Judy Warner (00:01.498) Matt, Toby, good morning. Thanks so much for joining me today. I'm delighted to have you and to talk a bit more about the partnership between Cellus and Siemens. Toby, why don't we start out by having you introduce yourself and then we'll do the same for Matt. Tobias Pohl (00:18.019) Perfect, thank you. thank you for having me. I'm Toby. I'm one of the co founders of Cellos and CEO. I started the company about eight years ago with my two co founders, and and happy to be on the show today. Judy Warner (00:32.956) Matt? Matt (00:34.144) Hey Judy, thanks for inviting me. I'm Matt Bromley. I'm the Vice President of Product Strategy for the electronic business systems part of Siemens ED8. we're responsible for the design tools for creating, simulating, testing, PCB designs and getting them manufactured. happy to be here. I've been working with Toby for a couple of years now, so really happy to talk about this relationship. Judy Warner (01:00.316) So can you both dive into what the strategic reason was for your partnership and what the problem was that that you're trying to solve by this partnership. Matt, you wanna start? Matt (01:17.24) Sure. I think if you look at sort of electronic design, a lot of it is in this very early stage of concept design. I've got a I've got an idea. I don't quite yet know exactly what components I want to use, but I've got these ideas and I want to be able to do some very quick let me build up a a concept design to see if this is going to work with this set of components or this set of building blocks. to solve this kind of problem, to a high-level problem solution. A little bit before you get into the area of okay, now I've got a schematic and I need to get into traditional engineering. And so, what CELUS really brings for us is this ability to have a really robust solution to that early concept design phase. And what Siemens brings to the relationship is once you've got that concept a little bit more solid. You can transfer into the more typical design process where we care about electrical connectivity, design rules going into manufacturing, and of course producing the board at the end. So, from that perspective, I think it's a great relationship that we've got to help engineers be as productive from that early part of the design cycle where they're ideating, right into the design and then the manufacturing parts of the design cycle. Judy Warner (02:42.012) Toby what it mm-hmm. Co. Tobias Pohl (02:42.114) Yeah that Yeah, that's that's exactly right. I would describe it the the same way. When we started out, we saw like a void in the market of that fuzzy space in the beginning where you don't really have hard requirements yet, you don't know yet which architecture route to take that project, where at the end the closer you get to the end. the more robust tool chains exist. Like right? If as soon as you're in a real schematic, you go down layout, you go closer towards manufacturing, there are robust tools in the market, such as the the solutions from Siemens. And we saw that empty space in the beginning of how do I get there? Like how do I get started quicker? where the software world has a lot of solutions to get started quickly, to generate the first le level of code without getting too much into the details. And this is exactly where we came in and what we want to fill that gap, carrying engineers from requirements to the first component selection, to finding the right chips, adopting them in their design and getting over into schematics. But what we also obviously in that journey then fairly quickly realized is the engineer still needs to do the entire thing. So connecting those two sides together is then the most powerful way way of approaching that. And that is where the partnership together with with Matt and and the this folks at Siemens really developed around together we can build a complete journey end to end. and that I think is what makes it so powerful. Judy Warner (04:20.658) So as I talk to engineers and thought leaders across the industry, the velocity and the complexity of designs these days, it just is a little mind-blowing for me personally, and I'm sure for you too. So given that complexity, you know, what are some kind of tangible bottlenecks, Toby, you identified and then you, Matt, and Matt (04:25.525) Yeah. Judy Warner (04:49.676) And and how you really close that loop between ideation and schematic, say. Tobias Pohl (04:56.214) Yeah, good great question. So to my from my perspective, there is there's definitely no shortage in problems. There's definitely no shortage in amount of solutions that need to be built. but engineers don't have unlimited time to work on them and don't have unlimited time to to to build different projects. And I would say in the the tool chains of actually building solutions can't support far more. We can already with the tools that are available today build significantly more complicated projects, build significantly more solutions. But an engineer needs to be able to handle all the options they have in front of them, handle all the paths that they can take, and handle an insane amount of data. I the just looking at comparing different main MPUs for your design, for example, can spend weeks on that comparing data sheets. And then after you made your