Hi, everyone. It's Judy Warner. Welcome back to this week's ecosystem podcast. Today I'm joined by Roberto Piazentini, who is the director of global marketing at Keysight DES, which is the software side of Keysight's business. We sort of have a friendly engineering riff to talk about and explore what he calls the most expensive engineering mistake. I hope you'll enjoy this conversation. I'll see you on the other side. Hi, Roberto. Thanks so much for joining me again. It's been a minute since we've had a chance to have a conversation on the podcast, and I'm glad to have you back. Will you please take a moment and just introduce yourself to our listeners and a little bit about your work at Keysight? Yes. So thank you, Judy, for having me again here. Your show, it's been a pleasure working with you for this past couple of years. So at Keysight, I'm the director for Global Marketing, the design engineering software organization. My background, I'm an electrical engineer. So and I've been working with electrical design, software design, you name it. I've been in the business for over 30 years. So that's a little bit of me. It's part of my hobby. I also really enjoy working with like software development, AI, websites, all sorts of things. But yeah, I'm an electrical engineer by nature. And a writer. I saw you post this. That's right. You're doing a little self-publishing now. That's right. But it's, but it's a self-publishing book around stories. It's a novel. It's a magic book. So all right. That's fun. Okay. Well, yesterday we spoke briefly. And when we finish, I think we both wish we had the microphone on. So I wanted to invite you back today and try our very best to recapture that conversation because I think it'll be really of organic interest to our audience and just kind of riff in about sort of the state of engineering. So Roberto, I'm going to start right in and ask you the question, what is the most expensive engineering mistake these days? Well, this is a really good question to get it started, Judy. And the most expensive is the one that you discover late, right? It's the one that you don't find as early as possible. So as simple as that. Yeah, that, you know, we've talked for years about the cost of late discovery. And again, I mean, people must get tired of us talking about the design silos or feeling it, right? Where they thought they did their part. And then, you know, a few stakeholders downstream, they realize that one decision they made, you know, stops the design. You have to stop, go back, or even at worst case, you have to re-spin the board. And so what are things you're thinking about both as an engineer and looking across all the software products that you're involved with at Keysight and sort of how to mitigate that practically? Yeah, I think as we started going this path of the most expensive problem, I mean, people have heard this all the time, right? It's really discovering late. But I think one aspect that people don't talk a lot is that I think that the complexity has really changed the engineering problem. I think that the engineers are not really optimizing today isolated components. We have all of these different physical pieces getting together, like electrical, mechanical, thermal, optical. There's software, right? We don't talk a lot of software as hardware designers, but without software, hardware doesn't do anything, right? Exactly. And all of those decisions that are made also impact later on how the device is manufactured. And I think that the complexity has changed from being purely an electrical problem, purely a mechanical problem. And what happens when you put all these things together? And as we start this conversation, the later you wait to start seeing how these pieces interact together, the more extensive it can get because it might be very late in the cycle. It might result in additional prototypes that you need to spin. It might be additional days postponing the release of your product. It sounds obvious, but it happens all the time, you know. And, you know, we all know then the finger paint, the finger pointing starts in, at least in my mind, Roberto, for sure, the complexity, I think, is driving it. But also, you know, I sort of came up, and you probably did too, in a more physical-based work environment. And the people who your fellow stakeholders, you know, could might sit in the next cubicle, right? And now the complexity curve has gone way up. And since COVID, the decentralization of engineering, which is always, which has been there to some degree, but now it's really decentralized and we don't look each other in the eyeballs anymore, or very rarely, it seems. So, and we keep talking about solutions like shifting left and then simulating the software. But I think for engineers, there's a lot of like, do I invest this much in software, you know, or that mechanical person's job is not my issue or whatever it is. So, you know, rather than just complain, I had a friend who said, let's not stand around admiring the problem, right? Let's keep saying, oh, here's the problem. It's decentralized. It's stakeholders and all of that. But like, how do we fix it? How do we make life better for engineers like day-to-day, Roberto? Yeah. And I think, and I think that you touch on something we talked about the complexity is kind of changing the engineering problem, right? And complexity is a vague word. And I'm a really black and white, simple guy, you know. So tell me, no marketing fluff, what complexity means, right? I think that the, and then a question always comes to my mind: why traditional development processes are breaking down? You know, complexity is always, it's a given. Products, electronics, we're always going to get more