The TechEd Podcast

How to Get Started with AI in Your Business (Practical Tips & Real Technologies) - Part 1

• Matt Kirchner • Episode 239

Share episode

What does it really take to get started with artificial intelligence in a small or mid-sized company right here in the U.S.?

We're breaking it down in this two-part series.

In part 1, Matt Kirchner shares lessons from his recent trip overseas and what he learned visiting 26 advanced tech companies in six days. From open-source innovation and mandatory AI education to the work ethic driving global competition, Matt explains why the time to act on AI is now, and how American business leaders can take practical steps to stay ahead.

He connects global insights to the realities of U.S. manufacturing and education, explores what it means to see before others see in the age of AI, and outlines the first practical technologies every organization should understand, from AI agents and MCP servers to embedded smart technology and digital twins.

In this episode:

  • What China’s open-source approach to AI is teaching the world about speed and innovation
  • Why small and mid-sized U.S. businesses can’t afford to wait on AI adoption
  • The two traits every leader needs to thrive in the AI era and how to apply them today
  • How manufacturers are already using AI for predictive maintenance, analytics, and smart equipment
  • The real-world technologies, like MCP servers, AI agents, and digital twins, that can start transforming your operations now

Including...the first 5 technologies from A Manufacturer’s Guide To AI Tech.

FULL SHOW NOTES (plus links & resources): https://techedpodcast.com/appliedai

Want to see all the videos and data? Watch this episode on YouTube.

We want to hear from you! Send us a text.

Instagram - Facebook - YouTube - TikTok - Twitter - LinkedIn

TechEd Podcast Introduction:

This is the TechEd podcast, where we feature leaders who are shaping, innovating and disrupting technical education and the workforce. These are the stories of organizations leading the charge to change education, to rethink the workforce and to embrace emerging technology. You'll find us here every Tuesday on our mission to secure the American Dream for the next generation of STEM and workforce talent.

Melissa Martin:

Hey, TechEd podcast fans, it's your producer, Melissa Martin. I am so excited for the episode we have in store for you today. It's all about artificial intelligence and the data and the stories and information we have to share with you is stuff you probably haven't heard before, from real examples, from how AI is being used on the other side of the world to practical strategies you can use in your business today, Matt has so much in store for you on this week's episode. In fact, there's so much to share that we turn this into a two part series. So you're catching part one today, and next week we'll air part two. The other reason I'm hopping on here with you is that this episode is also available as a video on our YouTube channel. So if you're a visual learner, if you want to see the charts, the graphs, the videos and the pictures behind all of these stories that Matt is going to share head on over to our YouTube channel and watch this episode there. In fact, you'll be seeing more of us in the coming months. As you know, our podcast streams on Apple Spotify, 43 other podcast platforms, basically anywhere you can get a podcast, but we're also on YouTube, and in the coming months, you're going to see more videos from us on YouTube. So if that's how you like to consume your content, and if you'd like to watch more of our podcasts, you can do that today. Head on over to youtube.com/at. TechEd, podcast. That's youtube.com/the at sign TechEd podcast. And now here's Matt,

Matt Kirchner:

welcome into the TechEd podcast. It's Matt Kirkner, and it is just me. Just me this week, talking about artificial intelligence and more specifically, talking about AI in manufacturing, and let's be even more specific than that, talking about AI in small to mid size manufacturing companies. You know, I still spent so much time around manufacturing. I'm on the board of several manufacturing companies, investor in several manufacturing companies. Still love manufacturing, where I spent the majority of my career. And one of the things I hear so often from our small to mid size manufacturers is I know I need to be doing something in AI. I know it can help me improve my business. I know that there are problems in my organization that can be addressed using artificial intelligence, but I'll be honest with you, Matt, I just don't have anywhere. I just don't have any idea where to start today we're going to answer that question. Today. We're going to talk about, if I'm in a small to mid size manufacturing company, or, for that matter, a small to mid size company in general, or maybe a small to mid size educational institution. Doesn't matter. We're going to talk about some really practical ways to apply artificial intelligence in our organization. I want to start, and if you're you're checking out the video, you will see that we have a picture of Shanghai, China. And the reason I'm starting with that picture is that I just, in the month of August of 2025 spent a week in Shanghai, and actually a week in China in general, Shenzhen, Guangzhou, Wanzhou, with an H and then Shenzhen, or Shanghai, should say, to finish off the week. And I learned a ton. And it really was an eye opening trip for me, and I want to share a little bit about what I learned on that trip. The first thing is that China is way more entrepreneurial than I expected it to be, way more profit driven. The free enterprise system, in many ways, is alive and well in China. I mean, we saw companies that were private equity backed, we saw companies that were venture capital backed, and not just venture capital from the Chinese government, but in many cases, Chinese venture capital from private investors. And also, believe it or not, venture capital from the United States that was coming from private investors here in the US super, super entrepreneurial and profit driven. Number two, and this is really important, the code for AI in China is all open source. And so what we mean by that is that here in the US, if I'm open, AI, if I am meta, AI, if I'm if I'm a US based AI innovator and AI company, I'm keeping all of my my code, all of my programming, all of my algorithms, all of my know how really close to the vest. I'm not sharing that outside of my company. I'm not sharing that outside of my organization. I've got that all locked down. That's my secret sauce. That is my IP in China, it is exactly the opposite. And in China, all of the code is open source. Once you innovate something from a coding. Side, from a software side, you know, that's available on GitHub to the whole world in a month or two afterwards, and so, so it's all open source. It's a much different model. And what, what that's doing is it's driving the innovation in China, not just the coding side, not just the programming side of artificial intelligence, but what we call physical AI, or the physical manifestation of artificial intelligence. In other words, how are we utilizing artificial intelligence on physical assets, like humanoid robots, like manufacturing equipment, like the Alexa speaker that some folks have in their in their kitchen. So how are we taking AI and then manifesting itself in its physical form? And that is really where the innovation is happening in China, and it is happening fast. We were at a humanoid robot, robot company that a number of my colleagues, a number of the folks I was traveling with, had been at literally, like three months before. And they said the rate of innovation in that company in those three months that it ensued since their last visit, blew them away. The management teams, the leadership teams in China all really young. It feels, felt like every one of them was like late 20s, early 30s that were running these tech companies. And by the way, I visited 26 tech companies in six days when I was in China. So we saw a lot of different technology and was there with some folks that kind of had a back door to a lot of these organizations. And we got to see some things that, you know, if I was just over there on a general business trip, I probably would have had a hard time seeing really, really insightful. But one of the insightful things was how young the management teams and the leadership teams were in these in these companies, a lot of them, in fact, I would say almost all of them, it seemed, were educated here in the United States, and either because they chose to, or because we told them they had to, they went back to China, and they're innovating there, but super young leadership teams, incredible work ethic. You know, we were at a humanoid robot company in Shenzhen, eight o'clock at night, so we had some long days, but 8pm at night, and here we are in this facility, and it's full of engineers. It's full of people who are working so 8pm mechanical engineers, controls engineers, software engineers. I mean, that company was alive with people working at 8pm spent some time at what we call the chatgpt of China company called Miramax, and that was in Shanghai. As a matter of fact, cots were lining the walls in that company. People were sleeping at work. People were working until they got tired, and then they then they would take a nap. You know, we were there at 230 in the afternoon, and the and the facility was still two thirds full at 230 in the afternoon on a Saturday, by the way. So that's the point of that. Of course, during the week you would expect that, not necessarily on a Saturday, really, really great work ethic to their credit. In China. I thought I knew how Amazon worked. I thought I, you know, there's like, some guy in a facility in LA who was, you know, putting some product up online, and then I would find it when I was searching for something, and it would land in my in my feed, on Amazon, and then I would look at the ratings and look at the price, and I liked it and I bought it. Not exactly, we were at a complex a cluster of E commerce companies. It was three buildings. Every one of the buildings was, you know, by my recollection, at least 70 stories tall. I think at least one of them was 90 stories tall. Within those three buildings, 1000 e commerce companies were working like crazy to reach the American market. And what they do is they have these e commerce merchandise managers, people that are managing maybe a series of 15 products, and they are battling with each other to make those products show up as high in whatever the marketplace is, whether it's Amazon, Wayfair, temu, Home Depot, all of these different companies that have marketplaces, and so they are competing with each other to show up as high as possible on those marketplaces. They were changing out backgrounds on their products, changing out the layouts, changing prices in real time. We even saw one of them that showed us a program, basically an AI agent that they had built that goes, get this to their competitor's website, scrapes the data from their competitor's website. And while it's scraping the data, pricing, data, product descriptions and so on, it is simultaneously burning their competitors ad spend and lowering their competitors conversion rate, meaning lowering the number of companies that were buying the product from them, from their competitor. And obviously, when that conversion rate goes down, so too does the frequency with which that product will show up on the marketplace is absolutely, absolutely fascinating. We visited a company called info mind. What they were doing was just blew me away. They basically, I'm not going to explain this perfectly, but they've kind of hacked into the algorithm of a platform like