Driving Innovation with AI: Transforming Automotive Product Development

00:05 Welcome, everyone. Thank you for joining today's webinar on driving innovation with AI Transforming Automotive Product Development. I'm your host, Matthew DiCecca, Director of Product Marketing here at PTC. And today I'm joined by two of my colleagues as we discuss how AI is reshaping the automotive industry across vehicle lifecycle. With us today, we have a Ayora Berry, VP of AI Product Management and Michele Del Mondo, Senior Director Global Advisor, Automotive.

00:35 Let's start with you today. Thank you for joining us. Can you introduce yourself and tell us a little bit about your experience in the field of AI? Good to be here. Been with in the software industry for about 15 years. Various product roles. Product management, part of design product enablement. And what I do here at PTC is I work with the different business teams and corporate functions to lead our AI strategy, incubate new offerings and, develop common components that we use in our products, like technologies

01:06 that can be used by two or more products so we can drive consistency and scale. That's good to be here. Thanks. Great. Thank. Good to have you. And, Michele, we welcome you remotely. Can you share with the audience a little bit about your experience in automotive? Sure. Thank you for, for being here and and sharing my experience in automotive. So, it's more than 25 years, and doing business in automotive, thereof. 15 years in the supply of business, working as, at one of the top hundred suppliers in Germany

01:37 and then at PTC, always in the business development, role and as industry advisor for suppliers and OEMs. So overall, more than 25 years. Thank you for those introductions, gentlemen. Let's dive into it. Michele, I want to start with you on this first question. How is AI reshaping core automotive engineering workflows such as requirements management, design optimization, test case creation to really help accelerate vehicle development? So that's a good question.

02:07 So, I believe that AI is profoundly, reshaping the engineering workflows. If you look at the requirements management, design optimization, test case creation, they are heavily, manual processes. So you have to read requirements, you have to, do quality checks. And the same thing for design optimization, test case creation. So and this is where AI is really helping and accelerating, these workflows by helping the people,

02:39 doing repetitive tasks automatically. So he can give you advice on how to improve the requirements, quality for example, once you, write the requirements. On the other hand, if you look at design optimization, which is, which is a time consuming, task where I can help you to iterate much more faster and generate that automatically. And if we look at the test case creation, which is typically the bottleneck in automotive,

03:09 because it's on the right hand side of the so-called V modeling automotive, it takes you also, a lot of effort, reading the requirements, defining the right test cases. And I can help you directly by, by reading their requirements and, suggesting the corresponding test cases. So what what we see is really a so-called shift left, from the the model to the, to more iterative design approaches,

03:40 whether it's requirements design or test case creation and a follow up for you, Michele. I've been talking to some customers in the automotive space, too, and they've been sharing a concept. Where is it possible to collapse innovation times? So using your V diagram, for example, is there a world where you see AI systems taking requirements, doing the things you share like running for quality, generating tests, but also jumping over across the other side of the V and go quickly do say test validation.

04:10 Go run simulations. Do you see that as a possibility for AI systems? Do you hear it in the market from customers? That type of concept as a vision? Yes, absolutely. You are completely right. So this is exactly, what what what they are looking for. Let's say we we call it in the automotive, shift left. So moving from the right side of the V model more to the left hand side. So immediately when you generate code or generate a new design,

04:40 you want to immediately test and validated against all the regulations that you have defined, let's say, on the upper, or on the, upper left hand side from the requirements, whether it's homologation relevant and so on. But that's exactly the, the goal will to, to, to get this kind of information and to directly link from requirements to test cases to simulation, run them automatically. This is where the the the big savings are which typically is the bottleneck in automotive development. Yeah. Yeah I like that shift left

05:10 kind of phrasing I heard a similar one around like supplier like supplier management can we shift left the planning and the engineering so we get the right products downstream in terms of manufacturing and service. So it's a common theme perhaps around AI that with more insights and automation we can do more shifting left to address, you know, business needs earlier on. Yeah. And we'll get to the topic of product data. But in terms of this question in regards to use cases, can you talk about how that goes into service as well?

