The Intelligent Factory: Exploring ChatGPT's Impact

Yeah, you can hear me, right?

Yeah, you could tell I'm not a marketer, right? Because I, I, uh— oh, we're live 6 seconds in. Welcome everybody.

We need to send out the link though.

Yeah, so let's go see. Yeah, let's send out the link because right now we have 0 viewers. So let's just, uh, pump that up here. Well, we'll finish at the top of the hour, so. All right, let me just get on to Carol. All right. All right. All righty here. Because right now we have zero viewers on LinkedIn.

I don't think we will.

No, I think we'll have time. So if you guys hold tight. So right now, hold on a second.

By the way, Jeff, why would you travel to Ireland? Did you have it on your bucket list, or what was it? Why Ireland?

It's just, we celebrate our 5-year wedding anniversary in year 8 because it was canceled 3 years ago because of the pandemic.

And you know someone in Ireland, or— I mean, you could have gone to Paris as well.

All right, so do you see on the— um, I don't think we have any live viewers yet. Is that correct? Actually, uh, Elad, can you—

I can't see anything.

Elad? Yeah, so you don't see anything live. So we're live here on— okay, are you on? Hold on. Okay. Hey. All right.

All right, looks like people are joining.

Mm-hmm. Alrighty, so let me just see if we're good to go here. So, so Jeff, do I get the job as your social media manager? What do you think? Did I pass the first test? Joachim? All right, guys.

I was wondering if an AI would do this better.

I know. Well, so we established maybe there's limits to what AI can and can't do here, 'cause I'm, you know, I sort of took for granted here. So great to see you guys. All right, so let's kick things off here and appreciate all your patience. Yeah, so I guess first, you know, for those who are here and, you know, were patient over the last 15 minutes, welcome to our AI in Manufacturing Episode 3. Today, Jeff, Joachim, and I are going to be discussing the impact that ChatGPT is having or may have soon in manufacturing as part of our ongoing AI in Manufacturing series. Just for, you know, for reference, Industry 4.0 Club is a nonprofit. Our goal is to accelerate sustainable adoption of Industry 4.0. through diverse interdisciplinary conversations with all sorts of folks, from people who are working in manufacturing to technology leaders, thought leaders. And before we start, just want to mention that we do stuff live on LinkedIn with technology, like we're doing right now, with some success. We also do stuff in person, which— so we're at the Manufacturing X.O Southeast in April. Are you there, Jeff?

You're I'm there, yeah, next week.

Yeah, yeah, you're running it. So we're there in Manufacturing X.O. We're at Eastech in Springfield in May, Tech Exchange, the Smart Manufacturing Conference, May 22nd, 23rd, and IME East in New York City in June. Jeff, do you have any events that are coming up other than X.O.?

The next one I'm doing after that's a virtual event with IIoT World specifically focused on manufacturing, both AI and Digital manufacturing. So that's just in early May, mid-May.

Okay, awesome. I'm at that one as well. What about you, Joachim?

Actually, I'm planning— this is in the making— I'm planning 2 events. One is in Dubai and the other one is in Delhi in August. So right now, but as you know, I mean, I had a surgery, so right now I'm in rehab, so I have to slow a little bit down. And from in the second half, I can Start moving again.

Perfect. So plenty of time for jumping onto LinkedIn and sharing with your audience. Definitely. So yeah, and I think it's interesting, the last time the 3 of us chatted was back in the Clubhouse days when I guess all of us were stuck at home trying to make friends and network. And Jeff, this was very early on in your journey, I believe, right?

Oh yeah, like the first couple months.

I think next time give you the keys to the car here and use your skills to get this thing kicked off. But okay, so just, you know, to start the conversation today, you know, this is not our first, you know, buzzy thing that's come around, but it feels like something's different. You know, I think in the past, if we think of the buzz around IoT, IoT already was around by the time the term came around, the buzz around, you know, NFTs and all this. So this one here, it feels like this is different than— I know, Jeff, when we spoke about it ahead of time, sort of describing how you believe this is different, but this is— it feels like something different's happening. I'm curious to unpack that with you guys. Like, is this different or is it just buzz?

I mean, I personally think it's a combination of both. It is buzz, but I think it is buzz because it is different. And one of the things that I think that makes ChatGPT so different than other disruptive technologies out there is the speed at which you're able to adopt it. Unlike the analogy I like to bring is blockchain is the big one that I would argue is very disruptive, but you need to understand how it works. You need to spend money, time, and resources to implement it, and that slows the delay in the adoption. ChatGPT, people didn't know what it was, had never even heard of generative AI in early November of last year, and by January 1st, half the world basically knew what it was and anyone could utilize it. So one thing that makes it unique from a company perspective is the speed at which you can implement it. Blockchain would take probably a year potentially to implement in a company. ChatGPT, I would argue, you could implement those models in your company in hours. And that's what makes this so much different than all the other technologies out there.