complex, right? I think that what is in my head, what's happening is that products operate as a system. You know, you have all of these different subsystems that together make this larger system. And the subsystems are touching base in all those different physical environments that we talked about. So the devices are becoming more complex, more physical domains. But teams oftentimes they continue to work in disciplines. As you touch base, I don't care about the mechanical part. We work in these disciplines, handing work downstream in our own silos. And the point here is that the failures actually happen in those gaps. When you finish your electrical design and you get to move it along to the optical interface and the optical side of the design, you say, oh, I wash my hand, my part is done. But then when the optical engineer gets and have to design it apart, there's some plays with the electrical side, maybe thermal and whatnot, that changes how the signal propagates into the electrical board. This back and forth, if you keep working on silos, I think it becomes complex. And this is something that if you follow the key side story, we've been investing a lot in the past several years in, since our early days, we started on the right side of the V, right? With instrumentation, with devices that make physical measurements. And then we started moving more towards the left side of the V, which is simulation side and whatnot. And we have invested in the most recent years, completing our portfolio of solutions in all these different physical environments. And we did it exactly because we are seeing that the challenge engineers are seeing nowadays is because of these gaps between electrical and mechanical, between electrical and thermal, optical and electrical, how the software plays when you put the software in there. So then you saw that keyside invested with that plan, for example, which is software test. How do you test the software that is running on your hardware? You might think that this is irrelevant, but if the software is not well designed, it might throttle your CPU too much. And the CPU could become hot, you know, and the heat, if it's not designed correctly, the dissipation, it could impact how the signals propagate in the board. So everything is kind of connected, you know. So then Keysight has been investing in all those domains. We have ADS for electrical, we have Photonic Designer for the Photonic side. We have now the thermal piece that we completed with ESI coming on board. We have now VPI Photonics coming on board as well, complete the system level of co-design. So we are very much attuned to making sure that we have this complete set of systems that understand the data between exchange between each other and the touch points between each other. And I think that this is how it can help engineers because you can parallelize the work, still have the mechanical engineering electrical engineer and whatnot, but the system behind the scenes making this work, helping them to share data more freely, more protected, and accelerate the work that they do. Does it make sense? Yeah, it makes perfect sense. And I just caught something what you said was about so you have these different tools to do different functions, but are you saying that you're actually building things into the tools to help the engineers? So it may be a photonics tool, but is also giving the engineers flags to think about the next step in the process. Is it? Yes. How are you? Okay. Yes, yes. So then I guess I mean by this, let me give an example. One of the big thing that people talk nowadays is AI, how you can bring AI into solutions, into products. I mean, not just helping engineers, but how to make AI available to anybody doing whatever tasks they're doing, right? Now, AI, it's something that it's based on data, right? AI needs data to be trained. And something we talked yesterday, you and I, which was, okay, so then, and I was quizzing, I was saying, what is the biggest challenge you think of incorporating AI into electronic design systems, you know, on EDA software, for example? And then people think, oh, it's because I need to protect my IP. I don't want to share my IP, but right, it is part of the problem. But if you think on the basic level, AI works, large language models, they work really well with text-based data. We live in a graphical world. All of the tools, Keysight and whatnot, they all use graphical environments for you to configure your designs, to test, simulate, and whatnot. So sharing the data with the AI in the backhand, it's somewhat of a similar challenge with sharing with the human downstreams with you. So then Keysight, for example, invested with CleoSoft, which is our SOS. ClearSoft's a company acquired a couple of years ago. And then we broaden our portfolio with a data engineering data management tool called SOS. SOS serves as this layer, this data layer where you can store your design data in it. You can actually version control the data that gets in there. And it serves as a foundation for you to pass this data back to AI. So AI can get trained, AI can respond better to your queries and your prompts and whatnot. And tying back to my point about graphical, the other thing that Keysight's doing there is that we have invested on Python capabilities on our platform. We started on the RF side. So then with the EDA for RF, you have the capability of, for example, exporting parts of your design as Python code. And Python code is ideal for AI to consume and get trained on it, right? If you recall back when we used Excel in the past, you could record macros and what you're doing with your mouse. So we're adding a similar capability to ADS where you can record certain things