Natan: So Jeff, welcome to Augmented Ops. It's really, really great to have you on the show. 

Jeff: I'm excited to be here. I can't believe it's taken us this long to finally get together. 

Natan: I know. And, and, and what an episode this is going to be, Jeff. It's, uh, kind of the end of 2023.

And, uh, you know, for the, for those of you who don't know Jeff, you know, as I'm listening to this, like, actually, it's a pretty tough thing to say. I don't know a lot of people who don't know you, Jeff. Why don't you introduce yourself for a second here for, for maybe the couple of people who don't know you.

Jeff: So I guess for those couple people, yes, my name is, is Jeff Winter and I'm kind of known as the industry 4. 0 guy with all my trends and stats and statistics and best practices. But for my day job, I lead the manufacturing strategy for Hitachi Solutions, which basically means I help guide them to how to become a best in class digital transformation provider.

And that's part of the reason why I'm so involved in the industry and speak at events and write and even do these podcasts is to help me. Learn more about what's happening in the industry and share my own thoughts so that I can be better at my day job. 

Natan: Yeah, that's, that's what we're going to get into, but, um, I'm going to share a little bit and I'm sorry, I'm sorry if I'm going to embarrass you here, Jeff, but, uh, not only Jeff spent, uh, north of 17 years in the industry, you know, spanning domains of automation, digital transformation, um, and helping, you know, catalyze a community around Industry 4.

If you ask Analytica for 2023, he is the number one global influencer. Um, and you know, we're going to talk in this episode on all sorts of numbers. Both Jeff and I are kind of, I think we like big ideas, but we also know that you got to look at data when you're thinking through them. So I'm going to use some data here.

Um, uh, Jeff has 86, 000 followers, which is an insane number. And all his posts on LinkedIn reached up the peak of 35 million views on, on, uh, on the content that he posts, which is really, really comprehensive and covers many, many trends in the industry. Highly recommended. So, you know, this, this gives him a, um, Point of view.

So with that, Jeff, let's get going. And, and, you know, we're gonna, we're gonna think about this through 2023, which was like a pretty interesting year for the industry. Do you, do you agree in general? Disagree? 

Jeff: Well, what do you think? I completely agree. We've seen one of the biggest great Technology disruptions or new technologies come out that we've seen in, in many, many years that caused basically every company to have to pivot quickly.

Natan: Yeah. And you know, for me, I totally agree. I mean, of course we're referring to, you know, we'll, we'll do the virtual whiskey shot every time we say Gen AI on this conversation. So I, I earned the first one. Good for me. But, uh, while there were a lot of trends, you know, today, And we've been back and forth of this on LinkedIn and other places.

We're like Industry 4. 0, is that still relevant? Is that just a buzzword? Tries to encompass everything digital and manufacturing, but, but people are still having a hard time telling you what it is. And we think about the roots. It's already kind of had, had its decade and, uh, you know, from the point it was kind of coined, initiated by somewhere in the EU, I think the, the, the.

Different opinions where it started, German government, EU, World Economic Forum, and so on, and I feel like I'm hearing less of it, and in a way, sometimes I hear, hey, we've, we've reached industry 4. 0, I want to say it's not uniform. So, and it's not super widespread, but what's your take? What, what's really the opportunity?

Is the term still relevant? What are the problems? How do manufacturers are really trying to solve it? How do you see it? 

Jeff: So great questions. And there's a bunch to unpack there. And it's, it's one of my favorite topics to kind of, to talk about this. So just to level set Industry 4. 0 is the. Name, or the moniker given to the fourth industrial revolution.

And as its name implies, it's a revolution in the way the entire industry operates, including how everyone within it works. And that often results in massive changes to society. Now, the term was, was famously announced at Hanover fair in 2011 and it, it marks the fourth phase of the industrial revolution.

So industry 1. 0. really thought of as mechanization. Industry 2. 0, mainly mass production. Industry 3. 0, really automation and digitalization. And industry 4. 0, if I were to simplify, it's more about cognition and cyber physical systems. Now this movement started to exploit the capabilities of smart machines and real time communication, kind of aiming to make manufacturers more efficient and even personalized.

