Episode Summary
The episode titled "Share PLM: The Case for Open, Collaborative Product Data" delves into the insights and experiences shared by a panel of experts from the Future of PLM webcast. Hosted by Michael Finocchiaro alongside Oleg, Rob, Yas, and Maria, this podcast explores the challenges and opportunities in organizing an event that focuses on human-centric change management rather than just technological advancements. The guests discuss their experience with the SharePLM Summit, a conference they organized to address the gap in traditional PLM events by emphasizing the human side of implementing new systems. They chose Jerez de la Frontera, Spain, for its unique Nest building, which serves as a physical space for design thinking workshops, fostering collaboration among global teams. Key insights include the importance of addressing human resistance to change and the value of diversity in perspectives, particularly with an equal number of male and female speakers. The episode highlights the significance of community and unity across different PLM systems, emphasizing that challenges related to change are universal.
The most critical technical or strategic insights discussed revolve around the necessity of focusing on the human aspect during system implementations and the benefits of a more collaborative approach in PLM management. The panelists underscored the importance of bringing together diverse teams for workshops and social interactions, which can significantly enhance their impact on business outcomes. They also highlighted the potential for expanding the conference to include more days dedicated to problem-solving sessions, thereby increasing the return on investment for attendees.
For PLM and engineering professionals, the key takeaway is the realization that successful system implementations require a balanced approach that considers both technological advancements and human factors. Emphasizing collaboration and community can lead to more effective change management strategies, ultimately driving better business results.
Full Transcript
Michael FinocchiaroMichael Fenicero here with the Future of PLM webcast. I'm joined very happily with my friends Oleg, Rob, Yas, and of course, Maria. We all had a fantastic time at this wonderful conference, the Share PLM Conference in Jerez de la Frontera in Spain. So we wanted to start just by kind of a, the Wizard of Oz, pull back the curtain. What's it like? Question to Maria, like organizing a conference at didn't exist before, finding speakers, I what was that like? Yeah, I mean, so when we first began thinking about the Shared PLM Summit, I was actually in Madrid with Beatriz and Elena, and we were discussing, it was the end of last year, and we'd all just come back from various different events. And we were discussing the fact that whilst all of the events that we'd been to had been great, We noticed that there was a gap that many of the typical events that we all attend, they're tool specific, system focus, and we focus more on the technology themselves and the features. And we noticed that not many people were talking about kind of the human side of change. So what was it like actually implementing a new system? Were people brought on board? And obviously this is something that we're very passionate about at SharePLM. It's what we support our customers with. So we thought what better opportunity than to kind of try and bring our own event to life, which is kind of slightly different in the sense that it is system agnostic and human focused. And that's kind of the inspirational kind of where the initial kind of thought process came to organizing the event. And what an event it was. How did you guys fall on the choice of Heras? wasn't exactly the most central place in Europe, to hold it? No, definitely not. So basically, I mean, I'm sure maybe some of our listeners today have already heard of the Nest. So essentially, this is a place that is currently finalizing its construction. But it's a place that's...
or a building, shall we say, that we are putting together in Jerez in the south of Spain, to essentially have a physical place where we can bring customers and people together to actually take part in many of our design thinking workshops. And this came out of the fact that, as you all know, SharePLM is a predominantly remote company. We don't have a physical office. And this was great, especially during the years of COVID. However, after COVID, everyone had been locked in their houses for a very long time. And we noticed from many of our customers, they were kind of asking for and wanting to bring a lot of their kind of globalised teams together into one place. So we thought, well, what a better way to bring people together than bring them to Spain. They can enjoy the kind of the amazing qualities that Spain has to offer, obviously amazing sun, amazing weather, amazing food, but also a place where teams could come together and connect. and be better aligned as they work through their transformation projects. And you managed to get an incredible group of speakers, three of which are present. Yes. At the event, Rob did this amazing job as Rob William Shakespeare Ferroni with a murder mystery. Do you want to tell us how you came up with that idea, Yeah, sure. mean, like Most of the way I approach work, I thought about it from the perspective of the audience. I thought, okay, what would they like from a conference? It was Share PLM, so I wanted to do something different. It had to be a human story as well, so I knew there'd be a lot of technology presentations. I wanted to make sure I talked about the human side. My background, did maybe 20 years ago or longer, I did some acting myself. And also recently, my wife and I, we've been going to murder mystery dinners with neighbors. And so I had that in the back of my mind while I was thinking about the presentation. I thought, you know what, I'm going to do a murder mystery presentation because I can also then get not only audience participation, but even get people up on the stage with me.