that you're doing in the ADS environment and then you can convert this to Python code as well. You can record certain things that you're doing in the ADS environment. And then you can convert this to Python code as well. Now, imagine you have all of this library with Python code for your design, for the things you do with the platform. And you can export this to a data management tool like SOS. And then you pass this to AI. AI can become very powerful accessing it. We're also starting to mingle the physical solvers from the different worlds. So then when we acquired, for example, ASI, they were really experts on the physical simulations, right? So then we are taking some of their solvers and some of the IP that we acquired and introducing it into the ADS tool, for example, for simulating thermal thermal simulation within our ADS tools. Does it make sense? Yeah, it makes sense. I like the word mingle. It appeals to my brain. And one thing we talked about yesterday, and we both landed on a word, which I think there's been so much fear-mongering. It's a messy field when it comes to AI. Some are really quick to embrace it, see the opportunity. Other people are scared. They're going to take their jobs. But you and I landed on this word sort of around the same timeframe, which was augmenting. And we think about when augmented reality came out or augmented reality goggles, it's like the engineer is still driving. They're just getting better tools. So as you said with Excel, with the macros, right? It's just putting another tool to save you time, right? It's not replacing you. It's just making your tool set better. And hopefully, you know, changing file formats, like turning something into a step file so you can upload it into your EDA tool for board design or whatever. Those are all opportunities for failure, right? When we're converting files and all that. So I like this idea that you're sort of creating this AI connected tissue between the tools that make it better. Exactly. And you mentioned something. I was browsing on one of those social platforms this past couple of days. And one of the things that I thought is interesting, I saw a post from somebody that I thought was so smart. It was saying that AI is not taking the job of humans. AI or AI will take the job of humans. No, AI is going to take the jobs of humans that don't know how to use AI. And then we talked about yesterday, Judy, right? Like calculators. If you have two, stupid example, you have, you need to make a calculation. How much is 1,334 times 510? Why would you calculate it by hand? Probably going to take a minute or two minutes for you to do it if you have a calculator there to help you do it faster. And if I have two professionals to hire, one that knows how to use a calculator and one that does not, I'll certainly probably pick the one that knows how to use the calculator better. It's the same thing. So then I think we shouldn't be afraid of AI. Now, one subtle point, I mentioned the word augmented. I see a bunch of really solid companies out there talking about AI-driven design, AI-driven this, AI-driven that. And it's putting the human in the loop. Honestly, as an engineer, I don't know if I like that. You know, I want to be in control. I think that it's the other way around. It's not putting the human in the loop. It's actually putting AI into the loop. So I think that exactly, exactly. So then I, and this is something that Keysight is, it's very careful about it when introduce AI into the systems, in our systems. You want to make sure that engineers, humans, are still in control. So you're never going to see anybody at Keysight say AI-driven this, AI-driven that. You're going to hear us say AI-augumed, AI-enhanced. It's a variation that you can see us talk about. So then it's high AI-enhanced design, AI-agmented design. It makes sense. Now, AI-driven design, a little bit, I don't know. I mean, sure, there's a place for it, but I like to have humans controlling it, you know. So this is what I think about AI. We shouldn't be worried about it. I think we should be worried about giving control to AI. So I'm a big fan of AI-augumed, AI-enhanced, responsible tools where there's a human driving it. Where does a human driving it? So it's not about bringing AI, I mean, a human to the AI loop. It's the other way around, is bringing AI to the human loop. Right. It's foggy times for everybody, but I'll tell you every time I just posted a podcast yesterday that was about AI, agentic AI, something that Siemens is doing with a company called Sellus. And it's really interesting, but as you said, AI is in the human loop. And so the way I expressed it there was that agentic AI is a co-pilot. You're not putting it on autopilot, which I think is another concern people, engineers have. I know they do because they talk to me about it. Exactly. They say, we're just going to raise a bunch of lazy engineers or incompetent engineers because they didn't learn. And I don't know. The more I hear about from people like you and Siemens and Salas, at least right now, I'm optimistic. And I think it's a great service of engineers. And I've spent most, a large part of my career, if not my entire career, helping engineers hopefully make their jobs a little bit easier or a little bit better. And why would I not give them AI, right? Exactly. And then, and this is my point. And I think that one other, one other aspect is how engineering needs to change, right? And this is really what we're talking here. What needs to change? So I think that we need to move from, and we kind of touched on this a little bit, moving from optimizing individual tools and not just tools, but tasks as well to improving decisions across the life cycle, right? Predictive