05:40 Yeah. It's interesting. So like across the thread of how data moves from requirements down to service, there's all types of opportunities for AI. And one that's interesting. That has been a tough nut to crack. But we're seeing more opportunity with AI is can we get that closed loop. Can we get for those companies that are servicing their products? Can we bring data around a faulty part, for example, and push that upstream or shift left like we were just talking about and enable product teams to look at that part failure and say, okay,

06:12 what would the new iteration look like? And at the same time, let's run some simulations on CAD models to see if it's got the right assembly interferences or not. Can we look at the property to see if it can perform well, but concurrently? Can we also look at inventory, connect to ERP or even potentially if it's a new part that we're going to service, look at, say service lifecycle management and use tools like Servigistics to look at where is the inventory at any kind of forecasting impact with those changes. So this idea of the closed loop, we see AI systems and in particular agents

06:43 as accelerators to those types of use cases. And there's two reasons why. Like one is on a technology front, these AI agents can now more readily access data. So we're kind of overcoming a technology hurdle that we've had. And then two, it comes back to an earlier concept, which is we're now live to automate things. AI systems can work independently to bring that data up. And bring those insights and do all those different processes that I just talked about. Yeah, going a little bit deeper there. And what ways does, you know, structured governance product data ensure AI outputs

07:15 meet, you know, the strict safety and quality standards, that automotive the automotive industry expects. Yeah. I mean, Michele knows this. This is essential, you know, in this industry, whether it's, you know, regulatory needs at a geographic level or just from, you know, industry standards. And I think, what's the good news here is that right now, there's already value in the software systems that these automotive companies are using. They're using ALM and PLM and, you know, CAD etc..

07:45 And these systems comprise their product data foundation, you know, so what what am I talking about here. So Product Data Foundation is the product data your engineering data like bombs and designs etc. stored, structured and managed in those systems. And what that's important for is it provides access controls and governance. And where this connects to AI is that when you embed I into those systems, you're going to gain the benefits of all that work you've established with those systems for compliance, for access control, for IP security.

08:18 So the good news is you have that product data foundation already to deliver on that. And now it gets better where we can now look at the use cases that McKayla was talking about. Like, can we look at AI as an engine to shift left and look at when I update a requirement, what's the impact on compliance? You know, is there a new regulatory needs that I need to have updated in order to meet these new requirements? And so think as AI systems, as checkers, whenever it's something new comes out, a new policy, it can check that and say, hey, this this has this impact on these requirements and go through that.

08:49 So there's two things that are coming through from a kind of, you know, product data perspective. One is that foundation and layering AI into the existing access controls and governance. And then two new use cases that are powered by this automation and insights. Yeah. I'd like to ask, you know, Michele, in terms of AI helping automakers and suppliers, you know, meet those evolving regulatory and safety requirements, how do you do that without slowing innovation on their end? As I mentioned, you can think about AI as being a proactive

09:21 agent and always listening to what's happening in the systems. If you look at these, these kind of, processes, regulatory stuff, you know, it's it's it's it's always document, you know, you you need to read documents, you need to compare documents against what you have done, against requirements against test cases, you know, and as I said before, that that's really time consuming and having a kind of AI. And I would call the AIS a catalyst for collaboration

09:53 because it brings together the different domains, you know, from electric electronics, software development, hardware development, which are typically working in silos. And I, or AI agents, they can be the, the one who talks the language of each of these domains, you know, and bring data together and so that you can ask the system with a natural language, about your problems and what you need and the exactly these kind of, of AI agents, they give you the answer

10:23 because they do not take care where to look in the data. As I said, we have the product data foundation. It goes there and gives you the answer. And this is if you embed I into the overall process. It's I would say it's also an embedded, safety and regulation process. And it's not something that you do after you have done everything as a POS process. It's it's embedded and it's a continuous,