And yeah, that was the kind of roots of it. I think the, met up with some old friends at a Mexican and we had some Mexican food and margaritas and we thought about the different perspectives, know, engineering, purchasing, et cetera, and all of the day-to-day realities, you know, of their experience in the PLM and BOM and PDM world. And yeah, I think the, let's say that the suspect stories came together quite quickly. And then the actual challenge was, you know, limiting it to 20 minutes. You know, I think I could have on a, you know, an hour long performance. for those that are listening, although the there's a thread and my feed with every all the presentations, including a video of Rob's thing. So, but for those that didn't see it, what can you just give us like two minutes resume of what it was like? that thing? Yeah. So there'd been a product life cycle murder. So that means that the product had failed to materialize it, it had been killed before it made it into the public domain. And there were five suspects. So you had purchasing, you had engineering. next to you there, purchasing, think. Sorry? On my screen, he's right next to you. Exactly. we had, what was the, can't remember the, but anyway, was project management. had IT. You got to give them the real names because that was another area. You might have to help me here. Maria will probably know them. Kent Chapman was program management. Architecht was the architect. And the fifth one was... Yeah, had purchasing. there's a comeback to me. Anyway, they all their stories. They they obviously blamed each other for the failures to have the information they needed in order to bring the product to market. And the detective below materials, which was made by me. Yeah, Yeah. He yeah. So he then questioned the suspects and came to the conclusion and.
And the audience was the jury. they decided who they thought was guilty or not guilty. And then obviously I revealed the true cause of the murder and then went on to explain my logic for that. That's really, really cool. It was a lot of fun. It was very entertaining. I think you had everybody nearly crying of laughter as you were presenting all of the speakers names, especially when Oleg came out in his cap and his gold chains. You look like a rapper from Brooklyn, you know? I'm getting invoices until now. I just put the link to that in the... No, actually that's one of the interviews with Rav. I'll look for the post. Well, like, why don't you talk about your, so you, you gave a really good speech, of course. and you're a regular at these kinds of things. How, you know, how did you perceive the event? How, well, how did you're the one, one of the people that traveled the furthest, right? You traveled all the way from Boston. So what is it about Maria's pitch that you said, yeah, I'm going to jump on an airplane. I'm going to go to the middle of nowhere, Spain, and I'm going to do this conference. You don't, you don't want to know about this. actually, we do. But I think, you know what it reminded them to me all the story about this event reminded me the event that now in the history, but I really like this event in the United States. was called Congress of a future for future of engineering software coffers. So I don't know how many of you remember this event. But the story of this event was very interesting because three people were traveling from existing events. And they said that the most interesting part in that event was actually time that they spent on breaks. during the breaks, they were able to talk. Yeah. So they were spending time during breaks and conversation. And then they said, unfortunately, we need to go back to session and listen to all marketing.
speeches and, you know, presentations. So they said, let's create an event that will be focused more on conversations and more communication between people. So it was amazing event. It's unfortunately dead before COVID. And I think there are some ideas how to recreate it again. But when Helena and later Maria were talking to me about humans, said, there is something. that is desperately needed in this space because I can get any presentation online and we can put all materials on YouTube and like everything is available today. can again, now even more, can ask chat GPT. So you say, why would you go anywhere? Even the US ask, why do we need to write blogs? Because kind of we need to rethink this as well. So I just found the idea of rethinking. of how events should be deeply important. I just said that I will go. But by the way, speaking of Boston, I heard from some other people stories of traveling to this place of Spain. And I think my travel was pretty smooth and some people were going. I mean, if I only will take your story. Michael about going to the airports and sitting. think my journey was much easier. you see, so was my own stupidity. lot of that. Yasser, you gave a really great presentation, that we really got to enjoy most of, but I know you were your passion about, mean, that's one of your, your passions as humans and PLM. And you just told us how you were fighting the Dutch government about humans and government too. So important thing. Do want to tell us a little bit about your talk? Yeah, well first of all, think, uh, CRPLAM is the modern version of me. mean, uh, Oleg, uh, James White on the chat saying he can't really hear you very well. So there might be a problem. I can hear you. And there is one question in the chat about AI technologies, but since that was Rob's workshop, I'm going to let Yoss open before handing that one to Rob. Okay. So don't worry, Volker, we're going to get to you. Okay. Good.