Tiktok, and they claim that they understand how to maximize the appearance of a company's products on a platform like Tiktok. You can take, I think they said it was like, take 10 videos, and they could turn that into like, 1000 different videos. So if they have 10 videos from from somebody like us, so you know, somebody that's trying to advertise on Tiktok. Um, they can chop that up in such a way that tick tock won't recognize that it's seeing the same video over and over and over again. And so it keeps coming up in your feed. If you, if you see a product, and you spend a little bit of extra time on your on your tick tock feed, on that particular product, it's, it's going to keep coming up. We know that. That's how tick tock works. That's how YouTube shorts works. There's a number of other social media platforms that work exactly the same way, and they have really tapped into that in a way that they can put tremendous amounts of content out on us social media and get us to convert into purchasing products from them by going to, for instance, Tiktok shop. Another really interesting insight that I looked at, and that I noticed when I was in China is that every single company had an education strategy. So a lot of companies would have their products for the commercial marketplace, and then they would have their products for education. They all had a separate product line or a separate focus on education. And here in the US, I'll just be honest, a lot of times we have a hard time getting employers to recognize the importance that they need to have a separate strategy for education, that they need to be engaging in education. But in China, it's, it's, it's endemic. In every single company, they're recognizing that they need to bring up the next level and the next layer of talent. The other thing that was interesting is we were told that artificial intelligence education, AI education, is mandatory in K 12 in China. And here again, I you know, there's a lot of great things going on here in the US, and K 12 starting to see more and more things on on the AI front but, but we're like evangelists trying to convince schools that they need to be doing this. And in China, they've already figured out how important it is, and that it's literally mandatory for K 12 students in China to have aI education. That's one of the reasons, by the way, that I'm such a huge advocate for the Discover AI platform that is getting tremendous traction here in the United States, where students are going through a 16 hour e learning course, and in learning about artificial intelligence and applied AI and what we call the edge to cloud continuum. And then they go through a series of experiences, and they're, you know, each of these experiences is call it, you know, 45 hours long, give or take. And it's experiences like drones and unmanned ground vehicles, or ground drones or autonomous vehicles, like like, you know, like a self driving car that you might see in a city like Phoenix, or they're learning about 3d design and fabrication, or industry 4.0 and and smart sensors and smart devices and manufacturing and smart energy. I mean, there's just so many different ways that we can teach students how AI manifests itself in the edge to cloud continuum. That's what's happening with Discover AI. You know, a very quick note that that I do have a financial interest in that, in that project, so I want to make sure that we're clear about that, in full transparency. But it's really, really gaining traction. We need to continue to do that. China's already made that kind of education mandatory. We're just figuring that out here in the United States of America. Final thing, going back to that final observation on China, I should say, going back to that example of spending time in Shenzhen at that facility, that was the E commerce facility, with 1000 companies, and by the way, 10s of 1000s. They said 100,000 employees. That was hard for me to believe, but easily, 10s upon 10s of 1000s of people working in that E commerce facility in that cluster, I asked one of the one of the one of the CEOs of one of those e commerce companies. I said, How do you train all these merchandise managers? He said, Oh, that we do that at the cluster. We all do that together. We have a common training program. Even some of these companies that were competitive, right? I mean, we saw companies that were selling very similar products into exactly the same market, competing with each other in the marketplace, but collaborating in the classroom and understanding that they were they all benefited from a rising tide, lifting all ships and investing in education collectively, really, really interesting, and something that, quite honestly, we don't see quite as often as we probably should here in the United States of America. So the reason I bring all that up as we start a conversation about AI and small companies is this is what we're up against, guys. You know, I'm a lifelong American. I am a huge fan of the free enterprise system. I am a huge fan of freedom. Mindy would call me a patriot. I certainly would refer to myself that way. I'm not terrified, but I'm they've got my attention. I mean, the things that we saw when we were when we were in China, we've got our work cut up for us here in the United States. You know, all is not lost. There's still we're still leading in a lot of technological areas. There's still a lot of reason for optimism, but it's not going to happen on autopilot. On autopilot. And you know, our economic success here in this country is not a birthright, and it is something that you know, that we earn every day, that that the previous generations earned for my generation, that I certainly hope to and strive to earn for my generation and generations to come. But that doesn't happen on autopilot. It happens when you pay attention to reality and you innovate, and it's time for us here in the United States to innovate. All right, so let's get into this AI concept a little bit. I had a great experience last August. I had an opportunity to meet an incredible author. His name is John C Maxwell. If you're not familiar with with John Max. Well, he has written over 70 books on the topic of leadership, several of them New York Times, number one, best selling books on the topic of leadership, servant leadership, ethical leadership, just, just a phenomenal author. And I had the opportunity to sit in the same room with John and hear him speak small room in western Wisconsin, and then, following dinner, almost by happenstance, sat down next to him just to introduce myself, and we talked for a half an hour. So here I am sitting next to this iconic leadership author, and we just, we just chatted and had a wonderful conversation. He wanted to know about how to how to apply artificial intelligence in his organization, which was really, really cool. And I just kind of took in his leadership, his sage leadership advice. Here's one of the things he said that evening that has stuck with me ever since. He said, you know, in the past, all leaders, you know, see more than others see. In other words, you know, if you're a leader of your organization, if you've risen in the ranks in an organization, it's because you could see more. Either you saw more opportunity, or you saw a bigger part of the business, or you saw how pieces of the business fit together, but you recognize things in that organization that others did not, and over the course of time that provided an opportunity for you to step into a leadership position and be on a leadership journey. He said, Look, that was great in the past, but he said in the year 2025 and going forward, he said, you know, from now on, it's no longer enough to see more than others see going forward. We have to be able to see, and here's what he said, we have to be able to see before others see. We need to be able to look into the future. We need to see what's coming at us and get there before anybody else does. I think, you know, I'll just be honest. That's one of the things that I've really worked hard