simulation, right? This is a great use for AI in simulation. Concurrent engineering, how you can actually have an electrical engineer, mechanical, thermal, optical, all kind of concurrently collaborating with this AI fabric, AI-ready fabric, like the data management solution in the middle in there. Trusted engineering data, automation, you know, and we talked about augmented AI, AI-augmented design. And this is another way to say it is selective use of AI. You know, I think that we have to selectively use AI. AI is a great tool and has its place, but humans need to be behind it. The intent is human, you know. Yeah, the architecture and the intent is humans. And even with, we have a sponsor called Quilter.io, and they're making a PCB design tool that can help with PCB design. But the owner is very quick to say, and he's a very bright engineer. Gosh, he has like a triple major in physics, engineering, and chemistry. And he was at SpaceX. So he's very bright, thoughtful guy. And he's like, there are things like geometry that humans still do way better than AI for the foreseeable future. And some of that human creativity, those things that you talk about. Absolutely. Most of us will encounter that if we're just doing a query and chat GPT or whatever. It's like, no, I didn't mean that. So, anyways, I like the way you're thinking. And again, creating this cross-functional awareness from whatever stakeholder. So, you know, sort of where we started is it's a big problem, those late discoveries, and it happens all the time. Exactly. And it's more common than not people. Exactly. And I think that your listeners, I might think, hey, wait, but what does it mean for me as an engineer, right? And exactly what you said, which is finding problems earlier. You know, it's understanding this cross-domain consequences. Whatever I do with my thermal dissipation for the CPU, how am I as a thermal engineer? How is it going to impact the software in the end that's going to run this and throttle or not, my CPU, exploring more alternatives in parallel and with the augment AI piece, you know, to that? It's reducing physical iteration, is you're requiring less prototypes, you know, and being able to deliver with greater confidence. It sounds like it sounds like vague pieces, but no, it's finding problems earlier, understanding the scrolling concept, as I said, of exploring alternatives, reducing physical iteration. You know, this is what we're talking about and having this platform of solutions. And having this platform of solutions that kind of talk to each other, that kind of share a common AI fabric, data fabric for AI to consume, really empowers engineers to really achieve those things. And that's something just thought about that. This is a really good point in our conversation to bring to something that's happening soon, which is something called Keyside Design Forum. I don't know if you heard about this, Judy, but this is a series of events that Keysight do. And this is a really good, a really good opportunity for engineers to see. And we talk about connected workflows, but what is connected workflows? It's really this. What is it? Getting all this different, is all of this different electrical, mechanical, thermal, all of these different physical worlds, spaces kind of converging and connected to each other through this common data fabric in there. All right. So in this design forums, engineers have this capability of actually seeing these workflows in action. We're going to have people from all of these different domains. Is this online or in person, Roberto? This is in person. This is in person. And the other thing I'm going to say is that it's, and this is the fun part, you know, the engineers attending those events, they can challenge our experts at Keysight, you know, pressure test our solutions with their own problems. You know, so then they can actually talk to these people that design the product, that created the products for them, and then see if how their products break, given their actual problems. You know, they can explore. Well, it's going to have some great hands-on demonstrations. And I know that there's a lot of protected IPs in all these discussions. We will have some capabilities of one-on-one discussions at those events. So then if the engineers have specific problems, they want to pressure test against our tools. Yeah. At those events, they can actually have one-on-one discussions with the experts from Keysight that design the product. And one last thing that people don't talk a lot, which is you'll be able to compare approaches. You don't need to compare your IP, but you can compare approaches. Whenever you are designing thermal dissipation or your next interconnect interface to your chip, there are common things that happen there, regardless of the IP that you have behind the scenes. So then you have the ability to talk to peers. They're dealing with similar complexities, you know, and share experiences there. So I think it's a really good opportunity for engineers to really explore those areas in person with the keyside experts and colleagues from the industry. Well, and Roberto, you know, for a few years I ran the Altim Live events. And the thing I liked most about them, again, I'm a big fan of in-person because we just learn differently. And it seems to stick better when we're having a two-way conversation instead of a monologue. And also, like you said, pressure tested. So, you know, you can call BS on something or really press the people that are making the tools. And like at Oltiam, that was really helpful for the developers. And a lot of times they would take that input or they could go click, click, click, and the engineer would go, oh my gosh, I