And one of the big focuses at the time of its launch in 2011, was around cyber physical production systems, or CPPS for short, which can make intelligent decisions through real time communication and cooperation between various different aspects of manufacturing with the intent of enabling flexible production of high quality personalized products.

You know, and with mass efficiency, but now after the launch in 2011, detailed plans kind of followed around 2013, where Germany emphasized interoperability, decentralization, real time processing, virtualization, and modularity. And industry leaders and, and even the government, including, uh, you know, Angela Merkel, but companies like Bosch and other political leaders really backed this vision.

And guess what? The idea went viral. Most advanced economies wanted to jump on that bandwagon. Uh, Italy established their Industria 4. 0. Sweden looked ahead with their production 2030. China initiated their 

Natan: China 2025. Yes. Which is, which is so, which is so crazy, which, you know, we're just going to be a year out of China 2025 in like nine days on the calendar.

Jeff: And they've invested A lot. They're actually one of the countries that has spent the most government funds in investing this. But here's where it starts to diverge a little bit, because then Japan was almost looking at it more one step further, and they came up with Society 5. 0 initiative, which mixes the digital revolution with societal needs.

But the big one of importance to us, because we're here in America, is in 2016, the U.S. set up Manufacturing USA, which created I think it's 17? 17 institutes to develop and democratize advanced manufacturing to improve workforce development through public private partnerships. And all the resources are free.

I mean, I'm personally involved with two of them, CESMII and MXD. But unfortunately, to answer one of your questions, there is no universally agreed upon or accepted definition of Industry 4. 0, because as a concept spread across the globe, it took on new flavors to kind of adapt to unique industrial landscapes in those countries. In the United States, it generally emphasizes the integration of big data and internet of things to revolutionize manufacturing processes. But at the same time, it's expanded well beyond production itself to include the entire value chain, focusing heavily on innovation and agility. Heck, we even hear terms out there now like supply chain 4.0, quality 4.0, to name a few. But here's where it starts to go in different directions. So Europe, while still committed to the principles of Industry 4.0 has already begun transitioning to industry 5.0, a concept that reintroduces the human element into the automated manufacturing equation. And it, it kind of champions a collaborative coexistence of smart systems and human ingenuity, kind of striving for a balance between productivity and social well being. Personally, I love the concept, but I think they should change the name, as Industry 5.0 as a term is kind of misleading, and it isn't a new industrial revolution.

It's kind of more of an evolution of Industry 4. 0. But even 12 years after its first public, uh, you know, announcement, The term is still very strong. If you use Google trends, it's just one source. It peaked in 2019 globally and still at about 65 percent search volume globally. And in 2021 is its peak in the United States with 85 percent of search volume today.

And if you look at Google Ngram viewer, which tracks references in books, it's continually increasing year over year, and it's currently at an all time high percentage in books, so it's not just. It's increasing, it's increasing its percentage over time. So I would argue it's still a very strong term, it's just taking on different meanings to different people in different places.

Natan: You know, one thing you mentioned, um, on the 5. 0 that is really top of mind for me from the work we're doing here at Tulip and, you know, we're really thinking about new types of production system that put people in the center is, and I've, you know, I've kind of tried to write about this topic a little bit in Augmented Lean, you know, the book we put out last year, where do you see humans fit in all this? Because, you know, I don't, you know, I don't have a gen AI— now I have to take shot number two—agent to, uh, uh, you know, uh, analyze your speech real time. But, you know, I think I want to say the majority of the description and the how you talked about Industry 4. 0 is really the, The, the system technologies, automation, et cetera. And yes, there is humans in all that, but where are the humans in industry 4. 0 and how should we think about it going forward? If it, again, I don't think about it through 5. 0, I'm just thinking about it. Like just what, what about the humans?

Like where, where do you see them? 