decision, you two weeks later figure out that that is not compatible anymore to the next part you chose. So it is a messy process in a way that is everything but linear. You can't just make one decision, go to the next, go to the next, and everything will just fall in place. And this is where a just a lot of time is is lost for engineers in cases going in circles and having to make the same decisions over and over again with new design considerations. And that is also where I think there is a lot of opportunity space for now what what we're also doing together and working on together of making that big amount of information data more digestible, allowing engineers to get short take shortcuts to the initial design to the initial bomb, the initial chip selection, etcetera. Judy Warner (06:55.718) Matt, you and I talked at Microchip, I don't know, within the last year. And that was the place that I really started to get a handle on what the work between microchip and Cellus and you what it could look like. and I remember either you or someone on your team was talking about the efficiency and and robustness of that you could go through. 200 page data sheets. But you'd also made a little system that was a board where engineers were able to go from idea to finished product extremely rapidly. So what is the part that AI plays here? I feel like the industry we're starting to get a little beyond the hype, and we can talk a little more thoughtfully about what it can and cannot do. And so tell us about what it does at that discovery process, and then how that flows into Siemens tools, say, and what that does for the designer. Matt (07:59.694) Right. That that's you know a great question. I remember when we spoke at at at Microchip and this is actually the board that we were talking about at the time. Just just happened to have it on my desk. so so yeah, I mean I think as Toby said, I mean, at one level now, what we see is that there's always been a tremendous amount of information in E in EDA, right? I mean data sheets are a great example. Tremendous amount of information goes into a data sheet. Toby is leveraging that. Judy Warner (08:09.157) There it is. Ha ha ha. Matt (08:30.051) his side, we're leveraging that as well. and so you mentioned Judy specifically around AI, and so technology to extract models and data out of data sheets is something that we already have available. So you can compress that cycle of creating new library elements, getting them into the design cycle as quickly as possible. That's only one element, and I think as you pointed out, we're starting to see, you know. the the hype cycle around AI pan out into areas where there's actually some real advantages. Toby at the front end was sort of mentioning a little bit about being able to choose the right MCU for a design. You know, we now have the ability to very quickly look across a bomb, an early bomb that may be sort of more of an ideation bomb, and then start saying, well, is this going to be right for me based on who my manufacturer is, what my lead time is, whom I'm going to have produce it, and have AI look at the set of data around that and very quickly give me answers about whether or not I want to make a change early in the design cycle versus find that later on in the design cycle. And I think, as we all know, the earlier that you can make those changes, the less the cost is to the overall, the overall production costs. And it's not just there. I think we're now seeing as we get into The actual schematic design and the layout as well. We're starting to see AI really have some applicability. One area that we're pretty excited about is in simulation. You mentioned earlier that designs are getting more complex. You know, that's definitely true. They continue to get more complex. Simulation is a fairly complex role within our domain. It tends to need Judy Warner (10:13.167) Yes. Matt (10:21.759) specialized engineers to do signal integrity or power integrity. But what we're finding is AI is really good at interpreting the results. And so now as I go through that design and maybe I do run a simulation on it, I can use AI as almost that that engineering design assistant and have it look at the results and say, well, you know, this channel is not performing as well as the other channels. You may want to look at changing this trace width to fix the impedance. And so I think you're spot on. We're starting to see some really good examples of where AI is applicable in compressing the design flow, right? Making the design flow more efficient, as well as the new areas that we're starting to hear a lot about with agentic AI, as being ways of tying the design flow together in a more a more efficient way where we can just have the human interact and say, I just need to run this simulation on this design and it will go off. and orchestrate that for you. So yeah, you know, still excit exciting times in EDA, right? I think there's a lot more we can do. and it's great to see these technologies start to add real benefit to to our users. Judy Warner (11:35.099) Toby, can you talk a little bit more about the agentic AI piece and maybe making sure that the designer has control of those constraints and what kind of moving past the LLM early hype to real robust things that engineers can reliably depend