to do. And I think had some success in working with with other leaders and organizations, had some success in being able to do exactly that, but, but we have to be able to, in our leadership positions, see now, ahead of the curve, and see what's coming at us. What we're talking about for the rest of this episode is really seeing and looking at what's coming at us in the field of artificial intelligence. So I want to introduce the the audience, at least in concept, to a great friend of mine. He's somebody that's a previous guest on the on the podcast. We'll link his episode up in the show notes. His name is Leo Reddy and and Leo is just a terrific, terrific friend. He ran the NATO desk at the State Department for several presidents here in the United States of America. And many people will credit the work that Leo did, along with several colleagues and under his leadership, with creating a platform to end the Cold War back in the late 80s, early 90s, and the work that he did was called the Helsinki Process. He was the architect of the Helsinki Accords, which many people believe led to detente, which was kind of a sharing of information and knowledge across political divides that laid the groundwork many believe for the end of the Cold War. Leo is still with us, and he's still a great friend, and he's in his early 90s. Now, I get to see Leo several times a year, and I love telling the story. You know, I visited Leo and his wonderful wife, Penny. They over the course of the summer, they spent time in a beautiful home overlooking Green Bay, not the city, but the water, the bay itself in what we call Door County, which is kind of the let's call it the Cape Cod of the Midwest, it's absolutely beautiful, great vacation place. And I get up there several times a year. So I was paying Leo a visit, and we were sitting out on his beautiful deck overlooking Green Bay and just getting caught up. And I and I said to Leo, I said, Leo, what's what's keeping you busy? And he said, You know what, Matt? He said, I am so fascinated by artificial intelligence I can't get enough of it. I'm reading everything I can about it. He's still, he's still working in his business. And he said, I'm trying to understand how, how it's going to affect our business, and how it's going to change our business. And I just tell people, you know, if you look at what's going on in China, where k 12 education is mandating AI as part of the curriculum, and then you look at somebody like Leo, and if China tells us that it's never too early to learn about artificial intelligence, you know, Leo, and my good friend Leo already shows us that it's never too late. You know, here at 92 years old, he's as fascinated as ever by AI. And, you know, I just love people. And as I get a little bit deeper into my career, and you look at some folks that are still innovating later in their careers, you know, there's a great article in the Wall Street Journal last February about an investor who's betting on people in their 50s and 60s. You know they're saying basically the concept being they're older, they're more experienced, they're in a better position to be able to gain our confidence as investors, because they're just maybe a little bit more dependable, or they know what's coming at them. And then there was another article that ran in early October of 2025 that said the new age of entrepreneurship. 70 to 79 it says septenarians, those in their 70s, of course, starting new businesses, leveraging technology experience and decades worth of contacts. A great article, by the way, by author Claire Hansberry. Really, really enjoyed that article, the one on the 50s and 60s, by Ben Cohen. I read both of them quite frequently in the Wall Street Journal. So. Here we are in an age, and I've got so many friends now, as I get, you know, deeper into my career, and I've got some friends that have retired, as you know, I'm never going to retire, but, but there's people my age that are looking to All right, you know, in the next 510, 15 years, that time is my life is going to come along. I can't tell you how many people have said, you know, I'm so glad I'm late in my career with all this technology and all this change, I just don't know how I'd manage it, and thank goodness. I'm not in my early 20s. And I'm like, Are you kidding me? Like, this is the best time in the history of humanity to be in business with all this change, with all this technology, with all this opportunity, I'm having so much fun, and I just, I just can't get over the fact that we live in this incredible time. And why not make absolutely everything of it point being, I don't care if you're 20 years old. I don't care if you're 80 or 90, or like my friend Leo, 92 years old. AI is going to change your life. It's going to change business. Let's get on top of this. Let's understand it, and let's enjoy the ride, because it is going to be a great ride. We really only need two things, by the way, two things to be successful in the age of artificial intelligence. Now this isn't original to Matt Kirkner, I'm stealing it from my friend Barbara humpton, who is the CEO for the time being, although has recently announced a career transition, the CEO of Siemens, 46,000 employees, $20 billion company, join me on this. On this podcast is very podcast for a great episode in 2024 we'll link that one up as well. Check them out. Check it out in the show notes. And here's what Barbara said, you know? She said, all you need in this age is two things, curiosity and initiative. And she said, if you have curiosity and initiative, the world is yours. That's true. If you're a 14 year old kid, that is true if you're an 8292 year old man or woman, the world is yours. With curiosity and initiative, hang on to it. I'm curious about AI. I'm taking initiative in AI in our businesses. We'll talk about that in a little bit, but those are the only two things we need. And by the way, Barbara humpton also told me that this is the greatest time in the world. She shares my opinion, greatest time in the world to be in American manufacturing with all the changes in technology. She said, this is the opportunity for our young people to get in on the ground floor, get in on the ground floor of a manufacturing Renaissance. And you know, you just look at some data, data that I was looking at just last year, sorry, just last month, at the construction spending here in the United States for new manufacturing facilities is, as it's at an all time high. I mean, it is orders of magnitude higher than it has ever been in the past. And so you look at the opportunities that are going to be presented as a result of this huge growth in manufacturing, infrastructure spending, greatest time in the world to be in manufacturing. Okay, take a deep breath. I'm going to ask the audience a question, and that question is this, if it takes five machines five minutes to make five widgets, okay, think about that. If it takes five machines five minutes to make five widgets. How many minutes will it take 100 machines to make 100 widgets? The obvious answer, of course, is, let's go through that one more time. If it takes five machines five minutes to make five widgets, how many minutes will it take 100 machines to make 100 widgets? And the answer is not actually 100 although it seems like it would be. The answer is five. And if we go back to our Eli Goldratt, if we go back to the theory of constraints, if we understand manufacturing throughput, the truth is, if it takes five five machines, five minutes to make five widgets, it will take 100 machines five minutes to make 100 widgets. The reason I bring that up, the reason so many people get that wrong, is that we go through life seeing the same problems, the same challenges, the same production lines, the same people every single day, and thinking that we need to solve them exactly the same way. Artificial Intelligence, by the way, is all about solving problems in businesses, right? If we're just implementing AI for the sake of implementing AI, we are we have a solution in search of a