Jeff: Another good question because humans are central to industrial digital transformations or any transformation, even as we integrate more sophisticated digital tools. And Augmented Lean, which I got a copy of earlier this year, you know, with its focus on human centricity, it underscores the importance of Empowering workers through technology rather than replacing them.

And this philosophy aligns with the concepts and principles of Industry 5. 0, which remember, we still need a new name for, but it, it places a strong emphasis on the collaboration between humans and machines where Industry 4. 0, once again, kind of focuses more on kind of the efficiency and how to harness and capitalize on all the data where kind of the Industry 5. 0 kind of builds upon by focusing on the human touch for innovation, customization, and social responsibility. And so I would say in this context, humans fit in as decision makers, like the innovators and the quality controllers who leverage digital tools to enhance their capabilities. So by using technologies such as AI, IoT, digital twin, data analytics, frontline workers can make more informed decisions and perform tasks with greater precision and reduce monotonous or even dangerous work. And this, this symbiotic relationship allows for more resilient and flexible production system that can adapt to changing demands and create more value, both for the businesses, the employees, and the customers. So ultimately Augmented Lean and Industry 5. 0, in my mind, they kind of remind us that while technology advances, the human element remains irreplaceable. And the creativity, the empathy, the complex problem solving abilities of humans augmented with digital tools are what drive the next era of industrial innovation. 

Natan: But let me kind of juxtapose two things here. And I'm kind of thinking about it through, you know, when we say automation, we really think about Classically, okay, we think about PLCs and maybe conveyor belts and robots and, you know, really serious sort of, um, you know, data ecology that goes along that.

And that's, you were saying, you know, we're automating the process. And often the mind goes to automating humans away all the way, if you take it to the full range of the spectrum, to Lights out factories, you know, sometimes here at Tulip when we were working with partners and customers, like the, when we talk about, uh, automations and human automations, it's about helping the humans automate their processes, which kind of has an interesting dialogue with, you know, all these topics around citizen development and, um, Having people with the context of the work, pick the tools to solve the problems and not just rely on pure automation and knowing that change is coming because when you're saying automation, and I think this is true in any engineering medium, if you think about it, it's like, you know, you got to know what you're automating, what's the spec, operating procedures, limits, you know, because otherwise you can't test the system and you can turn it on and say input, uh, Process output, whether it's a computer program or a whole, you know, crazy machine that builds, tests product XYZ.

So how should, you know, with all that, like two types of automation, what, what are you seeing? And how should people building and running production floors should think about this two juxtaposed approaches to automation? 

Jeff: Well, it's a, it's a good question because even the word automation has taken on different flavors over the years where traditionally it meant just the reduction of human intervention processes and people thought of it as robots and advanced machines.

But now that we've started to digitize. It's becoming automated digitized processes. And that's the whole field of robotic process automation, which really isn't the same as industrial automation, but manufacturers kind of, regardless of what you're using here, conceptually or similar, but just have a different flavor, manufacturers should kind of approach automation as a compliment to human labor, human labor, not a replacement, and it's kind of the vision of lights out factories, which we hear all the time, has evolved with an understanding that AI and robotics excel at repetitive, high volume tasks, whereas humans are unmatched in areas requiring creativity, critical thinking, and complex problem solving. The concept of lights out factories does scare some people because it's viewed as potentially replacing humans, but I would view it as just kind of shifting some of the work that humans are doing, not necessarily replacing them.

Because AI, for example, has been instrumental in the shift, demonstrating that while it can optimize production lines and process vast amounts of data, it lacks the nuanced decision making of humans in specific situations. And therefore, the key lies in collaborative automation, where AI systems and robots handle Tedious and physical, strenuous work, allowing humans to focus on value added tasks such as oversight, innovation, quality control.

And this approach enhances productivity, it reduces errors, and leads to higher job satisfaction as workers engage in more meaningful roles. Now, in addition, AI's predictive capabilities can be harnessed to improve anything from Maintenance to supply chain management, demand forecasting, all which supports human decision makers.