on? Tobias Pohl (11:56.829) Yeah, g great question because the robustness of the solution is really the biggest change that I think that developed now in the last year or so where with the switch from the more traditional traditional is even maybe the wrong word since this is a thing since only two years from now. But the more classic approach of having an LLM assistant alongside your design that allows you to describe Judy Warner (12:16.513) Yeah. Tobias Pohl (12:25.77) what you want in natural language and turn that into actions that we now see across many industries and many tools. But now with the switch to full agentic architectures, you can get so much more powerful and so much more robust. Where you and that, for example, on on our platform is is life like this today, you're actually not talking anymore to the AI assistant that makes all the decisions and makes all the changes. You as an engineer are interacting with an orchestrator that can connect to many AI agents that then do different functions, help you structure your requirements. Because we are we're in that fuzzy space in the beginning, right? Where a lot of what we do is also helping an engineer to actually really formulate precise requirements. And then have, to the point that you made around robustness, not just have an AI assistant creating a first result, but have for example another agent that reviews the result that the first agent created and runs on a different model. Or you have not just guardrails on a AI assistant that limits the inj the design to go completely into an unus like a unreliable, not useful space to this design, by having also a guardrail agent. that understands exactly where are are we still on scope with the design? Are we deviating completely away from what we're trying to build? And frankly, the most exciting in this agentic and multi agent world is it doesn't even have to live in one tool anymore. You can now have within one tool different agents. So there are many different agents that live within our environment, for example. Judy Warner (14:20.913) Mm. Tobias Pohl (14:23.276) But there are areas of the design where a joint relationship, such as with Siemens, is actually the then the real powerful element where we have parts of the design process that are not ours. We don't want to go into what implications does now this MCU selection to our example earlier, have later down the road on my layout? Or how should I run a simulation so that it can validate my component selection? And that agentic flow allows that I can have that question and one part of the design and have an agent send that off to a different tool that then takes that and runs a simulation, for example. Or I can send a command from my agent over to the Siemens side and check for implications on schematic and layout front. And that allows that you basically have. almost no limitation anymore of what you can connect with as long as it is in that within that agentic world, you can build kind of super tools that are the combination of many that then jointly can build the best possible solution and iterate over that. Matt (15:37.58) I think I think Judy there are two real good points to add to that. One is you know, historically in EDA we spend a lot of time on integration work, right? And whether that's through APIs or file formats, you know, e diff and those kind of things. one of the exciting things with agent flows is the ability to do that very, very quickly rather than have to worry about that level of you know, sort of historic detailed APIs, understanding what everybody's APIs are, how they work. Judy Warner (15:39.281) Mm-hmm. Matt (16:07.139) The other is is is sort of a corollary to what Toby mentioned right at the start, which is some of this is a little bit more well developed in the software world. And if you look at the software world, it's incredibly common that you have a large number of test cases, so that whenever you make a change to software, you can validate that against the test cases. and as Toby was saying, as you you start seeing these agents proliferate and maybe we start seeing agents that are starting to actually touch design data. It's it's going to be absolutely 100% critical that we have those test cases. And those so so that we can validate the you know what the agents have done are are still within it with within the requirements or don't break the design rules, etc. And that's going to require a very, very robust set of schematic analysis tools, DFM analysis tools, Other simulation tools that the agents can either run themselves and look at the results to make sure that design constraints are still being observed, or human in the loop is running them to make sure that the design is still meeting those requirements. And so I think we're going to see a need for a greater emphasis on that type of continuous verification of design. To make sure that as the agents are working together with the humans, we're not breaking anything through that design cycle. And that's going to be absolutely critical moving forward. As designs get more complex, that simulation gets more complex. And as as Toby sort of talked about working it at the front end, we're working that throughout the design flow with a complete scalable solution of both design and simulation tools. Judy