewer manual changeovers, really just a practical path to high-profit lights-out machining. Learn more about the Hennig Workflow by visiting hennigworldwide.com slash workflow. Yeah, it’s a lot of fun, though, being in the automation space as we get into this episode with Ryan. We’re going to get more into technology, and Ryan, maybe you could give us a bit of background on kind of your role with AMT and why they called and told you you’re going to be on today to help us talk about the tech at the IMTS this year.

[5:51] Yeah, sure. So, I’m Ryan Kelly. I’m vice president of technology at AMT. I guess that puts me in a good position to talk about some of the new, exciting technological developments that we’ll see at the IMTS trade show this September. I think it’s 14th through 19th in Chicago, as always. Yeah, I’ve been at AMT going on seven years. I was a member of AMT with our startup, MakeTime, and then after that, Xometry, for about seven years as well. So I’ve been around AMT, IMTS, for a decent amount of time. I think this show is going to be special. I mean, I think we say that about every IMTS. This one, we’ve been joking that it’s AI-MTS this time right now. In the last several IMTSs, there’s been, I would say, an increasing amount of AI, but this is really nothing new. Some form of AI has been in manufacturing probably since the 80s. So it’s not like this is the first time it’s showing up. I think what is unique about it this time around is that it’s going to be everywhere. In fact, I think it might be a little head-spinning. Everyone, I think, listening to this podcast is probably getting tired, right? Every single company on earth has some kind of AI offering. I’ve been doing a lot of thinking, and I think maybe one of the things that I can provide your listeners are a way to sort of make sense of the AI that you will see at IMTS and some of the offerings or experiences that we’re giving to the visitors to make a better sense of that.

[7:36] So I kind of think about the AI as sort of three layers and what you’re going to see at IMTS. So we have, at a more foundational level, we’ve got the hyperscalers and cloud platforms themselves. So AWS, Google Cloud, Microsoft, those kind of companies are going to have a presence there, really giving the foundational infrastructure to the companies that might want to build AI systems on top of that. Then there’s going to be… Yeah, it’s great. I mean, I think maybe last IMTS was the first time that we saw those hyperscalers have a presence there. That presence is expanding. It has to, right? I mean, for one thing, it’s well known that the industrial sector just produces a lot of data. Data is what really makes AI work. It’s been really, I think, underserved by the hyperscalers. And so it’s great to see them at IMTS and I’ll be really interested to see the offerings that they have in their booth this time. But to me, it’s a signal that manufacturing is getting the limelight that it deserves and should have had for a long time, right? To have those companies there. And we know all the money that’s pouring into the industry, you know, just the investments that are kind of outside into, you know, SendCutSend and, you know, Hadrian, all those things takes a lot of tech.

[9:05] And it’s really cool that those brands are now maybe going to be just common at these shows. Yeah, I really think that they should be, right? It’s, I really look at this, to be able to do AI in manufacturing, you need to have the whole stack and the hyperscalers and the infrastructure that they provide is a big piece of that. I think that they recognize that this is a huge and I think to your point, Paul, relatively underserved market. And I’m really glad that they’re going to be there educating the visitors and, you know, making more connections, because I think that’s going to help accelerate. So the next, I would say, layer is, you know, what I call AI native platforms. And so, like, and I’ll maybe say the third layer is sort of embedded AI. You’re going to see embedded AI is just like a feature, a lot of tools that you’re already familiar with. And that’s, I think, a lot of announcements that we heard that in 2024, but it’s going to be increasing the fact that, you know, a lot of these companies that already have some sort of digital solution, they are now going to have new features that are AI enabled, right? You’re going to see new buttons in your CAD system or your CAM system or your ERP system. So there’s going to be a lot of that.

[10:28] But in the middle there, you know, there’s a lot of companies that are going to be showcasing what I call AI native capability. And this is tools that would not have existed previously, because without all this access to AI, they just wouldn’t have existed at all. And so that could be, you know, fully automated CAM systems or like a company, C-Infinity, their auto assembler tool, where these sorts of things just would not work without the underlying advancements of AI technology. And so that’s maybe one way, you know, for people to think about, you know, you’re going to see like the two letters AI in so many different places. But they can say, you know, is this infrastructure? Is this an AI feature in an existing product? Or is this something really completely new that is AI native and would not have existed before and could not exist without the advances of technology? That’s, I think, a good way to think about all the AI that’s going to be seen at IMTS. A.I. is everywhere right now, and it’s going to be all over IMTS, but here’s the real question.