problem, right? What we should be doing is identifying those problems in our organizations, in our manufacturing businesses, where AI might be able to help solve them. How do we improve throughput? How do we improve yield? How do we improve employee experience or customer communication or lead time? So you fill in the blank. How do we do that? And can AI be a tool that is useful in doing that, and that's really what we're chatting about here when we think about artificial intelligence in business. And so let's go through and let's talk about a number of different technologies. As we slow our minds down a little bit, as we recognize that we can't just jump to conclusions answer a question, like five machines in five minutes, just off the top of our head, and we have to think through exactly what we're looking at and how we solve problems. And we're going to talk about several technologies that I think are totally transforming manufacturing, and we'll continue to do so. Let's call it a manufacturer's guide to AI tech really quickly the top the topics in the content. And if you're in manufacturing and you're like, where do I start with my AI journey. That question is going to be answered for you by the time we're done here. We're going to talk about some complicated things, like AI agents and MCP servers. And then we're going to get to the easy stuff, embedded smart technology, digital twins, vector databases, generative pre trained transformers, commonly called gpts, intelligent data prediction. Autonomous mobile robots, smart drones and manufacturing, AI powered industrial robots, next gen, AI metrology and smart materials. Are you ready? You ready for us to spend a little time together talking about AI in manufacturing? Number one, AI agents and MCP servers, building your own manufacturing? AI engine. So let's chat about what that concept looks like. I love the game of baseball. As I record this, we are in the middle of the MLB playoffs here in 2025 huge fan of baseball. I want us to imagine for the moment that you are a professional baseball player about to negotiate your first contract, and think for a minute about all the things that would be important to you in negotiating that contract. You're going to sign your first contract, it could be for just a ton of money, more money than you've ever imagined. What are those things that you want to make sure you're considering when you're signing that contract? And if you make a list, as I did, it's going to be things like the length of the contract, is it guaranteed? Are you getting endorsements through companies, whether they're local or whether they are national? How desirable is the city in which you're planning to live? When are you going to start? What does the coaching staff look like? What are their training facilities look like? Do they have a past record? Is there hope for the future in terms of winning and losing? Those are all the kinds of things that we're going to be thinking of when we think about optimizing a Major League Baseball contract, right in the world of AI. What we would call that combination of factors that we're going to consider as we step into our contract, what we would call those is our knowledge graph, right? So it's an interconnected series of different topics and different things, some of them more important to us than others. All of those things collectively, though, are what are going to define whether or not we're going to like our contract and like for whatever period of time we're signing that contract for, like the world that we're living in, at least as far as Major League Baseball. So now we have to prioritize all that stuff. Let's say we're in manufacturing. What do we use when we have a whole bunch of data, and we need to put a list together, and we need to manipulate that data, and we need to prioritize that data, and we need to analyze that data. Where do we put the data? And if you've spent any time at all in manufacturing, as I did for 25 years, running companies and another 10 now, serving them and investing in them, you know that all that data goes into a spreadsheet. Right where would we be in manufacturing without Microsoft Excel? And then once we have that data in Excel, we need to give that to somebody who is then going to negotiate our contract on our behalf and make sure that all those things, contract length, endorsement, training facilities, win, loss, record, all those things that are important to us, obviously, money being pretty close, probably for many, to the top of the list. What's the actual compensation we need somebody that's going to go negotiate that for us? Who do we hire to do that in Major League Baseball? Right? If you just pay any attention at all the professional sports, you know that that individual is an agent. So now we've got an agent. We've got somebody that's going to negotiate the deal on our behalf. All right, that's how we negotiate a Major League Baseball contract. Now let's flip that into I want to improve productivity in a manufacturing facility. So how am I going to go about optimizing production in manufacturing in the age of artificial intelligence, if I need to kind of use the same logic that we just used in optimizing our major league baseball contract. Well, we'd start by making another list. What are all those things that are important in optimizing production? So in a baseball contract, salary guarantee, training facility, coaching staff, production, what are we looking at? Quality, cost, lot, control, supply chain, workforce, production schedules, gross profit, inventory, all of these different maybe some environmental factors, all of these different criteria and all these different factors that are going to go into making production in our manufacturing operation run at optimum productivity. So we're talking about things like efficiency. We're talking about things like yield. We're talking about things like maximizing uptime on machines, all of those things. How do we go about doing that now? Well, when we were a major league baseball player, we had a knowledge graph, right? And we created that knowledge graph and we put all of our data together. Well, we're going to do the same thing in a manufacturing facility. We're going to gather and locate and identify all the data that we think we need to analyze in order to optimize production. All right, that list of data is going to be really, really complicated and really, really big. It's going to come from all kinds of disparate spreadsheets we're using across the business. It's going to come from our ERP system or our MRP system. It's going to come from our financial accounting system. It might come from our customer relationship management system. We're going to have all of this data that is coming at us. Might even come from software that's being run on different pieces of equipment, different machines. That is going to be a ton of data. That is going to be way more data, by the way, than we can analyze in a simple spreadsheet. What do we need to do that? Then what we use is what's called an MCP server, right? So we're gonna model context protocol server. And now the what that does is kind of outside the scope of this podcast. We could spend a whole hour, a couple hours, talking about an MCP server. Look it up. You know, there's all kinds of great training online to kind of familiarize you with that concept. Write those three letters down, m, c, p server that is going to be the aggregator for all of our data. That is where we are going to pull all of this district, disparate data and create our large language model, this whole combination of all this connected, interconnected and disconnected data that we're using in manufacturing to optimize our manufacturing operations. We're going to aggregate that at our MCP server. Now we need in our analogy about the baseball contract, where we were analyzing all the factors that would make a successful contract, we were putting it in a spreadsheet and we hired, remember who I we hired to do our our analysis and negotiate our