But by using AI as a tool that serves the workforce instead of replacing them, manufacturers can create a more agile, efficient, and sustainable production ecosystem that benefits the unique strengths of both humans and machines. In fact, there's a study done in 2020 by MIT with Boston Consulting Group that breaks down AI into five They call it levels or modes of integration into decision making, where level one is the human generates and the AI evaluates.

Level two is the AI generates insights and the human uses it in a decision. Level three is the AI recommends. Human decides. Level four is AI decides human implements and and the last one is AI decides, and the AI also implements. So you kind of see this progression of AI's involvement in the decision making from very little to completely doing it for you.

Now, what's interesting about the study is that says the companies that are seeing the most significant financial benefits, are those using all five modes. Mm-Hmm. not just. The highest level of automation where the AI decides and AI implements. They've figured out which tasks are most appropriate for which level.

So my take on this study is that those that are really using AI right, they know the goal isn't just blindly handing over control of decisions to AI. It's about understanding where AI can be most effective and where human intervention remains paramount. And it's, it's a delicate balance, ensuring efficiency without compromising the unique human touch.

Natan: Yeah, I, I, uh, I gotta tell you, this reminds me of, um, this kind of rant that went on Twitter. I, I don't remember, uh, the, the source, but it went something like, uh, what's wrong with this picture? It's like, humans are doing the hard jobs on minimum wage while the robots, uh, write poetry and paint. That's not the future we wanted, you know, and they're kind of like Because all these robots are doing like, uh, you know, kind of, uh, creating, uh, Hollywood class episodes, uh, scripts and beautiful images or horrible images on Dali.

And uh, we're still, we're still hard at work in many other things as humans, but, um. You know, taking it in a little bit to the, you know, the reality of automation and all this sort of, you know, unavoidable and, and kind of like my take on this AI stuff, you know, and I've been, you know, we're doing work on this, we've, we've put products that involve having a copilot and putting it to work to do things like, you know, you can ask it to analyze what line three did versus line four, and it will generate your.

Text response and charts and stuff like that. And you can ask it to build custom widgets in Tulip and that makes no code faster and things like that. But, you know, all this, all this is kind of an interaction modality, right? In some form of automation, classic or, uh, the new, the new kind, more like human process automation and, which is kind of an interesting way to call it, I think, because you just mentioned robotic process automation.

I'm like saying, there is like. HPA, which is like human process automation, which, or HDPA, which is like human driven process automation, because it's kind of funny every time I think about this, it reminds me also of HMI, which is like the most weird name to say human machine interface, which like this interface has nothing to do with a human and it's only like machine, so it's just need to be machine interface and like cut the human because the past 20 years has nothing to do with the human.

But anyway, Have So, all this stuff really creates a ton of data and, uh, I think that's another challenge we should be talking about because, okay, you implement all these things, you have automation and now you're, how do you get actionable insights from this data? How do you, you probably heard the cliche, data rich, information poor, you know, DRIP and I don't, I don't know.

I mean, I think I've been hearing it for, 15 years at least. I think we're going to keep hearing it, but what, what can we do? What can we do to get from all this data insights? Well, what do you, what do you recommend? 

Jeff: I think we're all trying to figure that out and, you know, kind of to add another phrase, uh, we, we often hear data is the new oil and I, and I like it because unlike oil, I like to compare it because data is not a scarce resource.

It's abundant, it's perpetually replenished, and more importantly, it can be shared and used without depletion, enabling endless innovation and collaboration. But, unfortunately, we are producing mind boggling amounts of data right now. According to Statista, we're going to generate, as a society, upwards of 120 zettabytes worth of data.

And even if you don't know what that means, it's an absolutely huge number. But what I like to point out about that is that, just to put it in perspective, That is 24, 000 times the amount of data that we generated in all of society up until 2003. And we're going to produce that just this year alone. And so, usually the follow up is like, wow, that's a lot, but, but who's producing and collecting all that data?

