Sustainable Supply Chain

From Assembly Line to Algorithm: AI's Revolutionary Impact on Manufacturing

September 01, 2023 Tom Raftery / Jan Burian / Lorenzo Veronesi Season 1 Episode 345
Sustainable Supply Chain
From Assembly Line to Algorithm: AI's Revolutionary Impact on Manufacturing
Digital Supply Chain +
Become a supporter of the show!
Starting at $3/month
Support
Show Notes Transcript

In today's episode of the Digital Supply Chain podcast we're diving deep into the world of Generative AI with two experts who've got their finger on the pulse of this transformative tech. I'm talking about Jan Burian and Lorenzo Veronesi from IDC!

Are you curious about how Generative AI is shaping the industrial sector? Jan kicks us off with an in-depth look at what Generative AI is, how it's evolved, and what makes it so compelling. Forget the buzzwords; we're talking real-world applications here! Jan’s experience in AI research offers insights that are not just academic but truly actionable for businesses. Spoiler: Generative AI is more than just a shiny new toy.

But hey, let’s not get ahead of ourselves. As Lorenzo points out, the tech may seem new to us, but it has been years in the making. No need to expect a 'revolution every two months'. It’s more about understanding how stable the tech is right now and how we can best leverage it for optimal results. Think of it as the internet – revolutionary at first, but then evolving at a more gradual pace.

What’s on the horizon for businesses looking to jump in? Both Jan and Lorenzo offer sage advice on scaling up these technologies and integrating them into existing systems. It's not just about pilots; it's about going industrial scale. And let’s get one thing straight—this isn't about replacing humans. It's about enhancing what we do, freeing us up to focus on the stuff that really matters.

Before you tune in, let me tease you with this: the episode wraps up with a look at the future, where Generative AI could be the stepping stone to fully autonomous production plants. Yep, it’s that transformative!

So, whether you're a tech enthusiast or an industry leader looking to make informed decisions, this episode is jam-packed with insights that you won’t want to miss. Hit that play button and let's explore the incredible potential of Generative AI together! 🎧

Reach out to our brilliant guests Jan and Lorenzo through IDC.com or catch them on LinkedIn. As always, thank you for listening - and if you're interested, the video version of this episode is at https://youtu.be/3-lnNUX9-8s!

Support the show


Podcast supporters
I'd like to sincerely thank this podcast's generous supporters:

  • Lorcan Sheehan
  • Krishna Kumar
  • Christophe Kottelat
  • Olivier Brusle
  • Alicia Farag
  • Joël VANDI
  • Luis Olavarria
  • Alvaro Aguilar

And remember you too can Support the Podcast - it is really easy and hugely important as it will enable me to continue to create more excellent Digital Supply Chain episodes like this one.

Podcast Sponsorship Opportunities:
If you/your organisation is interested in sponsoring this podcast - I have several options available. Let's talk!

Finally
If you have any comments/suggestions or questions for the podcast - feel free to just send me a direct message on Twitter/LinkedIn.

If you liked this show, please don't forget to rate and/or review it. It makes a big difference to help new people discover it.

Thanks for listening.

Lorenzo Veronesi:

This technology is coming to you, no matter whether you're an active participant or you're more passive, because it will be embedded in your email system, embedded in your operating system, your laptops, even embedded in your cloud system, your machine interface. So it's just coming because everybody is trying to find ways in the technology sector, how can you use this technology to transform the features and function on my products

Tom Raftery:

Good morning, good afternoon, or good evening, wherever you are in the world. This is the Digital Supply Chain podcast, the number one podcast focusing on the digitization of supply chain, and I'm your host, Tom Raftery. Hi, everyone, and welcome to episode 345 of the digital supply chain podcast. My name is Tom Raftery, and I'm excited to be here with you today, sharing the latest insights and trends in supply chain. Before we kick off today's show, I want to take a moment to express my gratitude to all of our amazing supporters. Your support has been instrumental in keeping the show going, and I'm really grateful for each and every one of you. If you're not already a supporter, I'd like to encourage you to consider joining our community of like minded individuals who are passionate about supply chain. Supporting the podcast is easy and affordable, with options starting as low as just three euros or dollars a month. That's less than the cost of a cup of coffee, and your support will make a huge difference in keeping the show going strong. To become a supporter, simply click on the support link in the show notes of this or any episode, or visit tinyurl. com slash dscpod. Now, without further ado, I'd like to introduce my special guests today, Jan and Lorenzo. Ian and Lorenzo, welcome to the podcast. Would you like to introduce yourselves?