contract? Was our agent? Well, guess what? We're going to do exactly the same thing in manufacturing. We are going to hire an agent to analyze all this data on our behalf and help us optimize production, although in this case, it's not going to be a baseball agent with a with a suit on and maybe a bunch of jerseys in his office. This is going to be what we call an AI agent, or a digital employee. When you hear people talk about agentic AI or you hear people talk about using AI agents, and sometimes people use words like that because they want to confuse us, or because it makes them feel important. It's really simple. It's just a digital employee. You're creating a digital agent that is going to go and do a process for you, in this case, go through all of our data so that we can figure out how we can optimize production by reviewing all the data from our Knowledge Graph, things like cost, inventory, production, scheduling, customer, revenue, all these other things that have been aggregated by our MCP server. Now, if that sounds complicated, the reason I start with that is because it actually gets easier from here. And a lot of times when we think about man, I got to start my AI journey. And I don't know anything about MCP servers or knowledge graphs or AI agents. I get it. I can tell you it's important for you to familiarize yourself with that sooner than later in several of our companies, and I spend most of my time now in the distribution space. We're using all of those things. And by the way, these aren't gigantic companies, right? They're small, mid sized companies. We have certain companies of 25 people, where we've got two or three people just working on the data side, believe it or not, so that's kind of the idea is that we don't need to be a billion dollar company to do some of this stuff, even at a really high level. But at the same time, if you're not in a position to hire a couple of data scientists, if you don't have people with that kind of talent on in your business, if you don't want to pay a qualified consultant, and there's a lot of unqualified ones, so choose carefully. But you don't want to pick a qualified consultant to do this work for you, that's okay. There's a lot of other ways that we can deploy AI in small to mid size manufacturing operations. Let's keep going, because for now, it doesn't need to be all that complicated. So let's go to number two, embedded smart technology, as folks know, if you pay any attention to this podcast, we're huge fans of FANUC robotics, largest robotics company in the world, largest CNC company in the world, as well. Mike Chico, President and CEO, four time guest here on the TechEd podcast and and just wonderful and wonderful people at FANUC. And I'm a huge advocate, and I've been to their facility in Rochester Hills, Michigan, more times than I can count. Same thing for their facility in Hoffman estates, Illinois, and I've had the honor of traveling to Japan and visiting their facility outside Tokyo several times as well. But if you buy a Fanuc robot today, collaborative robot, a six axis industrial robot, believe it or not, that's got enough smart technology on it when it arrives at your facility, smart sensors, smart devices, measuring things like force, disturbances, temperature, moisture, sending all that information up through the edge to cloud continuum that our kids learn about in discover AI, by the way, through the edge to cloud continuum. And those robots right now will use artificial intelligence, and probably more accurately, some data analysis and algorithms to predict their own future failure and order their own replacement parts before that failure ever happened. So many of the manufacturing machines and so much of the manufacturing equipment that you're buying right now today comes with smart technology embedded on it. We see that in education too, and I'll give you a couple examples from education today. One of the things that I think is really important as we think about our AI journeys as a quick aside, is that our educators, in many cases, are innovating faster than our employers are. And one of the things that I've been a big advocate for, and I've told lots of educational institutions, and many of them have taken us up on the advice, fortunately, and they may be getting the advice from others and not taking all the credit, but I'll take some and getting ahead of some of these technologies and recognizing that, look, we've got to we've got to put ourselves in a position as educators to advance research and advance practical applications of artificial intelligence. We wait until our. Employers are begging for it, it's going to be way too late, and we're not going to have a workforce available to them that they're going to need. They already need, if they don't know it. We need to create that workforce today. And so lots of great, innovative organizations that are taking the first steps, and even second and third steps into artificial intelligence. One of those great institutions is the Waukesha County Technical College. That's Waukesha County Technical College, their CEO, President, rich Barnhouse, Dr. Rich Barnhouse, great friend of the podcast. Great friend of mine, personally. And we'll link his episode too. He's been on the podcast as well talking about innovation and education. Reason I bring him up is that when we talk about embedded smart intelligence on industrial assets, that organization is leading in terms of training its industrial employers on the next step in manufacturing technology, advanced manufacturing and AI technology, proud to serve on the advisory board, by the way, for the Waukesha County Technical College applied AI Lab in Waukesha, Wisconsin. Great things happening there. And you can see this kind of technology there. And really at overthrow. You think it's 1600 fan ex cert schools that are you know, that have the same type of technology embedded on their robots that'll predict future failure. It's not just robots. By the way, just about any new manufacturing equipment that you're buying, if it's got any technology embedded in it whatsoever, is going to come embedded with smart technology. So our machining centers, conveyors, PLC, driven manufacturing equipment and so on. So much of this has smart sensors and smart devices, as we all know. That means, when we talk about a device or a sensor being smart, means it has embedded intelligence, it can think on its own, and it can communicate, and it can communicate on its own with other sensors and devices, so much smart technology. I mean, you think about a machining center that is measuring things like bearing failure, spindle degradation, wearing of ball screws. It's measuring the coolant system in measuring spindles and hydraulics and drives. I mean, at tool changers, there's so many different things that we can measure both predictively and in terms of monitoring machine health that allows us to predict future failure and then avoid that failure before it ever happens. And that's using artificial intelligence. When you're sourcing new equipment in the year 2025 and beyond, those are key questions you need to be asking your suppliers. Tell me about the smart technology on this equipment. Tell me about the software that I can use to monitor this equipment tell me about the predictive analytics capabilities that are embedded on this equipment or through the computer network or a Cloud connection that I can actually leverage the data that is being produced by that machine. And if there aren't good answers for those questions, keep shopping. Because the truth is that here in the year 2025 that is absolutely table stakes for manufacturing equipment. So then, once we have that data, like we said, we can put that up on the cloud, right? All kinds of ways to take data from a machine tool, put it up on the top cloud. Another great example, my friend David A Geary at Gateway Technical College, another Wisconsin example. I'm a Milwaukee guy, so So my my examples this month or this week, I should say, are coming from the