[11:48] How do you turn buzzwords into actual manufacturing outcome? That’s exactly what the IMTS Industrial AI Conference is built for. On Wednesday, September 16th, head to the South Building at IMTS for a one-day practitioner-focused program on reality-based AI for the factory floor. This isn’t vague futurism or jargon, Sue. It’s about understanding where AI really fits in manufacturing, from predictive maintenance and quality to edge versus cloud, small models, LLMs, and how to move from problem to purpose to process. If you’re trying to figure out what AI can do inside your shop, this is the room you want to be in. Learn more at the link in the show notes or head over to IMTS.com for more info. So, for a shop owner or shop leader that’s going to come to the show, and they have this dizzying, head-spinning array of AI everywhere, what would you advise someone to kind of how to frame their search or what to assess what they’re looking for that’s going to be something that’s tactical and meaningful that they can bring back to their own shop? That’s a great question. I think the good news is I would advise them to evaluate it just as they were evaluating any other technological solution. What’s your problem? And really bring your problem to the vendor and see how they would approach solving it.

[13:08] I definitely think that there really is a difference in what these tools can accomplish, but I think the way of buying it and evaluating really should be familiar and prospective customers shouldn’t be intimidated at all just because it’s AI and maybe they don’t understand it. I mean, I think if you go deep enough, very few people and maybe nobody really understands what’s going on with some of these tools. And so really, can you solve my problem? And if you think that they’ve got a compelling answer to that question, just as if you were, you know, evaluating buying some workholding or tooling or a new piece of capital equipment, I think the buying process is largely the same and you shouldn’t be intimidated just because it’s AI. Ryan, so, I mean, with the AI that you’re expecting to see there this year, you guys are going to have a whole area of the show dedicated to AI solutions, correct? Yeah, so we are doing a new thing this time around, which we’re calling the industrial AI arena. That’s going to be like, I think, you know, right into the grand concourse, so pretty central to the show. And there’s going to be a lot of booths, 32 exhibitors, plus India National Labs and an exhibit that we’re working on with Oak Ridge National Lab, FANUC, MESAC, PRUSA, and Edgewater Automation right there and what we call the Emerging Technology Center.

[14:37] So the industrial AI arena is going to be right there. And these 32 exhibitors we invited to participate, these are going to be more of these AI native solutions that I was talking about. And it kind of runs the gamut. You know, it could be everything from supply chain to, you know, working with your CAD models in an AI native way, but it really is going to, I think, broaden the horizons of the visitors that really understand what is capable with AI across the entire value chain of manufacturing. We also are going to have the first time that we’re doing an industrial AI conference, just a one day, a full event. And there’s going to be speakers also running the gamut from predictive maintenance, quality, Edge versus cloud type solutions, data readiness, really more practitioner focused. I think, you know, the idea is that this is going to be a lot of practical information that manufacturers can bring back to their facilities, as opposed to, you know, more TED talk, inspirational, what the future of manufacturing is going to be like. I think in both cases, what they’ll see in the industrial AI arena and at the industrial AI conference, we’re really trying to make this actionable right now. We’re not really interested so much in the future of industrial AI.

[16:11] We’re more interested in bringing what’s available now, making it real and really making sure that the experts that are going to be there really understand industrial processes and actually how to make this stuff really happen. Mike, Paul, do you remember the title of our main stage that had to do with AI? I don’t remember the title. At the last IMTS, it was Practical Use Cases for AI in the Machine Shop. And it went from, like, us talking about it for an hour across the lunchtime of one of the days of the show to now there’s a whole AI arena talking. And I love that it’s focused on practical use cases. Yeah, I do too. If you want to find out, like, the craziness with AI in the future, there’s a zillion YouTube videos about how humanoids are going to do all our work for us and we’ll be able to float around on a raft. You know, you can find that stuff. But I think what people like us want to find is, what can I use tomorrow to run my company better and specifically to make parts in a machine shop? Ryan, are there any particular use cases? I know you don’t want to, like, spoil what IMTS is going to show, but is there anything or any categories of kind of practical use cases that you’re especially excited about? I think it’s going to be different for different people.

[17:30] And one thing that I think is really important is going to be for people to understand their level of, like, let’s call it AI readiness. And so for folks that, you know, might be a little bit more, have made the investments in their data infrastructure, wiring up their machines and so forth, there’s going to be a different set of things that they’ll be able to take advantage of versus someone that might be not connected, not really have their data in order. And so it’s difficult to answer that question more broadly other than to say, I think at whatever place a manufacturer is on their sort of journey towards AI, they’re going to find solutions to help them. And that might be a good way to, and I think at AMT, we’re planning to put out some resources so people can kind of benchmark themselves before they go to the show, really understanding where they are. My take on it is you can only do so much if you’re not sort of building on a solid foundation. And so if you know that your foundation really needs a lot of work, then it probably makes sense to, you know, talk to the AWSs or the Googles or the Microsoft about how you get that in order.