never knew you could do that. And so this in-person thing, there's an energy. And I remember we were in La Jolla, and it was just one of my favorite moments because engineers were just talking to other engineers as well as people, but they were just having a beer sitting outside by the fire pit. And all of a sudden, all the laptops start opening. And peers are talking to each other. And then if they wanted to, it was kind of a non-threatening way to just have those conversations. And like you said, instead of us, you know, as people who are technical people, but people who market, it's easier to get caught in the fluff or the philosophical discussions we have about connected workflow. And it's like, yeah, what about this thing I'm doing today? Exactly. And so I love the idea that you guys create live events where sort of that magic can happen, right? And it is, it's a conversation. It's not just. Yeah. So, anyway, so then I want to, I want to make sure that your, your, your listeners, your, your audience, they hear about this. They think it's a fun event. Okay. And it's for them to come and talk, learn, pressure test our products, talk to the experts from Keysight. You know, I think it's, it's a really fun. I, I, I heard once from one of our customers coming to those events that they feel kind of like they're in Disneyland for engineering, you know, because you can you can learn about the cool new things you're working on. You can, and this, this pressure test that I'm talking about is a two-way judy. You point out something very important, which is the ideas that these people bring to the discussion, the people, meaning the engineers visiting those events and their challenges help us improve our products, you know, and help us make the products better for them at the end of the day. So it's an opportunity to influence, to provide feedback, pressure test the products, see them working, you know, and we take this feedback to our heart. You know, we are over nine decades, Keysight through all of the different companies that we kind of acquired and HP, Agilent, and so forth. So then we are an engineering firm and we take engineering to our hearts. You know, it's a fun, it's a fun event. Well, thank you for sharing that with me. I will definitely circle back with you and your colleagues and get that information and put it in the show notes below. Regarding this topic of connected workflow, do you have, besides the live events, is there anywhere you suggest that people go to learn more at a practical level how these things, you know, this concept that we've talked about, which is earlier discovery, is there other places you'd recommend that engineers go to learn more and do a little bit more of self-discovery? This is a great question. You know, I would, I would suggest that engineers go to, we recently revamped our website at Keysight. So then we try to make it more modern, streamlined, easy to find things. And in us refreshing our website at Keysight, we actually created a new home for design engineering software. So then, and if they go to keysight.com and they visit our, and you can put potentially, possibly can put to your listeners a link here there too. So then design engineering software is, and then you can see the various business units that we, well, the various areas, not busy, the various areas that we play, which is electronic design automation, computer-aided engineering, software quality engineering, optical design engineering, and engineering data management. You know, those are those are the main areas that cover all these aspects they talked about: the electrical, mechanical, thermal, optical, the software piece. You know, so then this is what I recommend them to go and take a peek on the design engineering software and what it can give to them. Okay, thank you for that. I'll make sure I get all the links. And, you know, I want to close with just saying that one reason I wanted to work with Keysight is because real engineers and experts like Steve Sandler or Ben Dan and a whole host of engineers that I just know as friends are saying Keysight is the only end-to-end solution. And I think we're sort of in a new era of that for Keysight and trying to figure out how to really fill in that tool set, but then fill in the gaps between hopefully augmenting with the idea. Before we finish really quick, there's something you mentioned there that I think is important to say is that besides the left side of the V, Keysight's very uniquely positioned on the entire V part. We must remember our over nine decades of measurement science, our instrumentation, right? So then Keysight is uniquely positioned on the right side of the V, the left side of the V, and everything in between. So I think you're going to see more great stuff coming from Keysight in the coming months to show more of this. But this is where we're going on the entire umbrella for the entire V side of things. Right. So simulation and being able to correlate it and maybe cut down. So, all right, Roberto, well, thank you so much for your time today. It's always a delight to learn from you and hear what Keysight is up to. I'll be sure to get those links from you on Keysight Design Forum and also your new website. So thank you again, Roberto, and hopefully we'll talk again soon. All right. Thank you. For our listeners, thanks so much for eavesdropping on our conversation today about what's happening in the connected workflow and why early discovery is so critical to your success and how Keysight is working to enable that. I'm going to put all those links below, plus some others. So please go check out the show notes and we'll see you next week. Until then, remember to always stay connected to the ecosystem.