Yeah, I think it's, um, I think we were talking about earlier this fact that, you know, there's the circle of potential that can be unlocked by a particular technology. And the fact that not only is this one And then there's the people that can actually do it and have the resources to do it. And in some ways, the limitation is, you know, what idea can you get? How can you think of using it? And the bar to actually begin to use it is quite low. And I know that, you know, Joachim, you know, back at your days in Izmir, that you actually implemented, I guess, a version of chatbots in it there. This is pre-generative AI, but—

Yeah, very pre. I mean, to add on Jeff's argument, I think it's so gamified. This is so much fun to use it that it is so convincing. It reminds me a little bit— I had a conversation with my son, he's 25, and it reminds me a little bit about the Pokémon Go story, you know, when you saw like thousands of people literally chasing whatever augmented reality objects in Central Park in New York. And this was insane in a way, but actually it was so convincing. that people would do that. And there are still people riding on a bicycle having 10 whatever smartphones and chasing Pokémons. And here, for me, this is the same phenomenon. Like, you have a technology that is actually— I mean, it used to be something for just really very, very nerdy people and people who are very engaged in this and really experts. And now it is so It is so gamified and it's made so easy so that everyone understands this. Every moderator, every TV show, everyone is all of a sudden talking about this. And that's probably the difference here. That's why it is so disruptive also.

Yeah.

And by the way, just to add on this, just sorry to add on this. In 2018, I bought an Amazon Alexa, so to say. And I loved this so much because I said, like, my mom, late 80s, she can use it because she just talks to a device. And then we thought, like, why do we have even managers tipping something on a tablet or whatsoever? Why could they not just talk to the tablet and say, like, what's my efficiency right now? And what's the absenteeism? When is my next shipment? And we did this, but it was way— I mean, we had so much work to do to actually make this run. It ran and it was very useful, And that's the difference nowadays. You can do this, but so easy.

But there's a couple of things here. So, Jeff, you're talking about who can actually begin to implement these kinds of things. And now it's clear that pretty much anybody can do it. But I don't know if that quite gets it yet. And you mentioned, Joachim, the Pokémon Go. So if we think about anybody being able to develop it and can get a lot of people to jump in, but I think the question is whether or not this will have staying power. Because in some ways, like Pokémon Go and Clubhouse and all this stuff. So everybody, like Clubhouse is a perfect example. Everybody jumped in. They said, okay, this is different. But then at some point it got to the point where it's like not different enough and not good enough. And I actually just want to hang out with my buddies in real life. So like it wore off. Or Pokémon Go, I walked around the neighborhood with my kids with Pokémon Go. And then at some point you say, why don't we just go throw the ball? So like at some point there's the You know, the fact that technology is exciting. And I guess maybe what I'd like to dig into here is that, is it just that it's exciting? Is it that it's new? Or is there something fundamental and core that will become like, will change how we work? And I think that's the thing that many people are trying to figure out. And, you know, I've chatted with folks who still say, you know, this is a fad. And I guess maybe I'll ask this straight up, like 6 months from now, will we think back to the good old ChatGPT days and sort of, you know, have a good laugh about how that, you know, that fad went up and down? Is this a fad or is this a change? Is this an inflection point?

I personally think it's an inflection point. So the difference between the Pokémon Go examples, that's more of a product. Same with like the Clubhouse. It's a single product. When I'm at least talking about ChatGPT, I assume all of us will be here today, is I'm referring to what's now become the verb to describe all generative AI. And ChatGPT is just the famous one that came out. So I see someone asked a question like, what's your view on Bard here? I'm kind of referring to all of them right now. We'll talk specifically later when we're talking about OpenAI's public ChatGPT, but just the concept of generative AI is the disruptive technology. And I think it's going to completely change everyone's personal lives, professional lives. If you look at the amount of companies that are startups that are building off of this, if you look at the amount of companies that are adding it into their company, if you look at everything associated with it, I don't think it's going anywhere because it's what I'm going to call a creativity enabler. And so it's what's allowing you to figure out how to use it to be more productive in your life. I mean, personally, I probably enter 50+ times a day into ChatGPT to help with anything that I'm looking up. It's just gonna be kind of part of our lives.

So not a fad and creative, and you would categorize it as creativity enabler is kind of like the catchphrase that you would put under it right now. That's the word you would associate with it.