[18:55] Whereas if you know that, you know, your data is in really good shape, then you’ll be, you know, maybe wanting to talk to, you know, some of these AI native platforms saying, OK, yep, here’s the state that my infrastructure is at, and that’s going to allow them to do a lot more. But the reality is, is I think that it’s pretty likely that at almost every booth that a visitor would go to, there’s going to be some conversation about AI that’s going to happen. And so I think it’s important to take stock before you go of really understanding your degree of readiness, you know, for that. Mike, you know, I know you go to every IMTS with kind of a wish list and like stuff you want to look at and add to your shop. Have you ever had that feeling where you’re like at the show and you’re seeing technology, but, you know, your company is just not ready for it, even though it’s the type of stuff that you’re like going to need and probably would have already implemented if you had that foundation set? Yeah, 100 percent. But I think I also think that’s why it’s so important to go to IMTS.

[20:00] Right. It’s it’s to see what is coming as much as what’s already here to me. It’s there’s there’ll be a lot of stuff at the show I’m not ready for, you know, whether that’s.

You know, AI and software technologies or machines, but in the next two years, I might be ready or I might have a need come up. So I think that’s one of the most important things going to show anyway. It inspires you to get ready at a minimum.

[20:31] Well, I feel like there’s a there’s always this you don’t know what you don’t know. Yeah. You know, and you can read Modern Machine Shop magazine and you can kind of look at things, but it’s entirely different to like literally walk the floor and talk to people actually implementing these and selling these things. You’re like, holy cow, that I didn’t realize that was possible. And yeah, lots of times at the show, too. You know, maybe even from a budgetary standpoint, some of those newer things are out of reach, but a year and a half from now, they’re not. Right. Because price of technology has come down. There’s new demand maybe on on our companies. But, you know, if you if you don’t see those things like the AI arena or automation arena areas and stuff, you don’t even know it’s possible. I think that that’s a good good point, and, you know, I kind of. I have two frameworks that I apply to this one, it was like what I’ve been talking about, the stack starting at like the level of are you able to collect data and is that data in good shape, you know, all the way to the application level in terms of the tools.

[21:36] But I also look at it from, you know, across the part lifecycle, you know, going through sort of design or even earlier than design in some cases of like just an idea for something that you want to make through the making of that thing, inspecting the quality and then actually putting it out in the field and using it. And if you if you lay those things on top of each other, you know, you get some kind of matrix. AI, different kinds of AI is basically coming into every square of that matrix and offering some kind of advantage, advantage of speed, of the ability to do more with less people, of the ability to squeeze some more margin out of things that you’re already doing.

And and that might be a good way to think about it, too, which is even if it is something that you’re not quite ready for, you can be inspired. But because you can say, wow, like this would bring so much advantage to my quoting system, this would really address a pain point that I have. What do I what what gaps do I need to close to be able to put that into practice? Yeah.

And so that’s that’s I think where AI fits. It’s not a specific thing or a feature of a thing. It’s it’s kind of like some capability that’s coming and making a lot of things a little bit better, a little bit faster or or something like that.

[23:01] But it’s happening, you know, across the value chain. You know, when before we hit the record button, we were talking about another category that’s pervasive throughout everywhere you look, and that’s automation. And it’s kind of like the AI arena, you know, it might be challenging to put automation in a sector because you see it everywhere you look when you’re at a show like IMTS. But those those two categories kind of are going to infiltrate every aspect of modern manufacturing if they haven’t already. And what are your thoughts on on that and the challenge you guys have at AMT trying to figure out how to have a whole wing of the show that’s dedicated to technology that’s kind of permeated throughout? Yeah, so we do have an entire sector, right, I think it’s the entire North building pretty much is just dedicated to automation. But the reality is, is there are robots in every sector of the show, pretty much what, you know, in the conversation that we’re having, what’s interesting to me is not everything that we’ve been talking about, right, because there’s a lot of just pure software that is AI enabled that doesn’t really need to be so connected to the actual physical environment.

[24:16] But I think most of what we’re talking about and a lot of what’s going to be most interesting to visitors at IMTS kind of falls into the category of like physical AI, right? And that’s, um, it’s getting a lot of attention in the media these days. It’s kind of the next frontier of AI, right? Yeah, that’s what I’ve been seeing a lot of just in the last, I would even say three or four weeks now, all of a sudden, you know, the talk of physical AI has been explored. Let’s frame that, like what the heck is physical AI for someone who’s like, what does that mean? Is this like, uh, yeah, it’s, uh, I think we should do that. Yeah. Yeah, go ahead. You want me to, you want me to do that? I’m not the guy to do it. So, so AI is only as smart as the data that it has to reason on. When you’re dealing with just large, large language models that have read the entirety of the internet and you’re having a conversation with it, you know, there’s a lot of basis it has for you to have an effective conversation about the history of the New York Knicks or the world cup or, you know, any, anything that you can just kind of chat about. And there’s a lot, and there’s like a lot of industries where there isn’t really this direct connection to the physical world.