state of Wisconsin. Gateway Technical College. They're using MT link i, which is another FANUC product, and they are pulling data from all kinds of manufacturing equipment, programmable logic controllers, machine tools, meaning machining centers, robots. You know, all of this data that is coming off of their manufacturing equipment up into the cloud, conveyors, autonomous mobile robots, automated guided vehicles, so we can pull all that data up to the cloud, create a data set, and then use and which is really similar to what we were talking about when we talked about our MCP server, use our AI agents that we've created, or Somebody else has created, or maybe there's a turnkey solution that a supplier has already created to go through and monitor and measure what is taking place on the floor based upon the data we're pulling from our manufacturing assets. So incredible things that we're able to do now in the world of artificial intelligence, as it relates to equipment on the edge in manufacturing. All right, let's, let's talk about digital twins. And I want to give us an example that has nothing to do, or very little to do, at least directly with manufacturing. But one of the most interesting studies that I've seen in really anywhere, and I'll tell you why in a minute, is what we call the Harvard piano experiment. It was a long, long time ago, or I shouldn't say, a long, long time ago, but over 10 years ago, maybe over 15 years ago. But here's, here's what they did in the Harvard piano experiment. They had a short song, right? So a series of keys that somebody would play with one hand on the piano. And so they came up with this little song that people could play. And they took a few groups of people to study how they learned that song. So the first, the first group of people, was what they called their physical practice group. These are the people that sat down with one hand, played the song five days in a row, two hours a day. So two hours every day for five days, they played the same song as part of this study, and that was their their physical practice group. They also had a control group. And of course, if we know anything about the world of research, in our control group, you know that's the group. In this case, they did nothing that control group did, didn't spend any time playing that little song on the piano. So those are their two groups. But oh, by the way, they added a third group. And this is where the story gets really interesting. They added a group called their mental rehearsal group. So whereas the physical practice group played the song every day two hours a day for five days on the piano, the mental rehearsal group played the song mentally every day for five days on no piano. They just did it in their heads. So they did it. They did this whole experiment mentally, and then they looked at the results. They said they were actually looking at what happened in the brains of the people that had gone through this exercise. So the first thing that probably isn't surprising to any of us is for that control group that didn't practice physically, that didn't practice mentally, they saw no significant changes at all in what they call the motor cortex maps of the brain that control that element of learning. So no change in the control group. Doesn't surprise us a bit, but it's important for us to recognize that as we compare it to the physical practice group, which actually saw significant expansion in the motor cortex area controlling the practice fingers. So in the part of the brain that controls those five fingers, they saw significant expansion in the motor cortex area for those individuals in the physical practice group. Then they looked at the mental rehearsal group, and you know, if you if you're probably expecting is it's like, it's one thing to practice something by hand, right? Another thing in total, to practice something if you're just practicing that particular thing in your brain. But what they found, and this is The fascinating part, is that there was a comparable expansion in the motor cortex area of the mental rehearsal group that was similar to that of the physical practice group. So think about that for a minute. This group that was practicing mentally their brains and their their motor cortex area that controlled the five fingers, even though they weren't moving those fingers, even though they weren't physically practicing, they were just practicing, they were just thinking about this, two hours a day, five days a week, showed a comparable expansion in motor cortex area to the group that actually physically practiced. And this has all kinds of implications for by the way, psychology and that you know, how we train our brains to think, and the power of our brains and our mental thoughts can really create a situation, in some cases, that is as real to us as the physical experience of that we could do, podcast upon podcast and that concept. But here's the reason I bring that up in kind, in the context of manufacturing, and that is this we can use the same context, not in terms of playing piano, but in terms of optimizing production and manufacturing, and what I'm what I mean by that is, I always talk about the scariest day in the life of a manufacturing engineer or an industrial engineer. What is the scariest day in the life of a manufacturing or an industrial engineer? And that is, if you've never worked in manufacturing, the day that you make a change to the manufacturing process, because if you're the person that makes that change, and all of a sudden, you improve throughput, or you improve uptime, or you're able to deliver more product to your customers more quickly, or you're getting more quality product and less rework in the manufacturing plant, if you're the person responsible for that, you are a rock star, right? I mean, the all the way up to the CEO of the company and the board, they're going to hear about how this industrial engineer had this beautiful insight and improved productivity and made a huge difference for the company, for its employees and for its customers. If it doesn't work, exactly the opposite happens. If that doesn't work, and you shut the line down, you shut a customer's line down, you create a quality problem, you have people working Saturday on overtime to fix something that you created during the week. If you're the person that does that, you are the biggest dog in that company until someone else makes a bigger mistake than you did. That's the scariest day in the life of an industrial engineer. Well, guess what? We no longer have to just go out to the shop floor and change what's going on in the shop floor in the same sense that the Harvard piano experiment said, Look, we don't necessarily have to engage with something physically in order to improve it or in order to learn it. The same is true in manufacturing, in the life of digital twins. So what we do when we have a digital twin is we created an exact digital replica of a physical asset. We connect the two of them together, or at least inform the digital twin of what's going on in the manufacturing facility. Now, when we want to innovate, we innovate not on the physical asset itself. We innovate on the digital twin, and not until that process or product is absolutely perfect, do we take the digital manifestation of that improvement and put it out on the shop floor. And then once we do, we've got a really, really high degree of confidence that whatever we change is going to be effective, because we tested it over and over and digitally. The same is true for a new production line. We can do that digitally as well. Create a new production line. Perfect it in the cloud. Perfect, it in the digital universe, and then when we get it perfectly running exactly the way we want it to the. Then we build it on the shop floor. Digital twins are leveraging artificial intelligence in all kinds of ways, because we can use AI to optimize that process, then digitally before we put that physical manifestation of that process improvement on the shop floor. So digital