10:55 AI agent for homologation, let's say, that way. And Michele, in your last response, you touched on an important topic of collaboration. Right. We know engineers don't work in isolation. Oftentimes they are siloed. You know, ideally they want to leverage data from across the enterprise, you know, efficiently and easily. How do you see I really helping to improve collaboration across teams and what's the importance of that? That's a good question. That so, and the very important one, as we know that, that, we,

11:25 we need to do or we need to work with different domains like, electric electronic software, hardware, but also further down the road, and a factoring and service. It's very important that, if you are using AI, you use it, as I said, before and to end in an end to end process so that you can leverage the full, knowledge that that the company has on IP here. And, I think AI is a kind of I would say, a kind of of, middleware which connects to each of these systems,

12:00 which, which, which gets the data, which provides then the knowledge, independent where you are an engineer and working only with CAD and PLM, and the same way for other people, like a software engineer more working with an ALM and, and, software IDEs. But also, getting the information, from the service department, from the warranties, you know, to get directly feedback so that you get, a continuous, closed loop system here.

12:30 So this is, very important one. And, I see AI as a catalyst for collaboration building on that. You know, we look at the market and different tools that are going out, the things we're working on at PTC and like our partners as well, there's a couple patterns emerging, like, how does that come to life in the software from, you know, exposing that data and experiencing that data. So one is more about bringing it together into a dashboard. So think like an engineering cockpit or, you know, this digital

13:01 thread cockpit where you have data that you can ask questions. And I like how you put that. Michele, which is what's unique here is we can now now ask those questions in natural language, you know, where is this information? What's the status of this project? It's very intuitive. Like we can't under, you know, under count how important that user experience is with interacting with these complex data systems. That simplification is huge in terms of speed and access to information. So there's that dashboard or cockpit and you'll have chat interfaces, but you may have some reports or ways to double click into other things.

13:33 That's kind of one view. That's kind of a unified view. But the other one that's also important too, is where you work in a product every day. Your requirements engineer, and you want to be in that system like code beamer. And in that system you want to access the AI system that's in code Beamer, but then tap into that data that maybe is in another another software system. That's where the agents, as you share, become the kind of middleware or the messengers to move that data. So that agent and code beamer could say, hey, I'm looking for this information. It knows it can connect to another system, say PLM

14:05 to look at parts and link the data between those two. And so there's two kind of engagement models are seeing. And it depends on the user and the need. But they're both very important in the future. Yeah. I want to stick with you for for the next question. You know, we talked a lot about, you know, the importance of AI. You know, what it looks like within an organization and the value that it can bring. But, you know, what challenges do automotive teams face when they do go to integrate, you know, these AI agents or different AI capabilities within their organizations?

14:35 And what are those risks? And and how do they overcome those? Yeah, no, it's a great one. And there's definitely a risk. And it's important to be aware of them and to to mitigate and around them. You know one way to slice that is let's use the software stack, very simplistic view like user experience, the actual like AI services. And then the underlying infrastructure because there's risk at every point and it's good to be aware of those. So on the user level, a key risk is around asking questions or executing actions and not getting the right answer.

15:08 It's all about quality and reliability. So what we're going to see more right now and and going forward is it's important to have humans in the loop to have the user be there in the software. And when I generate some output, you ask a question or executes the action. The person the user has a choice, whether they use that information or not. Click submit, save what have you and then it gets stored into that business record based on that authorization. So hallucinations and quality is one risk area. The next layer down is like how the agents operate in the technology.

15:41 And one of the risk is just around the pace of innovation, like the the speed of which AI technology is advancing in the market. It's massive. And so things are constantly changing. So you may invest in some infrastructure, you may go build your own software, and then the next month, the next three months the technology changes and you have to change out your system, change out the instructions for your agent. So there's risk around the technology burning up, having to maintain source software, especially stuff that you've custom built. And then on the bottom layer is more around like data and the models.