[25:36] And so those types of tools that are completely based on digital information, there’s a lot they can do. In our industry or sort of the collection of industries that are going and represented at IMTS, we’re really dealing with the physical world or really kind of going from the digital world to the physical world, back to the digital and kind of ping-ponging back and forth. And what we’re talking about when we’re talking about physical AI is the ability to take information from the physical world and translate it in some way so that the AI can understand it.

Or in the case of like robots and cameras and other sensors, giving AI some tools that it can interpret the physical world in some way that then is going to make it useful. That’s a big deal because there’s not really a lot that manufacturers can do without having this interface with the physical world in some way. Nick, in your case, I like a relative example for you is like what you guys are doing with monitoring the spindle load and tool life and then making a decision and then changing what’s going on in the machine.

[27:03] Yeah. Right. Like you’re, you’re getting that data, you’re putting it into some sort of algorithm that gives you an answer, like to improve things and you’re sending it back to the machine.

That’d be a physical layout. Yeah. It reminds me of kind of like the, you know, human body, how it works.

You’ve got your nervous system and that’s like how you know if something’s hot to move your hand or whatever, right? Like, like you have to have that connection between what’s happening physically through what you’re able to sense, your senses, and then making decisions like, hey, maybe I should stop burning myself. I think that’s a, I think that’s a perfect metaphor for it, actually. And so what’s missing today, like even in like we’re, we’re routinely getting digital information, figuring out some way to, you know, turn that into a physical part, then we’re using tools to then inspect that part and compare it to the digital artifact.

The, the differences I really think, and, and, and when we talk about physical AI, what we’re driving for is that that’s all happening as close to real time as possible. It’s feedback loops. If you’ve had this experience of throwing a ton of data at ChatGPT or Cloud or Gemini or something, it’s astounding at how fast it crunches in and gives something back to you.

[28:26] And then, and then you’re kind of the bottleneck, right? You’re like, oh man, now I got to read all this stuff. And then I got to think about it and make a decision and then I got to give it back. The amount of times I’ve told AI, I’d like explain it to me like I’m an eight year old.

[28:41] Right. And, and, but you’re doing that, right. So that you kind of alleviate the bottleneck, which is like your meat brain.

But, you know, in, in the context of, of what you’re going to see at, at IMTS, particularly in the robotics sector is lots of technologies that are supposed to, you know, either open up the bottleneck so it gets more information about the physical world or reason on information from the physical world in a way that is more useful for all these manufacturing applications. Because, you know, for physical, for us to really unlock physical AI, we don’t want to be the mediator of that information, either copying and pasting that information or shuttling it between programs ourselves, or interpreting it and then getting a new set of instructions. You know, the ideal use cases, you know, you, you get the system running information from the physical world goes automatically to, you know, some AI system.

And then it’s making some adjustment, it’s making the robot do something, it’s making the CNC machine doing something different. And then it gets the feedback from that in as close to real time as possible. So that that loop just continues.

So that’s not the way it is right now. But I think, you know, we’ve, we’ve, we’ve seen this happening over the last several IMTS show cycles, but, you know, the advancement seemed to be coming a lot more quickly. One reason is that the tools are just getting better, the computers are becoming more powerful.

But I think another significant reason is because a lot of the hyperscalers and other big companies are realizing what a tremendous opportunity this space is, they’re really starting to pay attention to it. And I mean, just to one example, it’s an IMTS exhibitor is Intrinsic, which is a Google company that’s in industrial robots. And they’re, they’re really, you know, have a pretty interesting solution to this problem.

Not only how can we program robots faster, tell them to do things faster, but how can that become a learning system, a dynamic system that doesn’t break when, you know, the table moves a few inches or something like that. So good. And so just so cool how you framed that because even in our context, you know, Mike brought up that like tool load monitoring, and it helps you predict when your tools are going to fail.

And the alternative to that is to like, go based off what happened to the other times you did it, collect a bunch of data that’s all theoretical, and then say, okay, based on what happened the last time we ran this job in this tool, here’s what we can expect in the future. Or even sometimes they do it in the software through like simulations. But you just explained to me who’s in this category, like the difference between physical AI and then kind of the alternatives.

And if you have like the loops and the sensors, it actually seems almost easier and more efficient than to try to collect what’s happened in the past and interpret it and then predict the future. Yeah, for sure. Well, that’s why it’s so beneficial.