It's not the banks. It's not healthcare. It's not even retail. It's manufacturing. According to McKinsey Global Institute, manufacturing is collecting nearly double the volume of the next highest industry. Yet, here's where the reality really sinks in. Out of the roughly 2, 000 petabytes of data collected by the manufacturing industry over the past 10 years, the Industrial IoT Consortium estimates that we're letting 99 percent of it slip through our fingers, just discarded like yesterday's newspaper.

Now, the lion's share of this data comes from supply chains, the strategic sourcing processes, the multi facet operations within factories, and even the stringent stages of compliance and quality management. But according to HBR article that I just recently read, on average, less than half of structured data is actively used in decision making, with less than 1 percent of unstructured data, which is the bigger portion of data, being used or analyzed in any fashion.

And that's crazy. So if you were to look and go, okay, so what are you seeing is some of the mistakes out there? I'm saying it's 

Natan: But Jeff, why is it so crazy if, you know, just to be a little bit critical here, work's still getting done, products are being shipped. Yeah, we had like a ginormous supply chain crisis a couple years ago, but also to be fair to the world, there was a global pandemic that kind of screwed things over.

So I hear you on the data, and I get the oil argument, but the question is like, you know, sometimes So what? Like, it's still going and, you know, the world keeps turning and you need some data, but like, I guess Not so much, so why, why are they even collecting it? You know, how do you explain this, such very clear dissonance between, you know, knowledge work that is so knowledge driven and so much, uh, you know, every little thing we say, you know, A, B test this and do that, you know, I'm speaking as a tech engineer type brain that I am.

And yet in manufacturing that is not less complex and it is all about data, the phenomena you're describing exists. So how, how do you explain this sort of. Conundrum, like what, what's going on here? Well, 

Jeff: and this is an interesting way that you phrase that because I would view it not necessarily as a problem.

And the reason why I like it as the oil analogy is it's an untapped resource that's free. And those that figure out how to take advantage of it will walk circles around those that don't. So it's not that you need it. Yeah. You can keep doing what you're doing. 

Natan: You're just going to stay behind. 

Jeff: The best companies out there.

If you look at the World Economic Forum, they just announced their next wave, uh, last week. They're up to 153, uh, lighthouses. And you start to look at what they're actually doing in the use cases. Every one of them had to take advantage of that data in order to get the output that they got. So it's one to go, you want to look at what the best of the best is.

They're harnessing that power of data, not only to help increase efficiency, but to reduce cost and even to provide an entirely different new way of providing value to customers. I mean, there's tons of stats out there that say things like University of, uh, Texas, um, Austin, Texas. Produced a study that says a 10 percent increase in data usability, just 10 percent results in the average Fortune 1000 company's revenue increasing by 2 billion dollars.

You know, McKinsey did studies that said data driven companies are 23 times more likely to acquire new customers over their peers. Not two to three, 23 times. So I would argue there's, to use another analogy, it's because there's gold in that data. And those that, that. Those that tap into it are just absolutely going to thrive.

Natan: So let's, let's build on this idea of, uh, you know, those who will kind of, uh, venture into the promised land of data would, uh, I think they would outpace the competition and like create, you know, insights that allow them to do things better, faster, cheaper, you know, that kind of thing, um, and just create more resilient businesses.

But the, but the reality, and I think, uh, you, you. I have a point of view. I know this is hard. I mean, we're seeing it day in and day out, uh, and, and this is not just a technology problem. It's a, you know, this, uh, we said digital transformation and, uh, uh, time someone says digital transformation, uh, you know, an angel falls out of the enterprise software heaven to the ground, loses their wings and goes, tries to help some company transform.

Um, so, you know, so it's like the change management and, you know, all this kind of stuff. It's pretty much augmented lean and. It's just not, not just a technology problem and a lot of things kind of falls into what, uh, my good friend Enno de Boer and his team in McKinsey, uh, started calling Pilot Purgatory.

Why is it so hard? What, what, what is the, what is going on with this like top down, bottom up dynamic? How do companies should think about this? How should we help companies prepare for these kind of projects? What, what makes it work better? 