So as I mentioned, I mean, I love working with shared PLM because we share the same vision that a lot of transformations need to happen through humans. And I consider them my younger version. So we are in a transition and it will get better. And if you look at my presentation, I realized that I always tell those stories about moving from coordinated to connected and now AI enabled. And I thought, but what does it mean for humans? And then looking back, actually, We are going now through a real social transformation. The first one was the agricultural, when we started to I would say, settle ourselves in small settlements. Then we had the industrial revolution, where we moved to cities and factories. And now we are in the age of digital and knowledge transformation, where society is no longer relying on human knowledge. We also have external human sources, and they're going to work with us. And that will have an impact on... the way we work and on ourselves. And my big advice, the human part on that, the technology is there and we have to learn to use the technology is that focus everyone now on your unique human capabilities. Learn and be ready to continuously learn. What you learn today might be obsolete in a few years. So be curious, curiosity is crucial and try to analyze what is happening. And I think the last thing is... communicate with people and that was exactly what we did in the conference. We need to have this communication skills between people to innovate and that's also what companies will be looking for in the future. They don't want the nerds to sit behind somewhere screen. They want the people that can combine this knowledge in communication. And in order to do so and this is also what we discussed with some other companies later, you need to set time to learn now. It's not that you say end of the year I'm going to training, no, it's a continuous process. So for example, spend half a day per week on learning, on reading and digesting. And that was my main message. It's us, the people that have to, change our behavior. That's really cool. Yeah. And I agree. It's the, we've gotten out of the habit of that self-training because we just have chat GPT and we could just ask the question, but it's really important that keep our
brain moving and thinking about it. But the ChetTPT will not innovate us. No, it doesn't innovate. It just rehashes what it already knows. But since we're on the top of AI, there was a question, as I mentioned, from Volkan Ceneci. I hope I didn't massacre that too bad. Obviously a Turkish name. How will AI technologies transform PLM processes? And what will be the most significant impacts of this integration on the future of PLM? kind of a big one, but that was more or less what we ended up discussing in your workshop, Rob. Yeah, that's right. So we ran a workshop as part of the technical tapas and lucky enough to have yourself, Michael and Oleg in those sessions too. And we looked at AI in digital engineering and PLM today versus what it would look like when it's mature. just maybe I start with what it looks like today where AI sits outside of PLM and it's a separate application that people are using to help them with some administrative tasks, whether it's data checking or looking for duplicates, et cetera. But when it reaches maturity, it will be fully embedded within the digital engineering environment. So you won't be able to tell what is actually a PLM system and what is AI. architecture, you will have it thinking and operating like individuals within that space. And I know, for example, Oleg, I'll probably pull you in here. You have some kind of thoughts about how AI will be operating, you know, especially with, you know, federated architectures. Well, I don't think you really know and I think the question that Alex so to stop you with the audio we can't hear you very well Can can you hear me? It's something with the microphone. It's very low. It sounds quite low compared to everybody else. It's on the top. It is on the top. Let me let me try it a little bit more.
But the anyway, just just while Alex fixing the audio, you know, we still we said that humans still need to be in the loop and be the ultimate decision makers. Although, you know, I will have the power to absorb vast amounts of data and present conclusions. It's still going to be down to humans to make decisions about the course of action, etc. Michael, what else did you take away from the session? was your, because you deep dived into engineering change management, for example. Yeah, that was an idea of one of the, actually two of the people in the session. we got to thinking of the problem. Well, one of the, one of the issues with the major issue we have with chat, GBT and all these things is it's not deterministic, right? You ask it the question twice and sometimes you get to completely orthogonal answers and the hallucination is a huge problem. The other problem is it just wants to make you happy. Is that answer great? Can I give you another happy answer? And so I don't believe, I don't, I don't think it was me that had this incident. I'm not that smart, but one of the people in the group said, well, what we need is a doubting version, like the, the murder bot version that's got anxiety and it's like, I'm not sure that's going to work. Are you sure? Can we check that? So you might want to dialogue. You might want another AI that's trained to like not believe anything like a scouting Thomas versus. is the like the really nice, okay, everything's fine. And can I give you a coffee now? You know, kind of version. It makes me think on we had in the past, the unconvenient truth. Now we will have the convenient truth in the future. And we don't. So I thought that was, it's sort of like we need a devil's advocate, advocate version of chat GPT, you know, because, and another thing I mentioned was that as we delegate more and more things to AI, Uh, we're taking a big risk because when those things, when something goes wrong, what guardrails are in place that we can actually reproduce the series of decisions that were made that led to a disaster. There's no standard for that. You back up all 3 trillion parameters and the 7,000 prompts you did to get to that. That's also a problematic. And in my opinion, how's your audio now, Oleg, maybe we can get your feedback. For some reason.
This application is not allowing me to switch the microphone. So I was trying to switch to another micro. But maybe just move closer to the one you're using. no, we just lost you completely. Switch back. was it. looks like can you hear me now? That's fine. Okay. That's right. Maybe I, maybe I just need to I will try for some reason the application is not allowing me to switch microphones. as trying to switch to another one. Anyway, I was trying to go back to the question that Rob mentioned about AI. I think this is something that we are trying to figure out is how actually AI will help. Just give you an example completely from non-PLM space. If I will ask you how many times you go to chat GPT or similar tool instead of going to Google or search. If you think about this, that would be the indication of the change of how we accessing information and how are we getting answers. For example, I know in many articles yours is saying we need to switch from documents to data. So if we go to Google and search for something, we're getting results, think about them as documents. But I mean, paying pages in fact is document. So we're getting to page and then we need to figure out in this page what we want. So you probably noticed, I'm pretty sure because I did as well. Is that Google sometimes starting to give you answers, even if it happened like three years and five years ago, starting to get some answers. So, mean, that was the early signs of AI issue. you would like to put it this way. Now, if I go to chat GPT, I can get an answer to my question and judge GPT is pulling information from different sources, especially now, but it get access to internet and many other tools.