16:14 And here the risk is around IP security. You don't want your data co-mingling in your vendor software with other, you know, customers. So when you look at your your products that you're using, think about okay one, where is that data being stored? Make sure that's an isolated ten. And then two, if I'm using LM models, for example, and say the you know, the vendor is doing training of that model, like fine tuning it again, like use an isolated kind of pattern. So for example, in Servigistics when we deploy machine learning models

16:45 for a service supply chain type use cases, we have a foundational ML model. And then we fine tune it on the customer data. But it gets sent to their isolated tenant and their environment to maintain that security and risk. So there's risk all all throughout the staff that you can think about. And that's just one way to position it. There's also an interesting risk, which is just like the risk of not moving if, you know it's an important time, like the AI wave is here and this is an opportunity to take advantage of that. To get back to Michele's point,

17:15 like go to market faster, improve quality, reduce cost. And if you don't take advantage of this AI opportunity, you're at risk also being out competed by your competitors. Yeah it's a good point Michele. Anything to add on your end? Oh, I would take the same analogy, with AI, with the, transition that the automotive industry is taking from IC combustion engine to, electrified one. You know, the the technology is moving so fast. The battery technology is moving so fast.

17:46 You implement something, but in the moment that you implement something, you get cut by a new technology, you know, but, which, which might be more powerful. And this is I think, the topic that we have these uncertainty, what is the right one? When should I jump on the wave or should I wait or. You know, and I think, you can't wait. You need to jump on. You need to try and and then you have to figure out, how to move forward. But, I completely agree with all your, Yeah.

18:17 And to build on that, like, as I said earlier, like you're in a good position right now because you already have that product data foundation, that data that gives context for the AI to deliver reliable and relevant results. And then just right now, as a starting point in your journey of AI, like start with known use cases where you have, say, the AI chat type interfaces where you can interact with your data systems using natural language like start there and then look at assistance where they're doing specific task. And the reason why that's important is, like you, and I'm

18:48 sure many of you are trying this out like it's important because you want to build the wherewithal like you need to build your AI muscle. And so you need to be familiar with the technology, the use cases, so that when the more advanced use cases that have more significant productivity gains, you know how to make the right decisions. You know where the right technology to implement. So it's all about building your organizational muscle to around how to use AI in your business. Micheli, you know, with your experience, you know, in the automotive industry and seeing how it's grown and sort of changed over the years,

19:18 you know, what makes this industry such a good sort of incubator of of new AI technologies? You know, what differentiates it as, as an industry that can really benefit, from I. Well, if if you look at the, at what what's happening in automotive and I especially want to mention these autonomous driving, for example, or the ADAS systems, you know, automotive, assistance systems, they need to learn, you know. Yeah. And they need to learn fast.

19:48 If you want to drive autonomous, you have so many cases that that that the human brain, I would say, can't oversee everything, you know, and with AI, it can help you to accelerate, to iterate faster, than before. It can learn, faster. It can even, generate, cases which, which nobody, maybe a human thinking, doesn't think about. And this is exactly what's happening. So the automotive industry is, is is is in the, in the transformation,

20:21 phase where, speed is key. And this is where I apply, you know, with AI, the promises that that you get faster, you can leverage these things faster, which were not possible before because the systems are getting so complex. That that that you need a kind of assistance. And what assistance can be better than that? A good implemented AI solution based on all the knowledge our customers, have in their ALM and PLM and other systems.

20:53 Yeah. You mentioned assistance and automation, which made me think of a framework that we use at PTC. You know, the triple A framework. So if you are thinking about AI and its maturity and what it can do for you, a simple way to break it down is advise assist, automate. So advisor type agents are ones that are giving you information, helping you answer questions, summarizing things and then assistance. They do specific task in a workflow. They they execute a specific task. They don't automate the whole workflow. And then automation is about automating, you know, more or less many of the tasks.