Warner (17:54.994) So this is all fascinating because of course I think about my career, about all the problems we had and gaps long before the complexity we have today between different stakeholders up and down the stakeholder ecosystem. And so this idea of having validated data, so you're teaching, but still having that engineer in a loop. As you know, there's been a lot of concerns about AI taking over the architecture part, which has sort of been guarded up to this point, that most AI companies have been saying, no, AI can help you, but but there's certain things that only engineers can do. So your job's not at risk. So and I hear what you're both saying is that's still not at risk. It's still the human has to be in the loop. because it had it doesn't have this great body of validated data and the data keeps changing because complexity keeps changing. So is there a use case that you can talk about with one or both of you that really helps show how this works in real life? And Toby, for you, a follow-on question, or you can integrate it into your answer, is what is this like? when it comes to a user interface and how easy is it to sort of onboard it and know how to drive the tool? So if we can do use case and user experience or user interface, that'd be great. And I'm gonna let you guys choose who wants to answer the those questions. Matt (19:39.897) So so Toby, why don't you start off and we'll work it sorry, why don't you start off and we can work it through an entire flow? Tobias Pohl (19:40.266) I would say Matt, let's do it by journey. Tobias Pohl (19:46.143) Exactly, I would also do that. very quickly before I dive into that, Trudy, to your point of the human engineer as part of that flow and as the kind of critical element of making decision how much AI will change that. I think like this is a very common fear that you have now actually throughout all professions, right? Everyone is is asking themselves that question. I think what we will see in Judy Warner (20:11.089) Mm-hmm. Tobias Pohl (20:15.028) most industries, but for sure in in our space of of engineers making such creative critical decisions is will have a similar effect as the let's take an an an accountant that used to run his calculations on paper and then is switching to Excel. that doesn't mean that you don't need the accountant anymore, but that does mean that the accountant with Excel will always outperform the accountant that doesn't want to use a modern tool. And I think a similar development will now happen in our space, where you still absolutely need the engineer, but the engineer with AI can massively outperform the engineer that doesn't use AI. And now if we apply that to a flow where in the very beginning, in that what I now call many times the fuzzy stage of the design, where you're trying to capture requirements, Getting started, and I think ever every one of us knows that, getting started is often the re like a really hard part. It is very difficult to find your place to get going. And this is where AI allows us to have completely different ways of starting and also completely different ways of structuring interfaces and structuring UI of tool chains, where on our platform, You start in by literally saying what you want to build. So you can kick it off saying, let's take the the port that Matt showed earlier today. I want to build a memory game. it should have four buttons, four LEDs, behave a certain way, build that game. I might have additional customer requirements that I'm building it for that are in a PowerPoint presentation. So I'll upload that. So you can embrace being messy in a way. of having giving it all the information, all the context, and you don't need to first get to kind of that rigid, precise definition before you can even start. And then we take that and translate that into an initial block diagram, an initial architecture sketch of what you're trying to build. And then you can iterate over that. And this is where, again, from a UI perspective, we can leverage AI quite a lot. Tobias Pohl (22:36.172) By having it as the companion alongside the engineer. So you, as an engineer, give more input, give more of your design intent, of what your what your requirements are. And the design assistant asks follow-up questions of what you should be also thinking of, what options exist. And you can play that ping pong in a way between the design assistant and the engineer until you really have described what you want. Judy Warner (22:57.713) Mm-hmm. Tobias Pohl (23:05.024) And then you can start diving into solutions. And we find the MCU for that memory game. We find the fitting battery connector and power supply and battery management. Whatever you have on your design, find fitting solutions. And then I think a a great opportunity of that new flow is you're not just forced to now limit yourself to one or two architecture decisions. And then go down the tool chains and build complete designs for. But you can explore many because you're just so much faster in building processing comparing solutions that you can spin off multiple different designs and benchmark how they compare to each other. And obviously, then at that point, when you reached a point of okay, I roughly got where I needed to be, I have solutions that all work together. I found my entire bill of