Jan Burian:

All right. Hey, Tom. And hi, Lorenzo. And good morning, everyone. My name is Jan Burian. I'm with IDC here in EMEA. And yeah, so my focus It's mostly the manufacturing digital technology, and I'll also try, at least try to understand the way the companies are transforming themselves into the digital environment. And generally speaking, also, I'll try to also understand the way how to make a Europe and a European manufacturing more competitive compared to the other parts of the world. So thank you. Thank you for having me.

Tom Raftery:

Okay. And Lorenzo?

Lorenzo Veronesi:

Thank you, Tom. So Lorenzo Veronesi here. I work in the team of Jan and I, I'm Associate Research Director at IDC and, uh, I, I lead the Global Smart Factory research here. So I look at everything that happens, in the shop floor and how technology can support this.

Tom Raftery:

Okay. Fantastic. And one of the big technology trends that we've seen emerge in the last eight months, roughly, or a little more, has been the emergence of Generative AI and large language models. Now, we've seen how this can be useful in lots of places, the likes of, call centers and service centers, these kinds of things, and in everything from making recipes and travel itineraries to all kinds of things like that. But Is it, is it something that you think is phenomenal? Is it something that you think is underwhelming? You know, where do you fall on that scale? Is it, is it overhyped or is it something that genuinely has legs?

Jan Burian:

Okay, I started this one, right? So I mean, we, we are getting a lot of the, the questions like, like these, right? So was it the hype? Was it like a, something really like a game changing, right? So maybe the, I mean, and here in IDC, we believe it's, it's also the so like a story off AI we call it AI Everywhere, right? So in every aspect off the companies, but also the personal life. So So definitely, it's more than just a hype. Of course, as any technology it evolves over the time, right? And I mean, personally, what I can see, what's different compared to the other technologies is that I mean, it started, I mean, with tools like, uh, I can name like a chat GPT, for example, right? So, which could everyone like literally test by her or himself, at home, try to, test the technology. And then, to transfer it, I mean, into the, into the business life, right? So, I would say it's, it's from a toy thing into the really, let's say business game changer thing, right? So this is quite unique way. So definitely that's, that's beyond hype. Of course, there's a sort of the adoption side, adoption curve. Now, I mean, the companies who started there, they are sort of, they could be considered sort of like early adopters. Now, as we go through the time we can see more and more companies are jumping on that on that curve and try to investigate the solutions and the technology itself, but also try to understand the risks and we're going to get to that a little bit later.

Tom Raftery:

Sure, Lorenzo.

Lorenzo Veronesi:

Yes, yes, I agree with Jan and actually, I think at the beginning of the year, myself and a colleague wrote a blog post on the IDC website about the role of the ChatGPT in a manufacturing company, trying to sort of the spell some of the needs that were coming around this technology and for a while, that was the first search result you could find in Google, with, my name and ChatGPT manufacturing, that would be the first result. So, it was quite popular. We knew that blog had a very short shelf life, though. And, in fact, after a couple of weeks, uh, Microsoft acquired, uh, acquired ChatGPT, Google released Bard. So on and so forth, ,So the idea that we move through 0-100, uh, very, very quickly in, uh, in a couple of months, and now we are sort of at the phase where, where we see the industrial world understanding the technology and looking at the potential of this technology. And in fact, when I attended Hannover Messe in April, May, this year, it was still, it was still not a bad, um, I mean, The thing was, no company was presenting it because they didn't have the material to show, but everybody was talking about it. So it was really at the cusp of transforming the manufacturing world. And in the past few months, we are seeing a lot of deployments in this space. So we are, we're going to, to, to be covered here, um, in this, in this podcast.

Tom Raftery:

Okay. And I mean, I can understand how chat GPT might be used, as I mentioned earlier in like call centers and, tech support and things like that. But how would you use chat GPT in a manufacturing environment? What would that look like?