If you think about AI infrastructure and AI agents, and there are different tools that can be used by agents, we're just raising up from the level where we're requesting the list of documents to the place where we are getting requests about some data and answers. And this layer is actually performs all these requests, performs all this logic and presents information us in the way we want. So that was my kind of main message in my presentation. I said, we're getting from get access to data to getting answers. And I think this is another layer that will be filled by AI, different AI tools that will tell, give us the answers rather than returning list of documents or list of some other data. objects as we say it in PLM tools. And this is how we will change our behavior. It is really a transformation. yeah, absolutely. There's a couple more questions coming in that are very relevant to this. I'd like Maria, she's been quiet for a while. There's one here that has a bit of a human angle on it. Maybe AI could be used to better understand customer requirements and assist in the setup of PLM systems and decision points are left to the human mind. think that's even one of the core things that you guys do with Shared PLM, right? Yeah, I think so. think, well, AI is obviously great when you're, especially when you're in a discovery phase with a customer and you're trying to figure out, know, first of all, what is their system landscape? How has adoption got in the past? What has been successful? What hasn't been successful? What's worked well? What's not? I think AI can be great in some respects to help you better understand the industry in which the customer is in, help better benchmark what systems are other companies using within the same industry, what sort of processes are they following, things like this. However, I still believe that AI can only go so far. So it's definitely when it comes to kind of more of the human side and bringing people.
on board with an organisational change like the implementation of a new PLM system. Ultimately, AI tools like ChatGBT can give you some kind of very high level overviews, whether that's of ideal processes, whether that's of ideal kind of onboarding strategies. However, ultimately, one thing that we feel very strongly about at SharePLM is that to really get people on board, they have to really feel like they are a part of it. And to make them feel like they are a part of it, you have to really take the time, go down onto the shop floor, speak to the end users and understand what are their specific pain points and what are their specific kind of wants and needs out of a system. And I think that's something that AI can only go so far because, you know, they're not the ones going down and speaking to the users and finding out kind of what would make them want to use the system and use it in the way that's, you know. the business wants them to. So I think AI can definitely support us when it comes to helping organizations with their transformation projects. But I think also we can't lose the human aspect. I that was the whole point of the Shared PLM Summit. That's why we wanted to get everyone together. And even with the kind of growth in AI and its capabilities, it's still important that we actually take the time to connect face to face and on kind of a human to human level, as opposed to the human versus machine. What do you think about that, Yeah, does AI understand culture? That's where we need humans. Imagine this. At some point, you'll be having a conversation with a chatbot where it'd be hard to differentiate between whether that's a human or a machine. And so, one of the challenges I was talking about earlier on with customers, let's say you've got to speak to 50 different people within the company and find out what their requirements are, what problems they're having with their current system landscape and what they would like in the future. That could potentially be quite time consuming for the team doing the work to get that information. But imagine you could deploy a chat bot, which would have chat interactions with those people at a time that's convenient for them and that would ask them the right questions and would modify the questions based on the answers that they're getting and even propose kind A or B scenarios.
you you can, the quality of the information you could gather would be exceptional. And that is using machine rather than, you what would take days of people's times in the past. Yeah. traditional consultant is gone. Yeah. I think he is there to form, create that questionnaire that Maria wants to use to go on the machine for, but a human needs to ask those questions. But then you record them with auto-data. otter.ai if the person's okay with it and then you get a resume from the AI. Exactly, yeah. Go ahead, I think if you said to AI, imagine you're a consultant and this is the end goal you want to achieve and you're going to ask five questions, put five questions to these people that you're going to be speaking to, whether it's engineering, manufacturing, whoever. I think you'd get pretty good results. in terms of the kinds of questions that you would ask that a consultant would develop. Obviously, when it comes to interpreting that information that you get, think then it is the human value. I think AI can do a lot of the heavy lifting. Maybe an anecdote on this. I was last year involved in the software selection for a company, a type of startup with four potential vendors. Up front I asked JetTPT, what would you advise for this company with these characteristics? And actually the end recommendation on JetTPT was at the end what the people chose also, without having shown them the results before. We never influenced them on all this. Yeah, and I think another great use of tools like Chats GPT is creating kind of user personas as well. So when we're working on projects, one of the kind of first activities we do is first of all, making sure that all key stakeholders and impacted users understand what is kind of the vision behind this, the implementation of a PLM system, what is the business case for this, but more importantly, to break it down onto a user level.