[31:51] Our part is just getting it, I think, set up in, in the first place, you know, and I think there’s a lot of people that are starting to take advantage of it, like, sounds like you are too, Nick, but, you know, I think the low hanging fruit is relatively easy. And I think that, you know, getting beyond that, there’s really the challenge of explaining to the AI what all this stuff means. And that’s something that, you know, we work on, you know, with MT Connect, for example, that’s, that’s what we call like the semantic layer, you know, can we structure this data in such a way that when we give it over to an AI, AI is like, I understand what this means.

And I can, I can do useful work on it. AI will, if you give it data, it’ll do something with it. But you might not understand it.

And you might, the result might not be what you’re looking for. And as manufacturers, that’s not acceptable. You need to have zero doubt about the quality of the input, and the quality of the output.

So that’s, I think, something that you’ll be hearing more about in the next months or years is, great, robots are getting better, the sensor technology is getting better, the integrations are becoming easier. But what about the data? Is the data good enough? And is the, are the semantics or, you know, maybe think about it as, is the grammar good enough for these things to actually make sense? That’s, I think, going to be a key part of it. And the part where, even as the systems become more connected, without that, there’s going to be lots of humans, you know, having to mediate a lot of those handoffs to make sure that the, what’s going in looks right, what’s coming out looks right.

[33:39] You know, that makes me think about you, Paul, you’ll be in the East Hall, right? The software and technology hall. And I think about companies like yours, where, like, it’s, it’s pretty easy to understand, okay, you have to build on a solid foundation, and then you have to give the appropriate context so that the AI can do its thing. And that we all know that, aha, like, oh, it’s got it, finally gets me.

But with you, you have to rely on your users, the Mike Paines of this world to help set up that foundation for success. How do you steward that? Yeah, I mean, there’s, there’s a lot to it. Having, having the right, you know, the right prompts at the right times, the right ways to collect human, you know, feedback on what they’re working on, how it’s going, problems they’re having.

These are all things that are really meaningful when you’re doing job costing, or looking at the type of work you’re good at, and you want to make decisions about, you know, future things. But we’re also starting to, you know, speaking of physical AI, I guess, you know, we just recently launched our own native machine monitoring. And so with MT Connect is the first protocol.

And so we’re sort of combining what the people are doing with directly what the machines are doing. And then that’ll be feeding into, you know, scheduling algorithms and things that can make sure that you can know exactly what actually happened on the machine, not just what the people said happened, but directly off the machines and which cutting tools lasted how long and, you know, and making sure your schedule adherence is great. So it’s always hard when you strictly rely on humans only.

So it is, it’s fantastic when you can get direct data feeds from a machine about, you know, were there any overrides in place, you know, and, you know, how long did this tool last and, and, and what was the actual cycle time compared to what you quoted directly with what the machine says, not what the people say. So it’s a big puzzle for sure. But it’s a really fun one to figure out because the, the holy grail of, of, you know, what a shop needs to succeed is to have presented to the decision makers, like the most accurate true state of what’s happening on the floor, and how that can impact the decisions they make about the work they take on and the people they employ.

And when you can get that dialed in, it’s a really virtuous, very positive thing. I think it’s really important, maybe this this one concept that that you articulated really well, which is any business, but I think that, you know, manufacturers because of this, the physical digital divide, you know, operating a business requires a sort of a model about how is my business doing? How are all my jobs doing? You know, how is this particular program doing compared to, you know, reality versus my expectation? If we can give AI the information to have a really accurate model, and humans and computers can have the same very accurate model, then it unlocks a lot of value. But if there’s any deviation between, you know, what me as the owner of this business, and this computer system, if we have different deviations in that model, then it leads to a lot of uncertainty, and you’re not really going to trust the output.

And so I think, you know, that’s maybe one way, Paul, that you described it, like the challenge now is like, making sure that the model that both have are accurate and the same. Then I think that, you know, the sky’s the limit in terms of, you know, what you can do because of how fast, you know, these AI systems can make decisions, but you only want to making decisions when they’ve got the right information. Yeah, when it’s congruent with the way that you would want someone to make a decision on your behalf as the leader of the business.

[37:51] You can’t spell high end without DN. If you’re serious about taking your shop to the next level, it might be time to stop thinking in terms of good enough and start thinking in terms of capability, flexibility, and my favorite automation. DN Solutions is helping manufacturers move into true high end machining with their SMX multitasking series, the DVF 5-axis machines, and advanced automation solutions designed for shops that want to run smarter, faster, and more competitively.

The SMX platform gives you the ability to combine multiple operations into one setup, reducing handling, improving accuracy and helping you get complex parts out the door faster. Their DVF 5-axis lineup delivers the precision, rigidity and confidence manufacturers need for today’s demanding aerospace, medical and high performance applications. And when you pair those machines with automation, you’re not just buying equipment, you’re building a more scalable business, high end parts, high end capability, high end thinking, because you can’t spell high end without DN.