Jan Burian:

Before Lorenzo starts, maybe let me, uh, share, uh, I would say one line, one data point from, from our, from our IDC survey, when we asked the, I mean, global level, the manufacturing organizations about, the way they say they were thinking about investing in, uh, generative AI technology, and I was quite surprised that 27% of the respondents, and it was, I think that it was over, say, at least like a couple of hundreds of respondents there, that, 27% said that they are really also investing in this, right? Already, I mean, in, in May, That's that's a very quite a like a short adoption period, right? So 27% it's quite impressive number, right? And then another, almost like a 40. They said they, they were doing sort of like exploration, of potential use cases, right? So you can see that companies are trying to, or taking that really seriously and maybe what's what might be the reason would be also that maybe some companies, they were just quite late with the digital transformation efforts, with adoption of AI or the other, I would say, significant technology, right? In a way they implemented it or adopting it. But now with the, but now with the generative AI, there's something really new but the managers or or C Suite level people, right? So, so they went through, I would say, these sort of like cycles for another technology. So they know that, I mean, the sooner they jump on this wave, the better, right? So, because it really helped them to gain that competitive advantage. Maybe Lorenzo could really reveal the areas of use and, uh, some, uh, some use cases we already, seen on the market.

Lorenzo Veronesi:

The sky, there's no limit here. It's really, it's really, it's really something that we don't. It's very difficult to say these are the use cases because every company is making up their own, their own approach. And so it's extremely variable, but we can summarize the use cases into four broad categories. The first one is about a user conversation. So, we can see, the system used as a way to, uh, remove the friction in the interaction between a person and in a system or a machine. We won't be able to speak to our machine very soon. And probably this is not even the most compelling use case. You would want in a factory with a lot of noise and distractions, but potentially, yes, this technology can help supporting a very natural interaction with machines beyond the traditional human machine interfaces. But also, for example, think about, the conversation you can have with your suppliers of your trading partners. A company can onboard the supplier by having a chatbot interact with the supplier in an extremely precise and extremely natural way, but also by bringing out all the relevant points that are essential for signing these very complex contracts. And this can be totally automated potentially. We can see the opportunity, for example, maybe not specific to manufacturing, but it could be relevant in a, in a manufacturing setting where so many different units and the, and type of people working in a manufacturing company, having a sort of a survey house system where you see, uh, you have a conversation between the employees and the, and the company um, in an automatic way to understand maybe what are the common problems, issues arising, what's the morale of the, of the workforce, and so on and so forth. So sort of a daily pulse, getting a daily pulse of the company, what are the most commonly, thought processes, et cetera, thought ideas, et cetera. So that, that could be a way. And there are so many other examples. Another one is, another area is knowledge management. So as we see, Bard or ChatGPT have been trained on pretty much every, everything the humankind has produced on internet. So they're very good at taking a lot of information and summarizing it. So the, the biggest help here could be, for example, in creating, a centralized repository of the company information that can be interrogated in a very natural way. And also, for example, embedding this, this sort of feature with a sort of a co pilot function so that every worker can have an assistant, a digital assistant that can suggest steps, action during their, their operation, their work, it could be in a factory, it could be in a warehouse. It can be in a service execution setting, it can be in an office, it can be, uh, when you're trying to maybe define a new marketing plan, you can see that how the system tell you maybe a lot about the latest trends. And so just driving and helping the work of, of, of people. Another area, which is really what generative AI is for is content generation. So you can have, we can think about having the system, supporting the design of our company, the design process of our company, providing ideas, but also for example, ideas and suggestions and alternatives. but also it can be how this can happen in the factory with designs of the factory with layouts with proposing different operational improvement strategies, and that doesn't have the time to be accepted, but they can be received by the user and, and, uh, and validated and, uh, and considered. And this can really open up the innovation space in an organization. Also something more manual, more, more simple, like creating automatic catalogs of, uh, maybe an engineering company or like a discrete manufacturing company with a product, configuration, very complex product configuration. It can be used as a way to simplify the process of, of showing the customer or prospect. What's relevant for them and not access, access information and also, uh, of course, the software development, uh, there's a big talk about how this, this, this application can, uh, can, can, uh, code and can support code activity, coding activities, and certainly this is, can be very relevant in our organization when you want to integrate machines, uh, machines, manufacturing the, the integration, machine integration, extremely complicated, uh, activity because the machine have different programs, they have different generations and the coding is, is, there's not a standard unified system despite the effort of, of the likes of the OPC foundation. So having a system that can simplify the integration or the machine by automatically coding the missing gaps, that could be quite quite helpful and generally speaking as a way to integrate different system together to jumpstart integration of different system without a massive human intervention. That, that, that can be very helpful. And, uh, generally speaking, it's, uh, it's, uh, it's a technology that can really, um, uh, I said, um, remove the barriers and the friction between the interaction people and, and technology and, uh, and make every every process much more flexible and much more efficient. Now, this is just about getting the right right approach and we can discuss maybe the next of this. The coming steps of this webcast.