I think right now that that's how I would describe it because the world is trying to figure this out. They saw it It's one that most people, within 5 minutes of playing around with it, a light bulb goes off and goes, oh my, this is powerful. And then it's going to be probably the rest of the year, I'm going to guess, maybe a little longer, where people start to go, I know where to apply this. I would draw an analogy to when Google first came out. People saw the power of it, but people didn't know how to use it. It took a little while for now people understand how to use it and they use it correctly. to search for stuff. I think after a year's amount of time, people go, I know what generative AI can do, I know what it can't do, I know what it's good at, I know what it's bad at, and I know how to use it for my personal life, my professional life, and how to impact every function in every company, you know, out there. So I think it's just going to be a learning curve, and those that take advantage of it now and are creative and innovative are just going to soar ahead of everyone else.

Well, I really believe that. But what's exciting is that I think the thing that makes me excited is different, but the fact is there's so many different ways that can be applied. And maybe before jumping into that, so Joachim, you know, for you, you know, what is it for you? Like, how would you categorize your experience with it and how you see it?

Yeah, I mean, one thing for sure, and this is basically the same with IoT and Industry 4.0 and everything, when it leaves the TV screen, then the value starts being captured. And that is something that will happen with this technology as well, I guess. When we get used to this, that it becomes part of our daily life, then it will start to create value that everyone will understand because it is so convincing and so easy. You know, I had a conversation, you know, my home turf is apparel and apparel There are probably 100 million people on this planet sewing stuff, making things, assembling things in countries, typically in developing countries. And I had conversations with executives about Industry 4.0 implementation. Then they said, ah, but I'm not sure if my people will understand this. And then I asked them, do they have smartphones? And he started laughing. Yeah, of course they have smartphones, no matter in Bangladesh or in India, Sri Lanka, or in New York. Doesn't matter somewhere in the States or in Germany. So the thing is that this technology here has been built by really nerds, so to say, like by really folks who really understand what it does. But now it's entering a market of users or a group of users that don't need to know anything. They just need to have a question or a task, something that they want to know, no matter if it is building a website or if it is their efficiency on the shop floor or it is like Write me like Alexa, write me a shopping list. So that's, I think, is the intriguing part here that it will become so normal as our smartphones are today. Not now, but I'm absolutely sure that this is. So for me, it's also like Jeff, you said it's a creativity enabler. Yeah, it's basically my, it will be my My personal assistant. That's what I see it, how I see it.

One thing to add to that that I found interesting is if you look at— so when I have been talking to various different, you know, clients out there, manufacturers, about the impacts of the public ChatGPT and the Microsoft Azure OpenAI models that you can get to base it off your own data, when it's explained to customers and then assign some homework to them, things like, hey, ask every one of your top, leads for all the different functions. If you could ask 3 questions to ChatGPT, what would they be? And to come back and see what they come back with. I've had every single company come back with totally different answers. Oftentimes where I come back and go, I never thought about that. That's genius. And if you think about that, every single customer that you're dealing with, that shows the power of the creativity. The other side I think is interesting is there are integrators out there that do the ChatGPT models through typically, at least right now, it's just the Azure OpenAI available ones for your own data that you can have. But when you have different, you know, integrators of this show their demos, I actually encourage companies to listen to 2, 3, 4, 5 companies do this because they're not even explaining it the same way because they're all taking the same fundamental capability and they've come up with their own creative ways to show how it can impact a company. And I can't think of other technologies where you have the base fundamental similarity, but then when you have someone, you know, explain what to do, They're apples and oranges. They're not even competitive because they're talking about totally different use cases, applications, business value. And I think that will normalize after maybe a year or so, but right now it is a creativity enabler. And those from the integrator, the customer, the person, you know, just on a personal side, those that have that light bulb go off are gonna thrive the most because it is so simple to use.