tools like ChatGPT can be great to ask questions like, I'm Angie, Angie Earing, whatever, I'm Jean Earing, as Rob called her in his workshop. And this is how I've previously been working and we're now implementing XPLM system. What are some of the pain points that I might be feeling? What are some of the emotions that I might be feeling? And actually this is a great way of kind of contextualizing before you actually speak to a lot of users to actually kind of put better put yourself in that position and kind of already kind of build out those user personas before you get the information from the actual person to better build those out as well. So I definitely agree that it can really help and it can really kind of make a lot about typical consulting work a lot more efficient. But I still believe that, you know, we have to find a happy medium between both the connecting on the human level and then using these tools to actually make our job even more efficient. Vinod Kadam asked about, hello everyone, there's so much everyone in AI in general, but where can we connect and contribute to use cases related to PLM across domains, different projects, like implementation, customization, migration, integration? It sounded like for Maria, that was like, you were talking about, we're interviewing users and we're seeing what they're doing. So now we pass the next phase where we're actually going to start implementing You know, how do we, where are the use cases where we use the AI in the most efficient way to keep the humans in the puzzle, but get rid of the really boring, tedious, I that's what we want to use it for, right? Is getting rid of that stuff. mean, AI has been great. So when we've moved, you know, kind of from the initial phase, we're moving into more of the implementation phase and we're starting to onboard users and train them. AI is great for content development as well. Of course, you have to take it with a pinch of salt. And of course, you need somebody with the technical knowledge to actually go through and make sure that the training materials are accurate, but actually as a way of making training content development more efficient.
creating something that we used to do in the past was we used to, if I wanted to create a tutorial on how to perform a specific action within a PLM system, I would record the video in the system and then I would have to record the voiceover myself, which would take so many tries to make sure that you don't mess up the script, et cetera. But now there are so many AI tools that you just input the script and it will give you the voiceover. Creating avatars has also been something that we've done as well when it's come to training programs. And similarly with things like chatbots, as Rob was saying, so once users kind of have already been onboarded into the system and had their training, having a chatbot that they can ask kind of day-to-day questions to regarding the system, regarding certain processes and get the results immediately via a chatbot. is also something that we're seeing a lot more of now as well, as well as things like digital adoption platforms that kind of lead you through the system as you're working on it. So yeah, we're really seeing AI's kind of dominance throughout the various phases of PLM projects. And then a company that wants to implement AI capabilities. I think that's where you see now a lot of discussion and thinking. Maybe Oleg, you are close to that also. Yeah, you know, it's very interesting to listen to how many diverse directions and applications we are taking it. So I can bring kind of my perspective from a specific side as a, what we do with OpenBOM. And I also can bring a perspective of writing maybe separately. But speaking about tools like OpenBOM and what we do with... AI, what is really interesting to see is how tasks that were very hard to automate before today is becoming much easier to implement and automate them because many, many, many tasks that we perform in the, the, in the customers and customers are getting data, they're retrieving information, they're transforming it.
They're moving it to different places. It's all can be described as a particular, I know everyone is using word agent now, but those are activities. And those activities can be really automated in the way that which look like more like human. in the past, we defined workflows. And we said those workflows are moving messages in inbox from one person to another person. And today it's getting some sort of mix. But some of these activities are operating with people and we can be much more human just asking what we want. This is what the agentic workflows are allowing us to do. But on the other side, having much more powerful transformation and information. So the information is getting very, very easy, transformable and used from different tools. I think this is where we will see a lot of breakthrough in the future. And we see those implementations are happening really well and lot of experiments are happening that will just simplify work for people. two kind of complimentary comments. Benedict Smith, who all of us have probably chatted with at some point or another, he said that what Rob, you described already exists inside Accenture. call it Lucy. I asked for a link. Maybe he'll put it in the chat for us. And then my friend James White, he said that he agrees with you, of course. And he says he would use discovery to survey hundreds to thousands of engineers to learn the as is of a PLM system. and a harmonization project because he's been part of PNM projects at Accenture and they could never trust that they got the entire as is, right? You only got the very small portion of the two or three people you interviewed over the many, many other people that were involved. So I don't know if you guys want to react to that. Well, when I hear the as is, I always say we need to work on the to be.
Because sometimes you lose too much time on studying the as is and you really need to first put your mark on the horizon and say, there I want to go to. And then probably everyone has his own point of as is and perception on how to get there. And that at least that's where I focus on. this is, I think this is where we get into the territory of, how creative can AI be so that, you know, you could use it to document the as is, but when you you know, start to think ahead to the way you want to go. You know, how useful is AI there? And I think I'd argue that you actually need more human creativity and vision. And that's where our value is. If I can jump on this, Michael, so I want to bring some examples that you might know from kind of human behavior. Do you have some stuff in your house that you don't need and you put in the box and then if you don't open it for 12 months, you probably don't need it, right? So that's kind of many companies behave, you know, they have a lot of data that they don't know that it exists because they don't know how to even access this information. So I think what we are going to see in this advanced data management capabilities today, I'm switching back to tools, is that if we will have an easier way to crunch all information using much more flexible data models and bring it in the place where we can actually get some answers and results, we might discover this information so we can move to to be like you suggested because the agents and the large language models and the new data modeling tools that we have, they already will absorb this information, but we don't even know how to search for this information. we know if you know the right keyword, you can find something in Google. If you don't know the right keyword, you will never find it. So it's a kind of discovery. I'm hearing from people today that say, no, I was talking to Chet GPT this morning.