Jeff: Well, and part of this, you know, is just the way that this kind of came to be.

So, not only are we producing more data, but at the time, even when we were producing data and not using most of it, we're still not using it, it's because we didn't have the tools. To be able to fully utilize it, this is where AI shines is being able to make sense, uh, you know, a vast amount of data to look for those connections, causations, correlations, to be able to provide meaning to stuff that 10 years ago you thought was meaningless.

But if you were to ask kind of like, what are some of the big mistakes I see out there and why companies aren't, aren't doing more faster when they already have the data? One is over collection without focus. You know, gathering and storing data with no clear objectives as to why you're collecting it or why you're keeping it.

Number two, and related to that, poor data quality. You know, if the data isn't accurate or up to date, it's, it's kind of like trying to build something with faulty parts. It just won't work. Uh, the third reason, I would say, is lack of sharing. So data silos happen when information is kept locked away in one department, which, you know, can stall collaboration innovation.

According to a PTC study, uh, last year, only 34 percent of companies reported that data created in their department is widely available on enterprise systems. And sadly, that's the best step. It actually gets worse from there. 16 percent of, uh, of company data outside their department, 9 percent of data from customers or products out in the field, and only 8 percent of data from suppliers, that's.

That's data silo right there. It's just not even being shared across the departments. And then the last one is ignoring context. Data without context is kind of like a puzzle piece without the picture. You don't really know what to do with it or where it fits in. It will take forever to make use of it.

So I think kind of those compounded just built up this massive amount of data that we never really looked at and now we're wanting to look at it and going, wow, this is a, this is a monster to tackle. 

Natan: Yeah, i, I, um. Yeah, I think context, context is king for sure. Like the, sometimes I show people, like we have like this dashboard, which looks on the wall with the, we call it our mission control, like with all the different sites and you see all the event emits and all that kind of stuff.

And, uh, show it to all various people and it's like, I can't believe that this can be done. No, this is 2023. You can actually log into your factory remotely and actually like see every work center, every station, every application, who's there. You can actually get a ton of context. It's not the future, it's kind of now.

I personally think a lot of the problem is how you use that to change how you work, like from the process and the Human centric aspect, because at the end of the day, companies are the people running them, kind of like, if you think about it from Conway's law perspective, so, you know, and, and, you know, we've been, we've been talking a lot about automation and generative AI and all that kind of stuff, and we're coming up on time here.

So, You know, we're sitting at the end of this very, very exciting year that we just covered. And I think this was, uh, this was a great, I mean, I think we'll try and distill some post out of this that kind of like what, what happened in 2023, but maybe we can, Jeff, maybe we can end on, on like, What's going to happen next?

So we're ahead of 2024.

What, what's, what's coming? What's down the pipe? What do you think? What are you tracking? What is the coolest, most important stuff that people should be thinking about? 

Jeff: Well, this will be the first time on this podcast that I'm going to use the term generative AI. I, there's, there's no way I can not focus on it because of how disruptive it's been.

And this is an example where one of the things that I'm going to see is a shift, a shift in how it's used. In fact, just today, IOT Analytics came out with their, their famous, what do CEOs talk about every quarter? And just today it was released and it talked about how the term generative AI dropped dramatically from last quarter in public earnings calls.

And they evaluate thousands of public earnings calls to see what words are used in there. And it went down a lot. And so you could take this as, Oh, is generative AI not as important? I would say not at all. If you look at the market trends, the spend is going up at an astronomical rate. But the sentiment is shifting from needing to struggle to needing to like address the hype and have a public statement on what you're going to do with it and what's your strategy for it to now it's time to show results.

It's focusing on the deployments, the scaling and the integrating it into the actual company. So I think we're going to see a bigger shift into the actual integration into all of our workloads. You have, you have companies, you know, like Microsoft being the big example have announced hundreds of copilots in basically every product that they've made.