Learn more at dnsolutions.com. I had a conversation with Eric Nekkich, who I host the Lights Out podcast with, yesterday about this exact thing. And we basically described like two extremes, a spectrum of two extremes, where like, on one end, it’s, well, the AI said it, the AI gave this result. So like, it’s AI, I mean, it’s got the whole world of information, how could it be wrong? You just trust it, no matter what.

And then on the other end, maybe it was, it showed you some errors, or it didn’t get it right. And you just kind of like write it off. Like, ah, well, AI doesn’t work for us.

I don’t know what all these kids are talking about. And, and I think what’s important, we had we had an example where we were trying to figure out like something related to the close rate of a certain product category in our sales CRM. One of our guys who’s really AI enabled is like, well, it says it’s this.

And Eric’s like, that doesn’t check with reality. I live in this business unit. Like, why is it saying that? It can’t be.

And then we checked it, figured out why the AI was wrong, taught the AI, hey, you’re making this mistake, because you’re pulling from the wrong source, or you’re not, you don’t have the right formula, corrected it. And now it’s back to helping you and giving you the answer right away. And I think that’s why the human element is so, so, so important in all of this.

Because like, we think we know how to run our businesses. I mean, we’re in leadership positions for a reason, everyone who’s listening. Like, you’ve got there for a reason.

You know how a good business should work. The AI is there to just like amplify that, speed up the decision making, help you be a thought partner. And if you’re too far on either side of the spectrum, it’s really dangerous.

Were you just write it off and don’t use it at all? Or were you just blindly trust it? Mm hmm. Yeah, good points. Yeah, love that.

[40:48] Yeah, we, you know, we’re, we’re using it, when we crunch data, we’re using it to basically identify exceptions, right? Like, you know, on any given day, I don’t even know how many jobs are running on any given day, you know, probably 50 or 60. You know, I mean, no different from before we were using any, any AI. 85, 90% of those are running just fine.

Right? It’s, but, but you had to look through all of them to find the ones that weren’t. Right? And, and now, you know, being able to use tools to, to, to say, okay, well, here’s the six you need to pay attention to today. And every once in a while, that list it gives you is wrong.

There’s nothing wrong with one of those or two of those on the list, but okay. So I looked at two extra jobs that I didn’t need to look at. Um, you know, that’s that, I think that’s how we’re handling that spectrum, Nick.

It’s like, yeah, it’s going to give me some bad data, but it’s a whole lot easier to filter through a little bit of bad data than filter through everything and find just perfect data. Yeah. Yeah.

That managing by exception. Really good at finding a needle in a haystack. Yeah.

And then over time, like, let’s say those six jobs that got flagged, you know, I could, I could just see how it evolves into, well, here’s, here’s the why behind it. It’ll, it’ll be, you know, you start plugging in stuff like what Paul’s talking about, where your machines are connected to your ERP and it’ll start to give you, well, I noticed this, this, this, and this. So here’s, here’s the why behind those, which would be the next question you would ask as a leader.

Well, why are they off? So it’ll be like, and then ultimately, like, here’s, these are the ones that are off. Here’s the why. And here’s what we think you should do about it.

[42:34] Right. And now you’re just kind of like choosing from the proposed solutions, which ones you think is best and iterating with the AI back and forth to fix the problem. What you’re, what you’re talking about a lot makes me kind of the terminology we’ve been using internally in AMT is like a learning system or, you know, a reflexive factory.

And this is the way we as, as people, especially those that are in a, that are owners, right? You, you learn from your mistakes. You really want the, your entire digital environment set up in a way that the AI can also learn. Right.

[43:14] That, that it’s like, if, if what Mike’s talking about is it’s helping me identify the mistakes quicker, like a next step would be, is the AI also learning from whatever we, we learned here so that we’re not making the mistakes next. And I think that at IMTS, you might find a couple solutions that help enable that. One that we haven’t talked about yet, but it seems like it’s pretty important is related to workforce.

And, you know, are there sort of AI enabled, AI enabled tools that capture training or, you know, or you’re able to capture knowledge right from the shop floor in some really easy to digest way. And the AI does what it’s really good at, and then translate that into lessons or a knowledge base or, or something like that. That’s, that’s this, this, you know, some way of capturing data from the physical world, which is your personnel’s meat brains again, and collecting that information in some way that it becomes actually accessible to the AI.