And this is what I want to discuss. I mean, I want the same to start happening with the tool. Like you're suggesting, I talked to the new AI tool from PLM vendor and now I know what I want to be. And then there was another two questions that are sort of similar. was a James said, if AI solves the PLM murder mystery, How do engineers know if the guilty party, if it found the guilty party or not? Sure. Trust but verify. But how can we know the truth without doing all the research ourselves? And that's similar to what Dan Theoden says. And I got to go shout out Dan, probably the first guy to run an MCP with Eris underneath and put it on GitHub. Bravo, Dan. He just demoted that to me about 10 days ago. Pretty nice.
So in that context, maybe I'm seeing that in the past we had single source of truth. Yeah. We, we, we at least broken once in a while. Now you're gone. And he took Maria with him also. Someone needs to fill the gap. So let me continue. So I think in the past with old PLM, we would talk about the single source of truth because we had this system thinking and now we are in the age of federated PLM. Now we talk about. single source of change and the nearest source of truth. And I think with AI, are probably in the most likely truth because we we don't own the truth anymore. We and we can't even control it. And I think those are the for me to the three taglines to tag every generation of PLM development. Michael, welcome back. Yeah, noticed him pulling. So Dan said, data quality is often the main focus when training LLMs, but with PLM, AI could also interpret ambiguous or incorrect data and still arrive at the right conclusion. Perhaps even better than a human could, this introduced both the risk and opportunities. How do you recommend we manage the uncontrolled interpretations or humans always remain Q and A? for the AI suggestions, hence they will never actually be fully autonomous. I can actually give the answer to this. think it's available in all best practices, as much as we can call it best practices of AI agent development. There is a stage in every AI agent that it's called guards. So you develop some guards in this agent that will not return wrong data. So I think some guards development needs to be done in all this AI tool development. So that's, that's where we are. We probably will need to learn more about how to do it, but that's, that's, might be the way to do it. Guardrails is what you mean, right? Okay. Yeah. I mentioned this during the workshop, but I went to a really cool conference with a cab, Jim and I, and there was a woman there from Oxford university that was talking about, she was a robotic specialist and she said she was pushing for an ISO standard.
for an electronic black box for robots so that when a robot that's interacting with humans does something really stupid and someone gets hurt, which will happen someday, we can go back and figure out like a black box on an airplane, right? And I'm thinking maybe we need the same standard in manufacturing in general, in PLM and manufacturing, because we're gonna, again, introduce this stuff and when something goes wrong, What is the what's, you know, what's the root cause analysis procedure that's, it's gonna be tough, right? Even just when you're trying to, I'm already, I'm already using like four or five chats at the same time between Claude and Ms. Gail. And I even get confused, like, which one was I asking which question, because I'm waiting for the other one to answer and the deep research is taking 15 minutes, I've already asked 17 questions on three other ones. I mean, this assumes that humans make good decisions or that we're a good judges. I think we've got so much bias and we attribute so much good fortune to stuff that we've had no influence over at all. It's just dumb luck. To assume that we could somehow make better decisions is maybe misguided. But still, mean... you know, ultimately as a human race, we should have some control over our destiny. So I agree that we need guard rails and, you know, black boxes and all the other safeguards. But yeah, at some point, maybe as a society, we will evolve and say, you know, put faith and trust into machine intelligence over our own. you know, we haven't overcome selfishness or any of these other human traits. So, you we're probably unlikely to. then there was that paperclip story that the guy in my group mentioned. You tell an AI to make paper clips and basically kills all the humans because it's easier to make paper clips. Right. So you get the T2, you know, it's always a T2 nightmare that that you get worried about. Christian Nyingger asked a really cool question. I like this one. I'll throw this first to Maria because she has she's been quiet for a while, but I like this one.
Has AI as a render tool been discussed, something that will render the current state of the bomb on all engineering attributes, weights, tolerances, joints, and connections? We know that engineers have shortcuts to get that, but can an AI be trained to present reality? I feel like that's a question I need to throw to either Oleg or Rob. I think their experience will probably answer that question in a much better and succinct way than I could. So I'll pass to Rob, and then maybe you can pass to Oleg. I imagine this is probably in Oleg's backlog and it's probably only a couple of weeks if it isn't already part of OpenBOM already. Yeah, thank you for credit. OpenBOM is answering all questions. Not there yet, quite honest. But I found it interesting how now all problems that we are aware about and all issues now somehow attributed to AI. think it's a kind of... presented as a magic. think it's just what I see it as a better way to communicate with data. I mean, this is like, this is not magic. I think it's just a better way to get data and to interface with data. So now we need to build mechanisms around them that will tell us like how to interpret this data and how to use this data. I will give you an example. I can write a prompt that will give me a cost of part out of ggpt, just a number. So I can put the detailed description of a screw, characteristics, and everything. I will get a number. So it will be based on prompt data. Can I trust it? That's a really good question, right? Because it relies on so many sources online. And the chat GPT does the best work to go and, you know, index all this, all the sources, got this information, can be trusted and use it to a certain degree. Yes, because we go to Google, we find some information and say, here is the data. So chat GPT probably does it as good as we do, but can it be trusted for some decision-making? This is where we need to have a better tools and we need to have a better data extractions, information and more validated records.