Well, maybe so that if the concept of creativity, you know, What you're describing is that it's extending beyond just writing stuff. And I think what's, you know, what I find mind-blowing is just the application. So for example, I was chatting with a plant manager recently and he spent— I asked him to spend 5 minutes, just describe your process. And with that 5-minute blurb, raw text was able to convert it to ISO-level documentation in a flowchart that described that whole thing. And so this is just something that is not writing a poem or, you know, helping to make my kids' Easter egg hunt more difficult, which I did, and it was fantastic. But so that's an application of something that's, you know, manufacturers struggle to do. They struggle to figure out how to document their process, you know, so that they can train staff that's coming in. This past weekend at Industry 4.0 Club, we decided to refresh our website. And if you go to industry4oclub.com, That everything there was created with AI. Everything. Photos, text, everything. It's, it's, and I think if you kind of look at all the directions that you can apply creativity to, you know, to, to create artwork, to create poetry, to create text and blogs, but actually create technology. So to apply tech creativity to create technology, this is kind of, you know, this used to be the exclusively the realm of product teams and software developers and technology developers. And now you can have folks that have an idea about some piece of technology they want to build, you just need to have a little sliver of knowledge or just get GPT to teach you. So I think this is sort of, I think maybe you hit the nail on the head, Jeff, with regards to it's a creativity enabler, but the scope of where you can apply your creativity is just, it's incredible. I think what, you know, a challenge for manufacturers is that because the scope of creativity is so broad, how do you not get distracted and then and spend a whole bunch of time? How do you not create massive waste because you're like, I have all of these tools? And I think that's something that we'd love to go into a little bit later in this chat here. You're like, it's overwhelming. How do you actually do something practical so that you don't, you know, basically waste time? Because this is, you know, it's such a big change. There's an opportunity here to drive value, but also opportunity to extract value by getting distracted. But, you know, before we dive into you know, what people are actually doing, you know, what, you know, talk versus actions. You know, I'd like to spend a couple minutes here just talking about some of the risks. I think, you know, we've heard stories about different companies uploading all of their confidential data to the internet. We know that's not a good thing. Don't do that, by the way, if you're listening in. So I'm curious to get both of your thoughts about How to do this in the right way so that manufacturers can explore, can get value without exposing themselves to risk? Maybe, Jeff, I know you're plugged in.

Sure, I'd almost say there's 2 components of risk there. One is risk within the data itself, and then the 2 is risk within the use of whatever the answer it is that's generated, your dependency on the answer. So I always use— there's 2 risks there as part of it. The first one I think is mainly a risk with OpenAI's ChatGPT, the public model that everyone goes to. And I actually think every single company out there should have a policy as well as a strategy around the public ChatGPT as well as any private instance that they create in their company. Because I don't think most people are aware that when you put your information into the public ChatGPT, that you're kind of giving away your information now. Even just to have policies around the use of it is highly important for people to be aware of because they're not aware that they're making that public available. Now, there are models out there because like I said, when I was at Microsoft and they made their big investment in OpenAI, they got access to the GPT models of which they built into their Azure environment. You can have all the protection that you're normally used to with your data built into an environment. I'm sure other companies will come out with competitive versions. But the point is you're building your own internal ChatGPT rather than the publicly available. So if you go that route, you would just have the same normal risk concerns that you would have with any of your other data stored in any of your other databases that, that you're utilizing. But the other part, uh, is the risk of the answers. And this is where my biggest education I have to most people is misapplication of ChatGPT. They're utilizing it to make calculations or even risk determination when it's not designed for that. So then you end up having people, you know, make poor decisions based off of an improper understanding of how it works. Good example, ChatGPT, generally speaking, except for a few situations, it's going to give you an answer for everything. There's very few times it says no answer. All right. So it's going to give you an answer for everything. If you depend on that answer, especially if you're asking a calculation, it's going to try and calculate it.

And then when you challenge it, because you can catch it, and then when it gives you an answer, that's wrong and said, hey, hold on a second, is that right? And he says, oh no, thanks for correcting me. I noticed that I am in fact wrong here. But yeah, so Joaquim, I guess from your perspective, how do you— in some ways, I think it'd be interesting to sort of think about this concept, the concept of the 8th waste. I think it applies very well here because the waste of human capacity and human ingenuity. And in some ways, if you take this technology and basically use it to replace your brain and thinking on the shop floor in particular, that is a huge risk. In some ways, it's a tool that has the potential to drive value if it's complementary. But to Geoff's point, if you place it and make it your decision maker, it's going to make a lot of boneheaded decisions and create a lot of chaos.