And they did this within one year of ChatGPT being released. You're going to see many, many, many, many, many companies have copilots like yours built into products, which are fundamentally going to change the way that we interact with all systems. Anytime you need to enter anything into a computer or extract anything out.

It's going to change the way that you do it. So we're going to start to see this trickle down into the manufacturing processes. Once people figure out where and how to use this stuff correctly, and we're—at least at Hitachi Solutions—we're starting to see a big uptick on Customers knowing more specifically where they want to use it versus in 2023, it was a lot of inquiring and investigating and how are people using it?

What are they doing it for? And now we're seeing the shift in, I know where I want to use it. I know where we have data available, where we have PDFs available, where we have access to information and how we're going to transform the the work of individuals and we know what individual we want to transform with it.

So, I mean, I really am still going to say that's going to be the biggest transformation that we'll see. We'll start to see more GPT models come out that are lightweight, that can be run simpler and less expensive than the big GPT-4. And that will, that will help, uh, skyrocket implementation as well. So, I mean, that one's going to, in my opinion, it's still going to trump everything else for, um, the way it integrates into all other technologies.

So, it's just kind of like the, the umbrella over all of them right now for what I, I predict we're going to see. 

Natan: What's the coolest application you've seen in, um, manufacturing in Gen AI? I'm kind of curious that, that kind of you believe in that is Not hype, because you know, this, you've seen all these like veneers on, you know, chat GPT for X type of things, you know, and you're like, okay, I get it, but is it really something we're going to be doing?

Is there, is there like a good example of real Gen AI applications that are, you're looking at this and you're saying, oh man, okay, this is a better way to do it. It's absolutely the thing that's going to stick. 

Jeff: So, uh, I would say. Sort of, because you have to remember it just became ChatGPT's birthday. So this has only been out for one year where the public kind of knew about it.

And companies started to purposely attempt to integrate this. That first year was experimentation. It was attempts at pilots. And just now you're starting to see some attempt at scale and some ability to look back and evaluate ROI or. Uh, outcome because there hasn't been the runway yet. You know, a lot of these companies, I mean, there are ones that we're just doing last week, so there's, there's no runway yet to know like, Hey, this has been in place for a year.

This is how it's been impacting the employees and the impact on, on the business. The most famous case study out there was the CarMax case study, because that was actually done before chat GPT. It was built off the GPT three model and they kind of in essence, made a little bit of their own chat GPT before it came out.

But they're the most famous public case study of having over a year's runway to show how it dramatically decreased the labor in their, um, their team to, to generate and analyze all the stuff on all their cars and all their websites. So it's not a manufacturing use case, but it's a very clear, they, they, they showed all the value that they got out of it. And it's the shining example of, hey, this is what it can look like. And this was before all the models that are out now. One of the biggest ones that I'm seeing though, that companies are having interest in right now. And part of this, we've actually done it for we Hitachi solutions for non manufacturers so far, and it's already running.

Um, but we're, we're starting to do it for manufacturers is taking advantage of all the PDFs that your company has. So we've already done one for a real estate company where we took all the listings of all the houses that they had, and we're able to load them into a GPT model to be able to access that information.

So we're starting it right now with user manuals, because every company has user manuals of your equipment. If you're. Machine was made in the last 20 years, you have a PDF user manual, and you can easily scan those with cognitive form recognizers to allow that data to be accessible to the employees. And we're seeing it for knowledge management as well as maintenance.

So imagine being a technician and you could just type into your company's chat GPT interface, if you want to call it, and you could just go, what's the blinking red light on machine 7 mean? And it can tell you what that means, you know, automatically. And then I've seen, that's like an entry level. Honestly, that's a pretty easy, uh, project to do.

Have you seen the, uh, the frontline co pilot, um, launch we did with DMG MORI and Microsoft, uh, at the EMO show 2023 in Hanover this year, September, because it does- 

Not the one at Hanover, but I paid attention to their, uh, you know, all their other major releases that they've had. 