Tom Raftery:

Sure, sure, sure. I can, I can actually think Lorenzo of two other use cases where chat GPT or larger language models and more generally could be used. One, there is a podcast called the BBC Happy Pod. And it's, it's a lovely podcast that goes out every Saturday where they ask people to send in sounds that make them happy, you know, and I decided, why don't I send in the sound of my EV, my electric vehicle. Now, electric vehicles, because they're very quiet, by law, they have a speaker built into them in some part of the car, which transmits a noise, so that when they're going under 25 kilometers an hour, pedestrians can hear the noise coming from the speaker and are warned that a car is approaching. Once they go over 25 km an hour that turns off because then the noise from the tyres is enough that people can hear it. But by law they have these speakers somewhere on them. My EV is a Kia Niro 2021 model and I had no idea where the speaker was because they're hidden. So I went into ChatGPT, told it, I have a 2021 Kia Niro EV, where is the speaker in this car? And it told me exactly where to find it, so I was able to record it and send it into the BBC HappyPod. So the point I'm getting to is these large language models know everything about every device. The most arcane device you can imagine, you ask it, does it know about it? It will know about it. And you can ask it, you know, questions. We had a problem with our dishwasher in the kitchen and I asked, you know, where's a particular component of that dishwasher? It was able to tell me. So for maintenance, for example, these large language models can be incredibly useful. And the other thing that I would say is in terms of data science. So, the latest plug in that OpenAI have released for ChatGPT is what's called its Code Interpreter, which is a terrible name because it's nothing to do with coding per se. What it means is when you turn on that plug in for ChatGPT, it becomes your pet data scientist. And so, you can upload files into it. And then ask it what would be an appropriate analysis to do on this data. So you don't have to be the data scientist. I, I downloaded, for example, the stats for this podcast. They come down in a CSV file. I uploaded that to ChatGPT using the code interpreter. And I asked it what would be interesting analyses to do on this data. I gave it no hints. I just said, what would be interesting analysis to do? It examined the file, it said what kind of file it was, it said what the data was, and it, it, it then suggested three or four different types of analysis that could be done, and then went ahead and started doing them. It produced graphs for me. It, uh, showed me, or it, it created, uh, download links to Excel data, Excel spreadsheets that it created from those. You know, it, so suddenly, companies can have their own pet data scientist to look at their data and to examine it and do analysis and in all kinds of ways. So, you know, the, the, the use cases you put forward are fantastic as well. The myriad of use cases, as you see it yourself, the myriad of use cases that are out there for this kind of stuff is just absolutely phenomenal. But I think we should also talk not just about the use cases, but to Jan's point earlier, what about the risks? You know, what kind of risks are there? And what kind of barriers are there for organizations to use this kind of technology?

Jan Burian:

Yeah, I totally agree. And I mean, before we start with that, let me just add that I mean, one of our end user client already, it's like a the global manufacturer manufacturing company, they already identified 50 possible use cases of Gen AI across, you know, all the, I mean, all the units and and process areas including the marketing, sales, after sales. I mean, you name it, right? By the way, for us, it's really important to understand, uh, I would say the potential for core manufacturing processes and for logistics, supply chain processes, right? Because it's like, it's like the automation, right? Or, or like an RPA thing. Like you have like some common areas, which are the same, pretty much the same for, I mean, In financial institutions, manufacturing, I don't know, the utilities, whatever it is, right? So HR, sales, marketing, but then you have like a core processes and this is where the value of the each particular technology should be tested and proven. And then there's another perspective and it's also related to what you just said, right? So for personal use. That's, that's, that's fine. It's enough. Right. And you also don't do any, activities which could destroy something or whatever, right? So, or distruct some, some, something, right? Because you have your own brain, you get information from a, chat and then you think about it. If it's, doable or not, or if there's some, there could be some hazardous element in that, right? So it's like a personal use, but for industry use, uh, for industry usage, right? So where you put also the company's, let's say, knowledge or some proprietary information in it, right? To get some output. Then you're practically giving your... Given your, I would say your, your own information, I mean, to the outer world, right? So, and that's, and this is the segue to, to, to the risks and, and challenges, of course, right? So, Generative AI, it's, it's, it's based on a data, which is being, captured. I mean, from the it's like a unstructured post, like unstructured data, right? You can get them across the network. So in the end effect, the technology could, create its own reality, so called hallucinations, right? For example. So you still need to be thinking about the outputs, right? So if, if they really make sense or if they can be a threat to your business, or it could be like a different biases, right? So, uh, I would say that in this stage of the technology adoption or development, there still needs to be that, I would say, human element there in place, right, which is, which is, just Really thinking about the, the output and about, possible, impact on, uh, on, on the reality. Like if you, if you use it for maintenance, uh, and so, I mean, of course, if, if you just looking for some, some information, like, how to maintain this and that, this type of asset and this type of asset, you can probably find information there, but if you really would like to use it for some sort of like, uh, like, Hey. Like the question could be like, for how long I can, I can run this engine until it stops or blows up. Right. Then, uh, you'd probably be very careful if the chat, if the, if the gen AI tool tells you, you can run it for next, I don't know, like six weeks. And then after two weeks, it just, you know, explodes, right? So this might be in extreme, in extreme effect. This, this, this might be a situation everybody wants to avoid, right? So, so we are somewhere, but we need to be really careful, careful about the, how we deal with the, with the outputs of such a technology.

Tom Raftery:

Sure. Lorenzo?

Lorenzo Veronesi:

Yes. And yeah, to this point, we can really summarize the main barriers to the or risks, which become barriers into two families. One is about the sort of cyber security and data privacy kind of story. And this Jan has highlighted and this is sort of mitigated by the fact that these tools are now embedded in the main, uh, infrastructure, solutions. So Microsoft, Google, AWS. So in a way it's, it's not impossible, for companies to build, their own data, data pool to which operate on which operate these models. So, and the model has been trained on a larger data set, but then it's used on a specific set of data. And so that can be, uh, can be very, um, mitigated here also with synthetic data, et cetera, so that, that's. That's not a big deal, but it's something that has to be considered very carefully. On the other side, and how the problem is, the output, how we can trust the output. And, uh, Tom, I encourage you to ask, ChatGPT for recipes about, for example, cooking fish legs. It will give you hundreds of recipes, very articulated, and that's the problem we want to avoid. And I think, uh, the, what would come to, my conclusion is that there are two ways. Either you know already the output, so you can tell this is right or wrong. This is good quality or wrong, good quality or not. So in this case, you're using the generative API just as an assistant. So it pretty much does what you would done already, but it simplifies your work, it takes less time, but then you are 100% the competence to review the output and judge the output and elaborate from that. The other situations where you don't know the output, like the example of Jan, or like say you ask the system, what's my OE? 75. 3%. Okay. Is it right? Is it wrong? What to do with that? If you don't know this, then, then you cannot use this information. No company will sign off for that yeah approach such a project, but I think the solution here is to is to have a system that you is as extremely precise. You define it in a very precise way. What are the steps that the system is is going through to get the output? So pretty much it's something that that is very prescriptive in the sense so that you know, the output is right because the system has gone through a very precise and and and and carefully checked steps. So in this sense, you can trust the system, but if you combine both things where you don't know really the process that generate the answer and you don't know whether the answer is right or not, that, that's a no, no, that will never be signed off in any company. So the solution is try to mitigate that by adding some check level, either in the process or having a, a real person reviewing the output.

Tom Raftery:

Okay. Okay. I guess. Another, point there as well, you know, in, in terms of the, privacy point that you, you both raised, the emergence of the open source versions of these, the likes of Lama 2 and the others that are out there now mean that it should be possible to download something like Lama 2, stand it up on your own infrastructure and then feed it your own data and keep it in house and that way, there's no risk of your information being with an organization like Open AI or someone else if you're if you're not trusting of them, you know, which is which is a fair point. And I think another potential thing that might help in terms of the checks that you talked about, Lorenzo, this is kind of, um, a bit of a, a hack, I guess, but because there are so many of these large language models out there and available now, ChatGPT, BARD, Bing et cetera, you know, if you do get an answer that you're not sure of, you could try run through three or four of them and see if you do get a consensus or not. And then if you do get a consensus, you might be more confident in the, in the outcome. But again, to your point, it's not really something you should rely on and no company would sign off on that to, to, to your point. Yeah. No, I'm blindly not. And, and, but, but mm-hmm. Go on Jan.