Yeah, I mean, 2 things here to add on Geoff's argument. I read a post recently and it said the 1st of April is the only day in the year when people question information because they think like, okay, 1st of April you get fooled by something and all the rest of the year we believe. I don't know if you made this also, but I saw it on Instagram, on my Instagram account. But anyway, I mean, This is so true, right? I mean, we had just this week, there was a discussion here in TV, in German TV, and they were talking about what ChatGPT is doing in terms of making the right or wrong decisions when it comes to applying law and whatsoever. And then they said, yeah, but there are lots and lots of studies saying lawyers before lunch also tend to make not the nicest decisions anymore because they're hungry and they want to go to lunch. So they start to sometimes do things. In other words, We humans, we are not perfect, and this system is also not perfect. And we have this strange belief that a computer does things right, but a computer was programmed by us and it was created. Of course, we can talk about reinforcement training and computers training computers and whatnot, but the idea that we have a source of truth in, in, in a, on a planet with 8 billion people, create 5 billion people being online, creating content constantly, I think this is not realistic. So when we think about how to use this information, there will be always a point where we have to question this. And even if I go to Wikipedia, if I have enough followers, they can write me a nice story about myself that had never happened. So in other words, this is something that a single source of truth is to be defined. You need a governance for everything. Also, when I look back to the Hugo Boss factory in Izmir, we had a lot, a lot of discussions about manipulation of managers of workers. When you start paying workers by the data captured by sensors, then you want to make sure that there is a log file saying like, okay, this data is true. You know, this is what you capture. If you have like 20% or 30% of your salary being bonused by data calculated by a system, then better make sure that the system is right, you know, and that no one is able, not in the IT and no manager and no whatever, is able to manipulate this. So I guess that's something that will need a lot of governance, wherever we apply this technology, no matter in public or on the shop floor. And then the use cases ultimately will be narrowed down. That's absolutely my belief that right now we are seeing an explosion of ideas of how AI can be applied. And there are lots and lots of creativity issues coming and lots of creative applications. But when it comes to value later, that's why I'm saying when it gets a little bit more silent, then you will focus on something where you say like, okay, what is the dataset that I have? What are the answers that I want to— what is the problem that I want to have solved here? And that will narrow it down and that is then manageable.

Yeah, so I think maybe to sort of sum things up here, so the idea, and Geoff, as you're describing, come up with a policy, I know what Most people are thinking here, hey, ChatGPT, write me a policy that makes sure that when our employees use OpenAI to play around with this. But I think that's probably one of the ways to get our companies to begin to interact with it in a way where people understand the rules of the game. I think right now, if you get the business version of ChatGPT, there's a sandbox and other things where you can— the way that they use your data is different. Yeah. So, you know, one of the things that, especially for organizations to do is get those pitches, as you mentioned, Jeff, from different firms, bring in a firm that is familiar with how to integrate this stuff to, you know, and I'm sure there's a lot of consultancies out there that can help set up these little, you know, controlled areas where, you know, people don't put in confidential information. But I think the key thing is to get sort of open people's minds to, you know, what it means to have this incredible creativity enabler in your companies. And I sort of mentioned in one of the previous episodes of this how there's an HBS study where they looked at folks doing regular office work and seeing anywhere between 30% to 80% productivity improvement for folks doing their job today. And I think if we think about the kinds of things that we do in manufacturing plants, whether or not it's, you know, writing an email describing, you know, your shift huddle or sending summaries around to the company or creating a process doc to remind people to clean up in the kitchen. Like, there's so much stuff that we do in manufacturing that would benefit from this stuff that could allow us to spend in other areas. So, we have to do it safely, create a policy, make sure that our teams are educated. The other one is, bring in some folks to help out. I think there's a lot of good technology companies and consultants who are likely out there that are able to help out and to make sure that the experimentation is done in a safe way. So I think that's a key thing here. So, you know, I'd like to jump in quickly to talk about some actual things that we've seen in manufacturing. And after that, I'd love to open it up and maybe we'll pick out a few questions from the message board here to sort of do a Q&A before closing things off here. But, you know, fact versus reality, what have you actually seen done versus what people have said? Like, what are some tangible examples of something maybe boring that's been, you know, that a manufacturer has done, you know, that has clearly delivered value? So not done that's interesting, but done where there's like an ROI that could be calculated.

Well, that's gonna be a tough question knowing how new this has already come out. So I have people that are in, I'm gonna call it production with it, but to have the ROI calculation takes a little bit of time to see if it made an impact in the particular task with it. So some of the ones that I've seen are, and this is where every company's different, is they're utilizing the data that they have readily available. So if you start at a high level and go, where can I easily apply, I'm gonna call it the ChatGPT, all right, at least conceptually, I usually start with going anywhere someone has to look something up or has to ever input anything into a computer. If you start there, which is probably a lot of people for a lot of tasks, That's where it can make the biggest impact quickly. Then you start to look at what are you looking up, what data do you have available so that can generate the answer either faster or easier for you, and same with inputting anytime you have to type anything in for it. Example use cases that I have seen either be completely implemented or in the midst of implementation is 2 aspects. One, is utilizing, I'm going to call it digitized documents to be accessible into an internal ChatGPT model that was built for their company. Things like company policies. Being able to load all the company policies in there, which typically might just have a search bar at the top. But instead of having a search bar, you have a, let's call it a chat. I'm going to call it Acme, the figurative representation for a made-up company. Chat Acme. You can now say instead of, hey, What is the travel policy? You can say, I'm going on a trip to here and I'm doing these 4 things. Tell me what I need to know about the company policy, and it generates the answer for you based off of everything in every file that's been digitized. So that I have seen be used. The other one is taking user manuals from equipment. So if you have those already digitized and you have them indexed the right way, all that information can be used to an operator, someone, to access. So you can use it to, in theory, look something up, but also to generate an answer for you like, hey, I found a blinking red light on this machine, what should I do? As part of it, or summarize my daily checkoff procedure across these 15 machines for me in English and in Spanish.