21:23 So as you think about the market and what you're doing, that's a framework to think about. And within that context, most of the markets in that advise and assist space. In fact, the research says about 80% of the market is in the advice space. And the rest is in the assist with a smattering in the automate area. That that's a good point. You bring up, you know, a crucial framework to, to serve our strategy. So let let's touch on that a little bit more. Or if you can give us, sort of a preview of, of what does PTC AI strategy look like, both from a product perspective and sort of general strategy.

21:55 When we think about, you know, how we look at the intelligent product lifecycle. Yeah. And how I really fuels that from my use case. And product perspective. Yeah. You know, central to our strategy is the intelligent product lifecycle. It's all about embedding AI throughout all the different domains of the life cycle fueled by that product data. And within that context, there's three pillars to our strategy. And it takes a role of like a maturity model, where first is let's empower industrial workers.

22:25 So it's more about individuals and teams. And then second it's about connecting enterprise intelligence. So that's about organizations. And then third, it's about transforming product lifecycles, reimagining what we can do in R&D or that close loop type scenario. So on that first one, that's where we're putting a lot of energy right now. Like let's empower industrial workers, whether it's engineers or downstream and technician, to be super users of their software to have better access to knowledge, engage

22:57 that knowledge with natural language, execute specific tasks so that you can cut down on the non value, add time in your workflow and focus more on the creative efforts. So that's where we have things like we're investing in different AI systems in Creo and Codebeamer, Windchill and service bikes. All of our products are having these new AI chat and type of assistant experiences. The next level up is that connecting enterprise intelligence, and that's where we're currently doing research and active development. For example, we're working with Microsoft and with Microsoft. We're doing research on how can ERP and PLM agents

23:30 work together to reduce change process times. So in this scenario, imagine, you know, an engineer kicking off a problem report and then enter an agent workflow where now there's this change process analysis where the change coordinator says, let's look at the impacted PLM items. Let's also look at the information that an agent is collecting that is based on past projects. Here are lessons learned and simultaneously, let's also look over an ERP and say what's the impact on inventory and cost.

24:02 So that collection of information then enables the organization to make faster decisions, correct decisions so they don't have to change anything down the line. So that's the type of work we're working on, that kind of connected enterprise intelligence, I think more aspirational where we want to work with the market is like, how can we reimagine product lifecycle? How can we shift left on that V diagram to cut down R&D time so that networks of agents through automation can do some of that iterative design and development and testing. And that's an active area, you know, you know, to go forward to.

24:35 Yeah. One thing you touched on there that I like is, is I allows, you know, users of our software or other vendors to be super users within that space. And it makes me think of, you know, when we use ChatGPT every day or copilot, you know, it allows us to be more efficient at our job. Can you talk a little bit about, you know, how that how that's going to change sort of the workforce in terms of how people are, are getting their daily tasks done, you know, easier and more efficiently through I. Yeah. No, it's it's an interesting space.

25:05 I think there's near-term stuff where we are going to be able to use agents and chat bots to just be more efficient at finding information. That's, that's pretty known entity. In fact, there's research on that that's already showing that we're saving 30% of our time looking for information using AI systems. I think we're the future is going. That's very interesting as more and more like agents being part of the team and you're more coordinating a group of agents. So like kind of an agent manager, if you will, in in that future,

25:35 you're going to see more of the scenario where people have a dashboard, where they're going to have progress on certain actions that they have, that agents are executing. And it's more about oversight and kind of updating the agent system so that while the agents do their work, you can work with the rest of the team to to like find out like on the value add perspective. How can I decrease time on preparing for supply chain, or how can I accelerate requirements authoring? There's all these processes they can focus on by collaborating with your peers and doing more of the rich kind of creative problem solving versus

26:08 the mundane tasks. All the stuff that we talked about earlier, there's a lot of like documentation management in the automotive industry that's required for good reasons if we can, you know, offload that to agents so we can work on the design and the creative spaces. That's a much more interesting space for us as workers in the space. Yeah. And I think it's important to sort of look ahead and sort of, predict how how valuable AI is going to be and, how it's going to drive continuous improvement and sort of scalable innovation across the industry.