But the other is like, be on the lookout for these features as part of quality systems, ERP systems, different robotic systems to really understand, are they just acting in a dumb way? Or, are they leveraging some AI capability to be more reflexive, capture the information or some, some sort of feedback from, from the physical world or from other system, you know, to Mike’s point, showing the exception, and maybe not just stopping there, but then recommending some action or, or acting on that exception in a way that saves operator time or saves owner time. Yeah, for me, I’ve been starting to use like co-work. Okay.

And so Mike and Paul in the middle of the night, we’ll get like emails, Mike, Paul, Jen, the whole team, Dustin. And I’ll be like, Oh my gosh, this is what I was able to do. And I think what, as you were talking, I’m like, what do I get most excited about? It’s that persistent memory that I’m building into it.

Because it was so frustrating before how like, you would go through a whole project, and then like the chats wouldn’t know each other. You know, that you’d have to like, kind of start over. And now I’ve got this like operating system that I’m building.

And the same can be true about factories too. I mean, this is just how I do like my strategic thinking work. But like, you’re building this persistent memory where it’s every time it makes a mistake, you correct it.

And then you say, write that to your memory. It’ll, it won’t make that mistake again. And like, isn’t that what we all want with our own businesses? Like, Hey, we learned this, never do it that way again.

Or we learned that this works really well, always do it that way. And so this persistent memory is what’s so exciting. I want to see that translate outside of my computer and laptop and into the factory floor.

And I think that’s what you’re going to show, probably at the Emerging Technology Center, I would imagine. Well, I’ll just mention that. I think in this conversation, we’ve expanded, I think, in some way, the definition of physical AI beyond what I think most people are talking about when they’re really just talking about robots, more capable robots, robots that have more sensory capability and reasoning capability than, you know, what they have right now or what they’ve had traditionally.

[46:40] But I think that, you know, especially in the context of like, a machine shop, like there’s a lot of things that we’re talking about that extend the reach of AI systems into the physical world that aren’t just robots. And that’s where, you know, certainly, it’s going to be very interesting, very dynamic, what’s going on in the automation sector. But the reality is, is that I think every sector, whether it’s like in the actual manufacturing equipment itself, it’s going to have more more sensors and ability to interface with these AI systems.

But on the software side, too, it’s just going to be more built to do what what you’re describing, Nick, about how to create more of this memory or knowledge base of how this factory should run, you know, participate in that in a really, in a really real way. You segued to the ETC. Is that what we’re going to be showing in the ETC? Yeah, I think in some respect, we are, we’re going to show, you know, high degree of connectivity between what’s happening right there in front of your eyes, and then showing how we’re capturing that information and, and being able to create a digital shadow of it, record that data and be able to reason on it later.

[47:59] What we’re showing in the ETC this time around, this is a collaboration with the Oak Ridge National Lab, which has been a key collaborator on the ETC. For the last several IMTSs, we’re really showing a system. In this case, it’s a system, a mostly automated system to create drones.

[48:21] And we’re going to have three different models of quadrocopter drone that we’re going to be making right there, making and flying right there inside of IMTS over the week of the show. It is cool. It’s a hybrid manufacturing approach.

There’s going to be some additive manufacturing component, five axis subtractive CNC. There’s a Mazak system that actually has got two five axis CNC mills, one that has been modified to be a metal additive system. And they’re sort of passing molds back and forth to one another.

Then there’s an automated cell that is sort of taking the parts of the drone and putting them together. And, you know, out pops completed drone bodies with the electronics all embedded inside. I don’t want to give too much away because like you just have to see it.

[49:16] It doesn’t it doesn’t sound like it’s making any sense, but it’s making me want to go see it. Yeah, absolutely. We’re targeting making drones every couple minutes and flying them right there in the show floor.

But making a drone isn’t really the cool part. The cool part about what we’re showing in the ETC is we have the ability to change the design on a dime without missing a beat in the production process. So, you know, our target is about for a minute, which, you know, translates to about 10,000 units a month, which is a requirement of the DOW’s drone dominance program.

So that’s kind of what we’re benchmarking against. But what’s unique about this system is you can you can take the mold, you can put it back into the Mazak cell, and you can change the design and then, you know, just pop that right back into production. So the idea that you can do low mix high volume and then immediately ramp back up to high volume in the same system is something that we think is pretty unique.

[50:25] So if it’s 10,000 a month, you’re going to be there for what, five, six days. How many are you going to get done at IMTS? We’re targeting 600 across the three different units. Are you giving those away or what’s going to happen with those? Yeah, so it’s a good, it’s a good point.

The electronics themselves are kind of the expensive part. So we’re going to be recycling some of the electronic units, but we’re going to have a lot of drone frames that, yeah, I think that there’s a possibility that if someone wanted them, they could take it away and buy their own electronics for them. I think you should have like one of those drone firework shows at night with all the drones that you made.