Jan Burian:

Yeah. But on the other hand, we still, I mean, but we still talk mostly about it. I mean, again, our focus is the enterprise focus, right? Yeah. So, and here we see that, let's say there's the dynamic is a little bit different here. And here we say, when we see that the organizations, I mean, manufacturers, right? So they, but also the other type, the other industries, right? So they work with the tech vendors to build they own, I would say, protected gen AI tools, right? So which are also being trained, on information or using information coming from the really, I would say, uh, trustful data source, right? So, that could be probably the, the, the way how the organization is going to operate are going to, to, to leverage, the power of, Gen AI, uh, technology, right? And also that some of them, they also try to build something which is really like industry specific, right? So if you're an engineering company, you build that model, which is, which leverages information really most, I mean, relevant for the, I mean, for the user in engineering, for example, right? So it's like mixing everything that comes around.

Tom Raftery:

Yeah. Yeah. And you know, what's the kind of next development of technology in the manufacturing area? Where to from here?

Jan Burian:

Okay. So maybe, uh, yeah. So, so maybe I give this to Lorenzo. I mean, I can just say just what I just said, right? So, I mean, generally speaking, I mean, there are two. Two perspectives. One is the, the way the technology develops itself or evolves itself. The other is the way how the companies are going to adopt the technology, how are they going to embed it into their infrastructure, into their processes. Uh, there, where we see that what's really need to be, I mean, the homework, the managers need to do is also to understand how, uh, impact their the KPIs, of course, right? And what kind of ROI there could be, right? So, and what kind of ROI could be, I mean, nowadays, I mean, with, when you apply the technology on a current processes and, and how does this affect the processes of performance or the processes of even like entire companies in the future? All right. So this is, I mean, this type of, let's say the mental exercise must be done by the by the managers, but it's, it's like with any other technology, of course, right? But this is where we see that the organizations are at the very start, most of the organizations are at the very start when, when, when thinking of how to, how to adopt generative AI.

Tom Raftery:

Sure, sure. Lorenzo?

Lorenzo Veronesi:

Yes, I just wanted to bring out the point that in addition to what Jan has said, that we saw these technologies like booming and coming to us very abruptly uh, last year, uh, because there was a release, public release of this, of this, uh, functionalities. But the technology in itself, it took several years to be perfected and build up. So it's not like we saw a trajectory of this technology coming very rapidly at us. Because we just, as users, we saw it arriving last minute. So we shouldn't be expecting that the same trajectory will be over, over the next few months. So... We need to consider and treat the technology as a relatively stable at the moment and just try to find the best use cases and the best opportunities for that. So, we shouldn't be expecting like a massive revolution every two months. So the revolution has come. It's a little bit like the internet. It took several, several, even decades to perfect the technology in a military or private setting. Then it was made public. But then it evolved gradually, not, not to the same speed. So that's sort of the same concept here. So, let's use it, carefully and when it makes sense without being, uh, distracted by the hype and treating it as any other technology that we have at our fingertips.

Tom Raftery:

Okay, and for org, for organizations who are thinking of jumping into this space, you know, what kind of guidance or recommendations would you have for them?

Jan Burian:

Yeah, I mean, so they definitely need to understand, I mean, the technology itself and also understand how to, how to, how to, how to identify the, the relevant use cases. I mean, again, I mean it's pretty much the same like with the AI or machine learn, uh, with machine learning, technology right? So, which by the way, I mean, we can have a say, we can have that feeling that machine learning was already or has been already fully adopted, like everywhere. And now Generative AI is another, is another evolution step, but it's not like that, right? Because these solutions or the use cases, they are running parallel and I would say from my own experience as I focus on, AI research for, for many years, I mean, I can see that the company, I mean, a lot of the manufacturing and global ones, they really fail to scale, right? So they do one pilot here, the other pilot there, but to really scale in an industrial manner, that's where they struggle quite a lot. And, if they scale, then another, I would say, challenge is to maintain those models so, so that they could run, I mean, smoothly in the next couple of years, for example, right? So, so these, these, I would say these are the lessons learned, which could be transferred from AI slash machine learning use cases, or adoption into, uh, Generative AI technology adoption, right? So, I wouldn't, I wouldn't be really reinventing the wheel here, right? Because, nowadays, I would say that there could be three main ways to, I mean, leverage this technology, right? So it could be publicly available tools like ChatGPT, right. Maybe like a couple of times here, but there's also some safety concerns. Then could be the technology could be embedded in enterprise solutions, right? So in CRM and P. L. M. And, in E. R. P. Right, which is happening right now. So this is also the way or those types of like co pilots being also embedded in this enterprise solutions. And maybe the third way is to really look at that as a technology which could help you to build the use cases and applications. All right. So, if you go these three ways, you know, combine them or in industrial use, maybe, earlier I could be more focused on embedded solutions and, uh, use case and applications, right? So that's, probably be the best way. And as I said, I focus on, let's say the way to move from pilot or piloting phase. To be really able to, to, uh, to, uh, to scale effectively.