Natan: Yeah, no, but this is us DMG Mori Microsoft putting the frontline, the Tulip frontline copilot on DMG Mori machines.

Jeff: I've seen your demo, yeah, just not at Hanover. 

Natan: Yeah, so we, that's where we kind of had a bit more of a kind of a marketing unveil, but, uh, it does exactly that. Like you can tell it, you can, you can, you can do it like from the, from the machine control panel, which was kind of cool embedding it into the, the machine. And, uh, you can even like use, um, Kind of text to speech.

So that use case, I, yeah, I've seen, I've seen people do it. And I think, I think it's going to become pretty much table stakes. Like if, more akin to how, like the world kind of, like, do you have, do you remember using Yahoo? 

Jeff: Yeah. 

Natan: Yeah. But do you remember like, I'm never going to use Yahoo again moment after, after you use Google, because I remember that. I'm like, okay, I guess I'm done with Yahoo. So I don't think it's there yet with GPT, honestly, because it's not close enough to the applications and to the context. Like we kind of mentioned before, another reason why context is so important, but, uh, it's definitely getting there. Um, yeah. I'm just like worried of, of, uh, you know, the, the people in operation, the engineers, the, the, the, the, the people who actually do the work, the, the lean, uh, people, operational excellence, quality, you know, they've been deprived of like great tools for the past 20, 30 years and like, if this gen AI revolution passes them on, you know, while we're figuring out the hype and what's real and what's not, then we're missing something again.

And then, then we'll have to do another episode on like, uh, How we left them behind again you know that's that's kind of like the stuff i worry about 

Jeff: the big thing that i am seeing with the shift in mindset of any of the GPT or generative AI and this is a personal revolution i went through cuz honestly i use those things dozens and dozens of times every single day it's about learning how and where to use them and one of the big areas where i think people fall down on using it or get discouraged or don't think that highly of it is they think– we'll just use ChatGPT as like the verb, the Google of, of, of all generative AIs.

So when I say it, I'm meaning any generative AI, but they, they think of ChatGPT as meant to be a replacement of knowledge as opposed to a tool that understands what you asked and can spit out answers in a way that you understand. If you think about it that way, you'll stop asking ChatGPT for answers to questions, of which it may get some right, it may hallucinate, it may make them plausible, but totally unrealistic.

If you start thinking of it as a tool to assist you in your ability to answer something, it's a game changer in the mindset that you have. I don't use it to answer questions for me. I use it to help with my creative thinking, to help with my ability to convey stories and information. I use it to help summarize.

And I use, I use it in a different way than I, I first used it where I would type in questions trying to get answers to things. And I got discouraged because I'm like, the answers aren't that great because it was never designed to know everything. It was designed to understand what you asked. And so people are using it wrong.

But if you use it as it understands what I ask, I can point it to my own data and then it can produce a very real. And good answer, but people aren't doing it that way. So almost all times I use ChatGPT. Now I use one that we built one inside our own company at Hitachi Solutions called Enterprise Chat.

So it's, it's not the public ChatGPT, but every time I'm using it, I'm loading my own information along with it that I want it to do something to rather than just asking it a question going, give me the answer. That's a fundamental shift on how you think about where to 

apply. 

Natan: Yeah. Fine tuning is a form of art right now.

You know, this has been super great and, and, and I think, I think like one thing we can definitely agree, we're going to have a very generative 2024, um, and, uh, we'll have to do a follow up episode on that, uh, but we're out of time, so I wanted to thank you so much for joining us on Augmented Ops, uh, it's been a pleasure and, uh, happy holidays.

Jeff: Well, likewise, thanks for having me. And I look forward to round two. 

Natan: Yeah, we'll have you in soon. And, uh, we, Hey, we might even do it in person in the Tulip studio in our headquarters in Boston. Um, and maybe, maybe shoot a segment at the Tulip experience center. It's pretty cool. Um, but until then, um, take care.

And again, thanks so much for coming, coming on the show. 

Jeff: Likewise. Thank you. 

Natan: All right, Jeff. See you soon.