That’s what you should do. I think they should go fly drones over the rock crawlers in Al’s section. That would be so cool.

Do you know about the rock crawlers? Have you heard about this? No, no. What are we talking about here? Our buddy Al Whatmo, he’s, he’s putting together like a whole track, like a whole… Community park. Yeah.

We, you know, those little like remote control RC mini monster trucks. Yeah. So that’ll be another fun thing to do at IMTS.

[51:30] I know, I know Al. I’ll talk to him about that. It would be a cool way to introduce it.

But yeah, I would say that, you know, the, the, the ETC are going to be showcasing a surge capacity, you know, that can be applied to maybe it’s not just drones. Right? I mean, I think you’ll, you’ll see this manufacturing process and be like, oh, there’s a lot of things that you could, you could sort of make this way. But it really combines a lot of the thinking about like, what are the strengths of additive? You know, what are the strengths of compression molding? What are the strengths of CNC? Combining those in a novel way.

But I think the cool factor is that we’re effectively flying drones right off the assembly line. That is so cool. Which I think will be, be cool to see.

[52:15] I mean, that is like what you imagine in some futuristic movie about like, you know, in the year or whatever, we’re preparing for a war and you just see like this, all these drones flying out of a factory going to work right away. Very cool. I think, yeah, I think, I think the concept that we’re working for now is that they’re going to fly themselves like basically into like, you know, the shipping box or like whatever, like, you know, the last mile delivery, they’re going to sort of fly themselves.

So. That is so, so, so cool. It’s like a Tesla car coming off the factory line and it’s driving itself straight to a dealership or straight to someone’s house.

[52:50] Yeah, right. Yeah, that’s, that’s kind of the thing that we’re, that’s our, that’s our goal. That’s what my wife and I are working on.

Like, we just keep her pregnant and then she has a baby and they just go right into the family business. They’re sweeping floors at six months. As soon as they can walk, they’re sweeping floors.

[53:13] Yeah. Gosh, Ryan, this has been such a cool episode because like, we’re, we’re even more excited about the show now after you’ve given our meat brains some vegetables. Yeah, definitely.

No, it’s gonna, IMTS has got, it’s still mostly for meat brains. Well, I’m one of them. And it’s like barely more than two months away now.

It is coming fast. Yeah, I don’t, I think, I think like nine, nine and a half weeks or something. I was just looking at my, my paper calendar right here.

So yeah. Yep. September, September 14th.

I guess I’ll see you all there. If people are not registered yet, where do they, where do they go registering, sign up to come, come to this place? You go to IMTS.com. That’s easy. Not AIMTS.com. Did you get that domain too? We should have, we should have probably done that.

You can probably get applauded. I’m going to go buy it. I already own it.

[54:11] That’s good. I’m going to auction it back to you. I grabbed it like when you said it earlier. This domain is for sale for $6,000. That’s right. So I’ll charge you seven, Ryan. Luckily that’s not, that’s not my, that’s not my department, so. All right. That’s awesome.

[54:32] Well, thanks. Yeah. Thanks again, Ryan.

This is a, yeah, this has been a really fun conversation. Always great chatting with you guys. And yeah, I think, yeah, I think it might be worth revisiting this conversation and talking about the AI.

[54:47] How many sightings of AI we saw and whether they were the real deal or marketing fluff. But the reality is, is I think it’s, we’re getting closer all the time to these, you know, reflexive factories and it’ll be cool to see. I think we’ve got over 1,800 exhibitors at IMTS at this point, you know, to really see how all of them are pushing the envelope along this line. Well, Nick, Nick’s going to play the AI drinking game. So as he walks around IMTS, every time someone mentions AI, he’s going to have a drink. So you’re going to have to get a humanoid to keep you upright. Yeah, I’m going to need a humanoid for IMTS for sure. And if, yeah. And if you want to come see Nick being held up by a humanoid, we’re going to, we’re going to be there all week. We’ll actually be in the Hennig booth. You know your booth number or anything? No, we’re probably not. No, I just know we’re in the South Hall. It’ll be easy to find us. We’re right near like the DN Solutions booth. And so, yeah, it’ll be really easy to find us. We’ll have a live studio recording. It’s a 339-133 everybody. Just lock that into the persistent memory and show up and look at me and Mike and Paul record at the IMTS.

[56:05] And if you haven’t already, of course, visit IMTS.com and register. And we’ll see you there. Because if you don’t go to IMTS, you’re not making chips. And if you’re not making chips, you’re not making money. Bam. Bam.

This transcript was auto-generated and may contain errors.

Connect with Ryan Kelly

Connect with MakingChips

Subscribe

For the latest episode straight to your inbox, business-building tips and other exclusive content, Subscribe to #MetalworkingNation today.