But just maybe to add to the training, and somebody just said training manuals exclamation mark. So one of the things that, that, you know, we need to wrap our heads around is the idea that it's not just reading a training manual. It could be provide, you know, create, teach me how to use this thing and teach me how to use this thing in Spanish because—

You can have it generate a quiz and have it test employees.

Have it test me. So like at some point, teach my kid how to do this thing. And I think that's the thing where once you have this data, recognize that the information that's contained can be shaped and can, you know, turned into different things, right? So ultimately, training, the training manual is a convenient way to capture all of the information. Like, you slam a manual down on somebody's desk and say, here you go, like, I'll come back in 8 hours, and I hope that you know how to do the process. And like, oh man, that's, that's not ideal, right? And I think the idea that, and I think this is the sort of that creativity piece here, that you can actually get people to interact with the information in the way that's best for them. So training manuals is one, but policies. Can I spend $50 on lunch? No, you can't. Why? Right? So Joachim, yeah, I'm curious to get your thoughts here, and especially given, again, that you've actually implemented kind of a version of it at your Hugo Boss plants there back when you were there, that I guess prior to them being smart, this is when they're sort of—

Yeah, that's why I said when it leaves the TV shows, then the money will come. And I'm pretty sure that it will come. So just give you a couple of thoughts here and proven thoughts because we did a very basic version a couple of years ago. So one thing is, like, for decades we talk about semantic search in corporate information. Like, I don't want to search endlessly in documents if I have a company with, like, for example, I'm working with a client in Sri Lanka, they have 65,000 employees. 25 factories. So they create an unbelievable amount of information, and if you start digging for search words and whatsoever, you get mad, you know. So when you have a crawler like an AI that is actually giving you semantic, reasonable answers that you want to have, that is for me one thing that will speed up searches in the company. And there are lots of searches for all kinds of information— manuals, training, and whatnot. So that's certainly one thing. Man-machine ratio in the apparel industry, man-machine ratio in certain areas is 2 to 1, 2 people, 1 machine. And the reason for this is because it is super traditionally organized. You have junior managers and senior managers and junior executives and senior executives and shift in charge and another one and another one. Then you end up having like you pile up 10 levels of hierarchy and everyone is reporting to everyone doing something. Now, if you have Industry 4.0 means applied, which means like you capture all kinds of human data, machine data, process and product data, and you have a system running on this, then probably you can— sorry to, to all the managers here— but probably you can eliminate 50% of the levels. if not more. And that's something that you can— and then you can ask yourself, like, who am I going to— who will I need, whom will I need in the future to run my business? Not to run my complexity in administration, but to run my business. So I can either release these people out of the equation, or I can put them on other tasks, which is something that we did, for example, at the Hugo Boss factory. There were lots and lots of people being scared of being removed from the equation, but instead we We increased our complexity, we decreased our cost per minute, and we were capable to handle much, much more products in the same factory. And that comes to my last point here, which is fluctuation. I mean, in the apparel industry, you have fluctuation sometimes of 20+ percent on annual basis, which means like after sometimes 3, 4 years, you have a complete new workforce. So if you have 65,000 employees and 20% of them are are removed and you have to hire them new, then you know how much training cost you have. So if you apply this and if you have this brain, talking about the beginning, the semantic search, if you have this brain that you can ask, then I can know, okay, Martin needs to learn this, Jeff needs to learn this, Joachim needs to learn this. Here you are. You just ask a question, you get your manual, you get your training, you get your quiz, and at the end you get your whatever survey or something like a check. that you do and off you go. And by the way, we did something but much more complicated because we didn't have this fancy tech, and we decreased our training cost by 50% and also the time. So there's lots and lots in it when you start combining all this and make sense of this.