26:38 So, Michele, when you think about, you know, looking ahead, you know, what are some of those benefits that, you know, as products become, you know, increasingly software defined and connected? You know, where's the value going to be in terms of leveraging and implementing AI with into the product lifecycle? Well, that that's an interesting question. And if you take the two words software defined, what does it mean? The the product gets more complex, right? Because it has more lines of code. And if you take connected with us, connected means,

27:11 connected means at the same time, safety and security, I have in mind because, once you have a connected device, you have to protected against vulnerabilities, right. And and this is exactly where AI plays. It plays a huge role. You know, you don't need to have persons which scans your code against vulnerabilities. You could have, these so-called AI agents, which can do that for you. And, and going back to the to the very first question, if you implemented really,

27:41 end to end, as soon as an AI agents detect a vulnerability in your source code with a connectivity label or whatever, it can automatically also start a change request, you know, doing an impact analysis, how much, of my product or suffer code do I need to change? How should I change it and immediately sort of change workflow? I think this is, the, the if we look at the future becoming more software defined with products and more connected, I think, the better it is for AI to help us to assist and then, go,

28:14 at the end and to automate tasks, which are not automated today. And this is how we can be faster and can scale innovations, with speed to market. Yeah. Excellent. And, just to wrap up here, I want to give each of you an opportunity to, respond to this last question. What's what's one piece of advice you'd give an automaker, an OEM or a supplier who's, either looking into AI or in the early stages, sort of implementing it, you know, what's, the 1 or 2 things you'd want them to really focus on or hone in on so they can achieve value?

28:47 Yeah, this is one that comes up a lot, which is, you know, getting your data set in order for your data house in order. Another way to put it is like to mature your product data foundation. The the there's a there's a saying from the internet era called content is king and and that that meant that the more content you had in your web applications, the better your SEO was. The more content you had, the bigger library of content, videos or music, which means you had a better moat. That was more value add for your customers. Well, in the AI age context is king.

29:20 With more information around what is that data? Where is it? Who's using it, how is it being used that gives context for the AI to deliver reliable and relevant results. So the one takeaway I would share is that continue to mature your product data foundation, go from doc centric to part centric and PLM, go from isolated requirements and test to traceable requirements and tests go from 3D modeling assemblies to model based design in CAD, for example.

29:51 Keep maturing that product data foundation because that gives more context for the AI. And the good news there is that out of out of the gate, you're going to get value from that maturing of your product data. But with AI, you're going to get even more value from your systems. Yeah. Excellent. And Michele, how about you? What advice would you give to, to automakers? Well, I would add on what what Ayora said is, start don't wait. Get into your seat. You know, fasten your seat belts and start within. But, think really the process from end to end.

30:24 So don't do do not end up in, in thousands of parts. Try to think about how the, the, the whole let's say, role of an engineer might change with AI because you have to think, what what tasks they are doing today may not be needed tomorrow So and how to enrich these kind of, let's say, engineering activities, really thinking from an end to end process. Yeah. Thanks, Michele. Right.

30:55 So what we learned one, you know, start with a strong product data foundation, to pick some of the high value use cases, you know, that are going to be most beneficial to your business. And three, make sure you have, you know, a sound strategy upfront to make sure that you can go and execute on your goals here. Well, that about wraps it up for today. I want to thank you, Ayora and Michele for sharing your insights. And thank you to everyone who joined today's webinar. It's clear that AI is really transforming how the automotive industry designs, builds and delivers products

31:27 and is excited to help our customers lead that change. To learn more about PTC's AI strategy and solutions, visit ptc.com or reach out to your PTC representative. Thanks again for your time today, and we'll see you at the next session.