material, I could populate it down to the last resistor. Now I'm at a point where I have a good feeling of where I'm going. I have a first level of my choices logged in, but I don't have a design yet, right? I don't have a product yet. And this is where especially to leverage that iterative design, a smooth handover to Matt, and that's where I would give it to you. Smooth handover to the next tools matters. and that's that's what we we've built together, where you can then spin it off into the next detailed phase of the design. Matt (24:36.92) And and and there's a lot that goes into that next detailed phase of the design. So one of the things that Siemens and Cellas has worked very closely on is making sure that handoff is as electrically correct as possible, right? Because as soon as you move in, once you've made that commitment into a schematic and then into layout, everything is is based on physics. It has to be electrically correct. The layout has to be geometrically correct. we've put a lot of work in making sure that that transition takes into account libraries, a lot of IP in in the libraries that you've got and already created. And so now you've made that commitment, and that's where you get into a lot of the traditional engineering choices and a lot of the needs for simulation. And so, yes, you've got that concept of what you want and you've got that initial schematic. You may want to go off and simulate part of those designs to make sure. That you're getting optimal use out of the maybe the battery that's driving the handheld device. and that may change, that may force you to consider some changes at that site, so a different buck converter or something like that. So so that I've got optimized power there. you know, we look at in in in the journey we're on with AI in in certain different levels. One is around how can I just make the engineer within their design process more productive? and so that's at the schematic side, for example, techniques like command prediction so that it's easier to drop the tools and know what they want to do next. it's capabilities like automatic component selection and connectivity, sort of almost like dynamic reuse. We know very often that. many designs are derived from other designs and if you can help accelerate that process, that's fantastic. I mentioned earlier about the design rule checks and having those available so that you know you can move through that process and sign it off very quickly. We move into layout, we have AI capabilities to assist in component grouping. we've very powerful auto routers so that we can do the geometric auto routing of the design very, very quickly. Matt (26:53.213) and a continued set of AI capabilities to help compress what's the engineering schedule today. Then you get the layer on top that we've been talking about is how can the more agentic AI or the large language model AI really become that design assistant? And so I mentioned earlier about that in the simulation space. We've got that in a lot of the reporting and the analytics that we do as we build out that agentic layer. So we're really focused on the expedition side and the hyperlink side in making sure that the design is correct. I mean, that's critical now you've moved into the design phase, that we can shift left as much of the analysis as possible so that you can find out any errors as early in the design cycle as you can, whether that's shifting left functional simulation into the schematic or pulling. DFM as DFT as early in the design process as possible. So I think when you get into design, we really still have to focus on correctness of design, compressing that schedule as much as we can, and then moving into this agentic layer over time with the agents that we both talked about that can then communicate to each other and the technology to to support that, such as MCPs. Judy Warner (28:19.098) That's amazing. And again, since we've talked, had conversations in the past with both companies, it really feels like it's the the ground's getting more and more solid and more and more clear. as the agentic tools does give you more information. And at least for me, I call myself a civilian because I'm not an engineer. is do you think it's still true that the designer is holds the judgment and architecture closely, but that these tools give robust, reliable productivity and probably some robustness because it is ru being run through multiple tools across this. Would you say that's accurate? Am I missing anything here? Tobias Pohl (29:10.848) I would definitely say so. Matt, go ahead. Matt (29:11.301) Yeah, sorry, I was I was gonna say, I think as Toby said, you know, the tools are there, are there to help the engineer, right? and the engineer is still going to be driving that intent, but he's gonna have more tools to help him do that more efficiently. We are gonna have to focus on verification to make sure the you know that level of correctness, Judy, that you said is there. That's absolutely critical. A lot of that we do today, but it's a lot of a manual process in running DFM. or running simulation and we're going to get a lot of assistance in doing that. I I think outside of maybe the most trivial of designs, you know, yes, of course, if you're just doing a flip flop, you can automate that. That's not difficult. But