So I like the way that you described kind of how to think about our data because in many ways, you know, manufacturers think about numerical data in a specific way because that's been what we've been driving our business with for a long time— financial data, metrics, signals, and all that stuff. So this giant pile of numerical data, and then we have this other pile of data that is all of our process documents, all of the text interactions, and really a lot of the context. So if you want to understand your business, if you understand performance, you can see that in the numbers. If you want to understand the reasons for performance, there's context that's not visible from those numbers. So I think what's really exciting is that you can now begin to bring in this text content if you only listen. And I think that's what excites me most about ChatGPT isn't its ability to talk, it's its ability to listen and to be able to pull in this data, bring it in, and then as you've mentioned, you know, find ways to identify problems and inefficiencies in the system. And I think, you know, for organizations, for manufacturing organizations that are, you know, may have some roles that become obsolete, one thing that manufacturers have no shortage of is problems that need solving. What is a big challenge is figuring out which ones are the important ones, which ones are the biggest ones, and how do you get it where you get your folks who are not on the shop floor to stand in front of the biggest problems. And I think that's the biggest challenge right now, that they don't have the time to stand in front of that problem and fix that problem because they're spending time transferring data from one data system to another, getting somebody who never wanted to be a writer who basically needs to spend 6 hours a day writing stuff. And then how many folks in manufacturing are writing page-long emails because it's part of their job to write page-long emails about who knows what? So I think the short-sighted view is, a whole bunch of people are going to lose their jobs who are in middle management. The fact is manufacturers are successful if they are really good at continuous improvement. Manufacturers that are able to find problems and get people to squish those problems, which we are really good at, you know, we'll do really well. And I think that's exciting. But I think I can understand how, you know, some folks will be nervous as well. And I think we actually jumped into some stuff.

Which is a good thing. Sorry to interrupt you, but, you know, I mean, when the PC was introduced, people got scared, like all the secretaries got scared that they will be removed because now the boss can write his own emails and whatsoever. And before, he had to dictate everything and then people would do this. Did you see any secretary being removed? No, they do new things. And that's something, I guess, that is very important here that Jeff talked about, the creativity. And I guess the People talk about hyper-personalization. We live in a world that is getting ever more scattered and ever more complicated, and the order sizes go down, the fluctuation goes up. I mean, things are changing so fast. Someone has to handle this. And you cannot sit in a meeting room anymore with 10 people talking about the quality results from last month. This is gone. This is history. But these people can talk and spend the same amount of time to think about, like, how do handle 10 times the complexity when this rises up, when this goes up, without employing 10 more people in the order entry, for example. So I have absolutely no doubt that all these people will get a job. It's just different, very different.

The other thing that I would add to it, so the first was kind of, at least I was describing it around documents. Words basically that have been put together that are accessible. The other is accessing systems. So imagine now where this can completely revolutionize your company, where people don't need to learn your systems because they can go to the ChatACME inside your company and just ask the question. So they can go like, who's the truck driver today for this delivery route? What was the last time this machine was maintained? What is the OEE on this particular line? And it can pull from your MES, your warehouse management, your CMMS, your ERP, and it just, it can access that information. You now don't have to take a week-long training class on how to use that particular software. You just type it in, you know, casual conversation, and it's able to report the answer to you. That, that can completely change the productivity of people if they can get access to that information without needing to look it up. And then if you go the generative route, Imagine being able to type in, can you create a report for me? Can you create a visual for me? I actually think dashboards, this is my personal opinion, are going to start to be less used because you'll just type in every time you want something what you want. Instead of having a daily dashboard of your productivity of your operations, you just type in, create a dashboard, but today I want you to show me this compared to this compared to this compared to this, and it will just create the image immediately on the fly for What, what's, um, I think if using the example you just mentioned earlier, so, um, not only can it just tell you like create the report, it can actually walk you step by step.

So if, if for example it has access to all information, say I, I need to purchase like, um, a, a drill, how do I do it properly? Well, the first step is you go, go to Bill because, because Bill will be the one who has authority. Then you need to fill this form out and I can help you fill it out. So the fact is it can actually walk you In some ways, it's similar to GPS where you don't really need to plan the different steps in the journey. I just want to get here. I want to buy a drill and I want to follow the process properly, step by step by step. So in some ways, to, you know, make these tasks that are unfamiliar to you— and I saw this when I was making a flowchart with LaTeX. I'd never heard of LaTeX. I'd never heard of Overleaf. And it walked me down these steps. It gave me the chunk of code. It basically made a flowchart out of— basically walks me step by step. So it's really interesting to sort of see how basically when it holds your hand, you can do things very quickly. So we're sort of nearing the end here, and I wanna make sure we get a chance to give our parting thoughts. I think it seems like from what both of you guys have described, this is not just a blip. Time will tell. I think there's still a pretty big camp that are saying it's a blip, it's waste, it's a fad. And I think the staying power will be there if I think manufacturers in particular are able to actually deliver value, deliver ROI. You know, I think there's been challenges over the last decade where some projects that in theory should provide a ton of value don't provide tons of value because of, you know, them not being directed in the right direction. So I think, you know, focus is key here. But parting thoughts, you know, what should manufacturers and vendors and all folks, what should they be thinking about today?