because I want to have a tool that will on the background of the communication with AI agent, will go to the all invoices that we purchased this component in the database of this company and get this result. that would be correct results. So we don't want the random information about this. But again, the same agent can get communication aligned. So that's why I'm saying the scope is important. So it's just a way of data extraction. we need to control it. Maybe this one will work better for Maria. One of the projects we, this is from Vinod Kadam. One of the projects we started as part of co was social PLM. Now it looks more sensible to make it work as all the elements can be integrated properly with inner communication. So, so in other words, how, do we use AI to create this social PLM, which is sort of the things you guys bring people together. I mean, ultimately, that's the whole purpose of using PLM. And it's one of the main kind of selling points is that, you know, we have a single source of truth. We're all able to collaborate in real time and always have access to up to date data. I've learned from you, Juss. And but of course, you know, you need to in order to be able to make the most of those kind of functionalities of a PLM system, you need to really convince people of the importance to collaborate and to better communicate. So that's one thing that we focus on a lot in many of our projects. So actually helping organizations with their communication strategies, how to actually bring people together, not just within the PLM system itself, but actually outside of it. So looking at things like creating user groups, also creating kind of communities of practice. How do users best want to receive information? What channels? And so I think it is definitely important that we give, you know, once a system has been implemented, we still go back to users and we still give them the opportunity to have a space where they can come provide feedback, lessons learned, and even just ask questions. Because, you know, we don't always want to be asking chat GPT for advice. Sometimes we do want to come together and ask a question in person. And so yeah, Josh, you can add to that. Yeah, because let's agree AI is not social.
Here you go. You can't take AI down to the pub and have a pint A glass of sherry with your AI, whereas I can go to the SharePLM Summit and have a glass of sherry with everybody. I just want to remind everyone, if you forgot, what I just found, I wrote an article 15 years ago. It was quite a lot of debates. Can we design products on Facebook? So that was back in time when everyone was excited by Facebook. And said, we can bring the same environment to design products. I mean, this is how everyone like 15 years translated social system. Now all systems now have this chat and the capability to create the threads of the discussions between the users. I I think most of tools today have it. Did it create more social systems? I don't know. Just, just, just wanted, just wanted to, just wanted to remind that. We've been there already with another technological loops. With all of these AI discussions, we talk about the upsides, which there are risks as well as upsides, but there's a cost as well, not just the energy cost, but for example, there's a cost to implementing PLM. It's not that PLM is a magic wand and you just apply it and everything's great. It's the same way there are. need technology, you need servers, you need people who are going to do the coding, need strategy. So there's a whole lot of work that goes into this. And so I think that this should also be part of discussions around AI. Otherwise, it's all just nice to have and sweet dreams. Someone asked, but I think that we need a whole webcast for that about what are the classic PLM vendors doing in AI integration today? I mean, can say that AI has been in generative design for a while, right? 3D printing is primarily ML based and there's a lot of AI that goes into a sub D, sub D modeling and implicit modeling. But in terms of PLM, other than the chat bots that were announced by propel and that's a somewhat vague 3D plus universes. I know
vendors are working on it, but I'm not sure we're really at the same level as some of the other industries are, particularly CRM and industries that are more social by nature. I mean, I've seen messages like AI is an adaptable interface, that the menu changes based on your work. I think this is not really AI, it's physics. So I think the jury's still out on all the things we're going to be able to do. That being said, more technical question someone raised, Mohamed Asghar asked, will ALM be integrated with PLM and acts as a single component? ALM and PLM? Yeah. For me in the long term, they are the same. But is it just a document or does it be... ALM is the Asset Lifecycle Management in my definition of ALM. That's the question. Because with connected products, come close to... products in operations, assets in operation. So there the concept merge. If we talk about ALM and PLM, again, we should take care that we are not talking about systems. We are talking about an infrastructure and in PLM infrastructure, there should be software and there should be hardware because it should be data-driven model-based. So maybe the question can be refined, Michael. We also feel like we now need to share PLM Summit purely on AI and how humans are going to be interacting with it after this session. As long as it's in the bodega. James points out that at least one vendor does actually have an AI bomb ranking capability with sourcing costs, weight, compatibility, 85 % rankings, alternative and raise it to 95 % results. So some of that stuff's out there but... It hasn't penetrated a lot deeper than some of that superficial stuff. And another, so another, we've got about, we've got nine minutes. So can ask this other one from Dirk Alexander Molitor who's at Accenture industry XES. Are you aware of any knowledge graph based, that sounds like an Oleg question. Are you aware of any knowledge graph based network of engineering artifacts along the product development lifecycle? For example, connections between requirements, architectures, CAD files, simulation results and test cases.
for chain impact analysis. That's an only question, right? mean OpenBOM has knowledge graph, but I don't know what was the question really, if there is a knowledge graph. Are you just connecting the BOM or are also connecting simulation results, test results, architecture requirements in order to do chain impact analysis? That was really the question. I convert this to an MBSC infrastructure, which is necessarily the tool. It depends what line we like to use. There is a knowledge graph, we call it product knowledge graph, sometimes manufacturing graph, but this is the technology and knowledge graph technology that runs underneath of OpenBOMB. But if this is a question about technology or this is a question about implementation, in particular data, that's actually where it's made it. It sounded like a technology question and I think that's very much digital twin. technologies in it because that's effectively your digital product twin.