Tom Raftery:

Okay. Lorenzo.

Lorenzo Veronesi:

Yes. And to add what, what Jan has said, I think there are, there's one principle of three principle one is, is, is to make sure the deployment is right and think carefully about, especially what is done at corporate level. Because the risk I typically have is that company do a botched development, just maybe out of a rush. And then this doesn't deliver the results, expect the results. So defining the right KPI is also clear, how I measure the success of that. It's very important. But if this process doesn't go well, then company may start thinking, Ah, this is not for us. And they miss maybe a cycle of innovation. And so that's a big risk. And another point of view, I think is very important to remember is that this technology is coming to you, no matter whether you're an active participant or you're more passive, because it will be embedded in your email system, embedded in your operating system, your laptops, even embedded in your cloud system, your machine interface. So it's just coming because everybody is trying to find ways in the technology sector, how can you use this technology to transform the features and function on my products and my viral proposition. So it will arrive and a company need to have a plan and need to think about how they can transform the way that people work with this technology, because ultimately this is a third point is not about, automating people away in a way. It's not about getting rid of people and replace them with technology, but is, assigning tasks to a technology that are not efficient anymore to be done by people so that people can focus on tasks that are more value added and can be only performed by, by real people. So that's, that's sort of the, target here.

Tom Raftery:

Cool. Cool. Cool. Gentlemen, we're coming towards the end of the podcast now. Is there any question I haven't asked that you wish I had or any aspect of this we haven't touched on that you think it's important for people to think about?

Jan Burian:

Maybe not a question, but I would say that, this particular technology and generally AI could be also understood as another sort of like a stepping stone or building brick for a fully autonomous, production plant of the future. Of course, you mentioned that once or twice during the podcast here, but I mean, it's ability to write a code to also, but also to let's say, write, sort of like a program for the, for the robot, right? So it's, it's, it's, I think this is the one of the superpowers here, which could really, which could really be a game changer in a way the, the plants or manufacturing environments, production lines, are going to be operated in the future.

Tom Raftery:

Okay. Okay. Cool. Cool. If people would like to know more about yourselves or any of the things we discussed in the podcast today, where would you have me direct them?

Jan Burian:

Yeah, I mean, we can be reached out through the IDC. com, right? So where you just, you know, you can see our names. We also publish, I would say a lot of, like opinion pieces, thought leadership articles, I would say worldwide, right? So it's, it's, it's, it's very easy. It's very easy to reach me or, or, or Lorenzo on me through the web.

Tom Raftery:

Okay. Fantastic.

Jan Burian:

Or maybe also the LinkedIn. Yes.

Tom Raftery:

Or LinkedIn. Great. I'll put your LinkedIn links in the show notes of this podcast, as well as the link to IDC and, uh, people can find them there. Superb. Jan, Lorenzo, that's been fascinating. Thanks a million for coming on the podcast today.

Jan Burian:

Thank you. Thank you. Thank you very

Lorenzo Veronesi:

much. Thank you, bye bye.

Tom Raftery:

Okay, thank you all for tuning in to this episode of the Digital Supply Chain Podcast with me, Tom Raftery. Each week, over 3, 000 supply chain professionals listen to this show. If you or your organization want to connect with this dedicated audience, consider becoming a sponsor. You can opt for exclusive episode branding where you choose our guests or a personalized 30 second mid roll ad. It's a unique opportunity to reach industry experts and influencers. For more details, hit me up on Twitter or LinkedIn or drop me an email to tomraftery at outlook. com Together, let's shape the future of the digital supply chain. Thanks. Catch you all next time.

Podcasts we love