Well, the one I'll add to is everything we've talked about really so far has been operational improvement. We've not talked about the other side of the coin, which is changing the customer experience that manufacturers provide their customers. Imagine now where ChatGPT is your interface to the world, all right, your suppliers, your customers, and they want to work with it because it's better, it's faster, it's more responsive, it can give you more information that you want. I mean, imagine being like, uh, my best example is like your Nordstrom Nordstrom as a retailer and you're a consumer and you go to the website and go, I really liked Pierce Brosnan in Thomas Crown Affair, put together an outfit to match his style, and it can do that. And like, that will completely change the way that you shop. Now imagine if you make tools or drill bits or, you know, it could be anything that you make, food, and you just go, hey, I'm in the mood for this, tell me what I should buy from you, and it can generate that. It will completely change the way that everyone interacts with companies going forward.

Yeah, well, you know what, actually, that if we think about what manufacturers fundamentally do, it's really providing that thing of value to the customer. And, you know, we've been— we created silos to be able to manage our businesses. You know, we can't really think about everything from the operator building the thing to the customer receiving it and being happy about what they purchased. But I think that's the kind of thing where the more we're able to map out that entire journey and manage that entire journey, the better experience there'll be for the customer, the better experience there'll be for the shop floor worker, and the better more profits there'll be for them. So I think that's a neat, yeah, I think a good reminder of the fact that let's not get our head buried in the sand or buried on the shop floor when the reality is that really we need to be, we can't forget our customers.

Yeah, I mean, manufacturers exist because they do something, they create something that someone wants. Otherwise they don't exist for like thousands of years always, which means like now if the world, if we have 5 billion people being online, and we have everyone having a personality and having a say and having a voice, then these people will want different things. And so the customer journey, that's what Jeff just said, is what SHEIN is already doing. It took them 48 hours to create a complete black and white collection when Wednesday Addams came out on Netflix, which none of the other apparel industry brands, the big ones, the hotshots, could ever do because they were crawling social media, putting all this together, and then put it on their app and start selling. Which means, so we live in accelerating times where you have to handle complexity. And this for me is one of the easiest ways here. In Izmir, we had to hire 30 people in the IT department to build the MES system and IoT and whatnot. and build the bot by ourselves because there was literally no one who could do this, what we were imagining and wanting. Now, I mean, you just crawl the web, you find hundreds of startups and also corporate bigger organizations who do this. So for me, this is like Disneyland nowadays. And so you have an idea and you have a wish and you find software for sure.

Yeah.

You increase what you do initially, which is you don't want to become an IT company. This is the very important part here. As a manufacturer, no matter if you have put screws together or you build apparel, you're a manufacturer and you should use technology that makes what you do better. And I think this is an unbelievably easy way to make things better.

Yeah, just remember the fundamentals of why you're even in the plant. And I think that's a good reminder. And I think, you know, especially given, you know, it is Disneyland and Disneyland's not about productivity. necessarily. So I think that's a good reminder that— but yeah, no, so, you know, we're kind of at the top of the hour here. You know, Jeff, Joaquim, thanks for joining the 45, 48-minute chat that could have been a bit longer. You know, as mentioned, we've been running a series discussing ChatGPT for the next several weeks, or I guess until it gets boring. So you guys let us know if you're sick of hearing about ChatGPT and we can flip on to something else. But AI in manufacturing, I think, has been Been hot for a while. We have conversations on LinkedIn Live here. We also, as I mentioned off the top here, we're involved with several conferences. There's the X.O. Manufacturing X.O. Southeast. I know, Jeff, you're there. I.M.E. East in New York City and a bunch of other ones. Check out our website, www.industry40club.com. The site may or may not have been created exclusively with AI, so use your judgments. Follow Jeff, you know, put the little bell sign on here. Jeff posts a bunch of amazing content. Follow Joaquim, put the little bell sign on. Both are sort of experts in multiple areas, and I think sort of experts in a way that understands the technology but also can help, you know, organizations, you know, make some smart decisions about how not to, you know, go running off and slam into a wall. Because right now everybody can move so quickly. But the chance, like, when we move quickly, if we don't know where we're going, you know, we get— we slam into a wall. And it's kind of human nature to get excited and run quickly. So I think, you know, having folks around you like Jeff and Joaquin and others in the industry is kind of— I— it's good to have them in your corner. So reach out. We're all pretty friendly. Pretty friendly. We're all pretty friendly, right? What would you say, Jeff?

Yeah, indeed.

Yeah. So thanks for tuning in, and I hope you enjoy the rest of your day. Thanks, everybody.

Thank you. Thank you. Take care.