Agreed. Well, it's been a great exchange. don't know if anybody wants to give a last word before we say goodbye. You want to start Maria? So, well, I think we moved slightly away from kind of the recapping of the Sharefield of Summit, but this was a really interesting conversation. But I guess to bring it back, I'd like to ask you all a question or maybe even just less a question, more of a comment. But what would you say was kind of your biggest takeaway from the shared PLM Summit and maybe what are you looking forward to seeing more of next year? Maybe we can start with Joks. Well, my takeaway was I've never seen a conference in the last 10 years with so much energy spin-off, where people were so happy that at least they were talking with humans, I think also in a very human-oriented atmosphere. So I think that brought a lot of energy instead of the the list of sessions to attend. And like Oleg said, also the other conference before COVID. Oleg? Yeah, think just echo what Joss said. There was a lot of energy in conversations and I think that's the switch that we need. So it's less about... technology and more about how we can use this technology with people and how it can help to solve problems. This is where I see it as the biggest thing because it was very much technology oriented for many vendors and vendors typically say, my technology is better. mean, that's obvious, but I think the real conversation about real problems, this is what was really really interesting in this conference, in this event. How about you, Rob? Well, for me, it's unity, unity in the fact that PLM is a team sport and that everyone that was there was focused on the same goal. so, yeah, as you said, we came away, everyone came away really energized, but in a way that you felt part of something bigger. And actually, there's been a lot of collaboration that's come out of the conference. And to answer Maria's question about
you know, next year, I'm so excited to see how this evolves because it really is unique. And I think there's going to be a lot of people who hopefully have, you know, sensed what it's like to be there. And I'm really excited to kind of see how this evolves and meet the other people that are going to be part of it and, and, you know, see all the new presentations, et cetera, and the new social interaction. It's going to be fantastic. Yeah, I think that's a really good point as well that you'll make. think the feeling of community, think what was so great about this conference is that it wasn't discriminatory in any way. know, it was completely system agnostic. We had people coming from using a variety of different PLM systems. I think we pretty much had all of them represented by at least somebody. And but it was great to see that, you know, everyone was able to connect over similar. problems they had when it came to kind of human resistance to the tools. And so I also loved that level of kind of unity that, you know, people, the way in which people approach change and people's kind of natural born resistance to it is the same, no matter kind of which tool it is that they're being presented with. yeah, I forgot the diversity, Maria. Yeah, I was gonna say that. The parity of having almost an equivalent number of men and women was fantastic because we got a whole nother perspective than we usually get at the traditional PLM 90-10 conference. Annali Ufkagan says, where are all the women in the chat here? It's amazing to see so many powerful people. It's true. It was absolutely astounding to see so many brilliant and women and men, but also how everybody was just together to, in this community spirit. And I think next year I'd like to see that again. And like it just to be expanded. And, you know, I thought, I thought it would, with the workshops could have easily been a whole day. Like we, with Rob and I, we were felt really constrained by the time and then we were sort of, my God, everybody's leaving because you know, the, the fire out. And so we were always like, you could easily do two days of,
sessions and social networking and then one day of workshops where you get a whole day to work on one of the three problems that you guys set up. So that's been that's been very common feedback and now I'm wondering how is everybody going to convince that. manager or their employer that they should take three days off work. Maybe we have to do it on a Saturday. We're just going to have to make sure we have a really compelling agenda, just as strong as this year. A bit of a sales pitch for you guys here, but I think it's value adds. So it's like an accelerator. So anyone that attends the conference, more days they spend interacting with other people, the more powerful they're going to become in terms of their impact on their own business. So it's actually, there's a huge return on investment of going to the conference. In that case, we'll make it a week next year. We haven't formally set a date. I think the next meeting of some subset of this group will be talking to a couple of analysts that I know, because I think it would be interesting to understand how they view the world and what that job is, because it's sort of mysterious. And I think we could probably learn some things there. And I would really appreciate in the comments feedback from, you what do you guys like about these webinars? What other subjects you'd like to do? I think maybe we could dive again into the AI digital thread as a service thing, because Rob missed out on that first one. So anyway, give us your comments. Thank you very, very much. And we'll see you in a couple of weeks or in September, because of vacation. Yeah, no, thanks for everyone who joined and thank you for the questions. Appreciate it everybody.