really, if you look at a lot of electronics and and you keep pointing out the increase in complexity, that continues, right? I mean we see that continue. That that is unique IP Judy Warner (29:40.378) Mm-hmm. Matt (30:08.165) That the engineer is bringing to the table. And that's going to stay there. He's just going to have tools that help him do that more efficiently. Tobias Pohl (30:18.194) Would totally agree with that. I mean ultimately w we have architected the entire platform around the idea of maximising the productivity of the engineer instead of taking the decisions away from from from the engineer. And I've very convinced that that's also the only the only right way of doing it. that does not mean that there are no micro decisions that you can take away from the engineer and that can be automated in the for the sake of being more productive. But What a big decision and what a micro decision is depends on the context, depends on what the engineer wants, depends on the project. So an engineer going through that journey can make the decisions that matter, make the decisions that matter to that project, or where they can add their secret source that they know best, but take other parts of the design where they either don't have much experience in or don't care at as much and hand that over to AI and automation to ful fully cover that. And ultimately the question of where that line is is very much a judgment of the engineer building that project. And I see that very much comparable to the development we've seen in in the software world. Software engineers can nowadays hand over a lot of decisions to code generators, AI development environments, but that doesn't mean we need less software developers now. We have just software developers pumping out a lot more output by being productive and maximising the value that they have. And I I'm very convinced that the exact same will happen in in our world or is happening, of us and Matt's team and and our team really pushing for the tool chains to enabling an engineer to be as productive as possible. Matt (31:52.505) Should- Tobias Pohl (32:17.706) and making the decisions they need to make and not making the ones they don't care about. Matt (32:22.575) And I I I think Judy, what's what's really interesting as we look a little bit forward, we've talked a lot about the the the sort of PCB design process, the electronics design process, but really when you talk about developing a product, and particularly when you're in into that physical design stage, you have a lot of more domains that need to come together. You've got how does the electrical and the mechanical interact with each other, what's the thermal characteristic so that you can you can plan your cooling or your power. and we've really got a great set of capabilities now where where you do have agents that can interact, you can have those decisions and those conversations much more easily than a single person who's focused on a single domain. and so I think that's going to start getting really exciting when you start looking at the capabilities across multiple domains. And you've heard a lot about digital threads and comprehensive digital twins. And that idea that you can build up this model. And I think this is going to be where you really see that realized. and and more than realized, but actionable as well, because now you can go off and talk to these agents and say, well, what's the impact if I move this from a a thermal perspective, right? I mean now I move the heat sink, what does that do to my my thermal cooling challenges, etc.? So I think a lot of lot of opportunity in front of us here. Judy Warner (33:40.729) Well Judy Warner (33:44.064) I really thank you. Go ahead, Toby. Sorry. Tobias Pohl (33:44.086) That that's actually Tobias Pohl (33:47.889) that's actually a great point because the with is that I think is one of the best benefits of that agentic flow. Because now you can connect those domains and those different tools together without an engine an engineer having to orchestrate that themselves. Because I think one thing that an engineer does not want to lose time on is now being the babysitter of many different agents that are not connected with each other. If you describe what you want in one place, you don't want to turn around and tell it to the next tool all over again, because then we're killing productivity. so that having that full connection where those different tool chains, different domains can really link together and an engineer has that orchestration taken care of, that I think is one of the places where they can copy completely out of the loop and have the agents share information with each other where it's needed. Judy Warner (34:39.32) Right, which does like you said, if we add more to the the workflow, we're slowing down, we're going in the wrong direction. And Matt, to your point, it's exciting thing to think to me, to think about agentic agentic AI being able to look across the system and help at that level. again It it's AI. We're all in it, whether we like it or not. but we're going here. But again, I'm always because our audience is comprised of engineers, I want to try to tease out what's real and hopefully lay aside some of those fears and concerns and paint a more hopeful future in which the designer