twins absolutely huge. We're through three of them. We're on to number four. Number four is vector databases. Vector databases. So I'm going to ask you a true or false question. True or False The day is coming where we can put all of these things, our work instructions, our standard operating procedures, our product manuals, our troubleshooting guides, product schematics, past procedures, default settings, operating variables, all of these important data points in manufacturing true or false, the day is coming where we can put all of these in one spot, ask any question and get the answer on demand. All right, so in your head, how many say true for all those things that we'll be able to put them in one spot, ask a question and get the answer as soon as we want it, and how many say false and the truth is, I'll accept either answer on this episode of The TechEd podcast, because if you said true that certainly we are getting really, really close to the day where we can do that, where we can put all of these things, product manuals, past procedures, default settings, SOPs on the cloud. We can pull all these things together, and we can just ask a question. The reason I'll accept false as an answer is because I would tell you that that day is already here. It's not even coming. We are already at the point where we can do that. I've got a good friend who runs a company out on the East Coast, and his company did exactly this, troubleshooting guides, product manuals, supplier data, put it into one what we call a vector database, which is basically a database that's pulling data from multiple different sources, multiple different areas, and made a queryable database using artificial intelligence. In his particular case, he transformed his ability to perform technical service for his customers and in his plant using a platform called pine cone. And pine cone is basically a vector database that takes data indexes, it allows you to query that data, find similarities or commonality in the data, and then produces results. Is it perfect? No, nothing in the world is but it is really, really close to at least providing insights to technicians and to technical people in terms of going, knowing where to go to find a solution. Now pine cone AI is that is that product. We'll link it up in the show notes. No, no economic interest on our case, in our case, in any, in any way whatsoever, with pine cone AI, other than it's kind of a cool product. There's other ones out there that do the same thing. Milvis, we the eight PG vector Chroma quadrant, which is Q, D, R, A N T, by the way, there's a lot of different vector database platforms that are available widely. Pick one that's right and load up that data, because it's incredible how quickly we can take troubleshooting manuals in the manufacturing plant, standard operating procedures, past procedures, all of our equipment, documentation, and put it into a database that allows us to query it and ask questions. And how much faster will your maintenance team, for example, or your electromechanical folks, or your automation folks, troubleshoot a solution or find a solution, troubleshoot a problem, find a solution, if they just have to query a database on their phone, as opposed to going into some disorganized in many cases, Library of supplier data and other documents to try and find the root cause of a problem. So vector databases write those down. Super, super important, all right. Number five, generative, pre trained transformers. By now, almost everybody's using some version of these, right? This is your chat. GPT, I like perplexity. I like Claude. Some people use meta. AI, a lot of folks in some of our businesses leverage leverage copilot by Microsoft, but that's what a GPT generative pre trained transformers has the ability to generate content. It's been pre trained on a large language model, and it can transform that data in the large language model into usable data for us in this day and age, you can go on to chat GPT or perplexity, and ask it any question, get an answer. And that answer, by the way, is usually pretty darn close to the right one, if not perfect, and getting better every day. So so we, a lot of us, are using gpts. Let's talk about where some of this technology is going. If you're a fan of Netflix, and you will take this into the the entertainment realm. Here a little bit. Some of you may have seen, I've seen all these painkiller with Matthew Broderick. Really, really good. If Matthew Broderick's name doesn't ring a bell, if you're a youngster here, but you've watched Ferris Bueller's Day Off, you recognize that name, or at least you recognize his character as Ferris Bueller, same guy in painkiller great series called Narcos, which is, which is about the drug ecosystem, really, kind of shows the underbelly, in some ways, of illicit drugs, how they get to the United States. Another one called Griselda. Same kind of a topic, great, great series on Netflix. Well, here's the reason I bring all three of those up. There's an executive producer of all. Three of them. So same, same individual was produced all three of those series which were just, just huge hits on Netflix. His name is Eric Newman and Eric, and He is the executive producer of all three of those Griselda Narcos and painkiller so why do I bring that up? He was on the podcast. We will also link up that episode in the show notes. But here's what he said, and it's just really, really stuck with me ever since he said it in the exact quote is, you often hear people say, well, a computer is not going to write the great American screenplay. And he said that on the TechEd podcast, and he followed that up with this. He said, Yes, it will. At some point, it will. You know, of course it will. Here we have the one of the greatest, most iconic executive producers in all of Hollywood saying and predicting that at some point, a computer, at some point, artificial intelligence, is going to write the next great American screenplay. And if any of us are using regular gpts, like the ones I mentioned before, you know that that's true. You know, they're super powerful. I will also tell you that in as much as a lot of folks are just using their gpts to ask it a question, where should I go for dinner? How do I make chicken? Marsala, what? You know, whatever question we have over the course of the day, the things that we can use gpts for in business are often overlooked by a lot of folks in and around manufacturing. I mean, you can use it for brainstorming content. If you need to write an article or a blog, you can do that. We've used it in several of our businesses to perform a first review on contracts. We use it to write the podcast show notes. We'll talk about that a little bit later. Organizing content and ideas. Got to draft a sensitive email, putting together an onboarding plan, drafting abstracts, drafting sponsorship models, verbiage for negotiations, putting together a learning plan for a new employee. And even by the way, putting together this list, which our team as they put it together for me, used a GPT to brainstorm what should be on the list. And all these things are things that we are doing in multiple businesses of ours. So easy, really easy. Lift, light, lift, easy starting point if you're using chat, GPT, perplexity, Claude, co pilot, whatever. There's probably a lot of applications in your business that maybe you haven't thought of, and you can expand into that.

Melissa Martin:

Well, that is going to do it for us this week. For part one of this episode. I know there was a ton of great content in there, and believe me, there's more. I know you were probably waiting for the next item on the list, and we promise you'll get all of that next week, Tuesday. So make sure you're subscribed. Subscribe to our YouTube channel. Subscribe to us on Apple Spotify. You'll find us there. In the meantime, you can catch show notes for everything we talked about on this week's episode at TechEd podcast.com/applied, AI. We'll see you next Tuesday. You.

Head Shot

Matt Kirchner

Host