brings all the judgment, all the critical thinking. And and like Toby said, be the orchestrator of that. I I have to do one push back, Toby, because our listeners are not gonna forgive me if I don't. You mentioned software engineers, but didn't a lot of software engineers lose their jobs when AI hit? Tobias Pohl (35:52.759) I would but I I think that this is now a very would be a very short term view, right? Because I mean you had an existing workload being covered more efficiently, so less people could do that. And that's that that that's what what that led to. But I think it is a matter of time until with different i i economic pressure, with new new growth, all all that does is it turns us to be able to build more. and it's not the first time, right? AI is now a very quick change, a very powerful tools, but software developers before modern high level languages and compilers were significantly less efficient and you needed much more people to create the same quality of output and better tools also only led to them doing more. So I would be I like I I'm I'm with you Judy, I can be wrong, right? but I would be massively surprised if that doesn't fully swing back to us being able to solve even more challenges than we we could pre to pre before. And if you in a couple of years from now zoom out, there are bumps in the road, like with any the development of new tool chains, but I I think it will be a rather short bump. Judy Warner (37:09.954) Okay, that makes sense to me because you know, for people like Matt and I that have been around a few decades, we've seen, you know, the sky is falling so many times across different technologies. And it's like, music is going digital, the world's gonna end, or d whatever. TV networks are dying, it's gonna be the end. And it ends up creating more opportunity and a bigger pull over and over again. Matt (37:31.153) You mean you're absolutely right, Jimmy. Judy Warner (37:39.226) What were you say, Matt? Matt (37:39.408) Yeah, the you're you're completely right, Judy. I mean, it y y you mentioned the AI hype cycle earlier on, right? And I mean ev every new technology tends to go through that cycle. those cycles are compressing, and I think with AI, you might almost argue, you know, within the space of five years we've gone through a couple of hype cycles from, you know, standard LLMs to agentic AI, et cetera, right? And they do they do filter out and then you find out where the real value is. From those technologies. And if you take Toby's analogy of software, you know, years ago everybody was writing machine code and then it went to assembler and then it went to, you know, kind of compiled languages, et cetera, right? And each one really created a new level of efficiency, an opportunity, right? And so it gave us the opportunity to have more complex software. And I think that's what you'll see in electronics. We'll have an opportunity to innovate more, to spend more time on innovation rather than. than than some of the more mundane tasks that everybody has to do today. and that's going to enable the companies that embrace that to to to be ahead of their competitors. So so as you pointed out, we've been through these cycles before. Those who embraced it are going to be able to innovate a lot better Judy Warner (38:57.646) Well, it ends up opening up things we haven't even thought about yet. And we can't because we're not there. And so that's why I'm optimistic because I've lived through so many of them. And we see what's shrinking, but we don't see what we're growing into. And anyways, that's been my life experience and I trust it's gonna be that way for design engineers too. And to be incredibly more confident and productive. and really focus on the things they love to do rather than all the grind and the boring tasks and the spreadsheets and all the inefficient. And there's so much risk there. Like you said, with you know, engineers changing file types and design reuse you mentioned and all of that, and doing that s sort of manually introduces so much risk. and this allows us to have hopefully less Matt (39:26.482) That's right. Judy Warner (39:55.31) Less spins, less design cycles. the shift left thing, you mentioned Matt. We've talked about a digital twin and shifting left for a long time, but it feels like AI is going to help us do that better. which I think is exciting when we think about the existing roadblocks and bottlenecks we have. Well, gentlemen, you've given me so much of your time and I've learned so much. Thank you so much for this thoughtful conversation. Before I let you go, you want to tease like what's next, maybe what you're working on now and what we might see next, and then also tell us how our listeners can learn more about the integration between both companies, and then I'll let you go. Matt (40:46.665) sure. So thanks Judy for the opportunity, right? It's always great to talk to you. So I really enjoy that. you know, I I think as we've mentioned, the area to look at now for for sort of new innovation is going to be around the agentic side. And and as we said, we're all working on that. Toby's working on agents, we're working on agents. The ability that when we connect those agents, you can start to see some really interesting things happen because now you start to get these iterative cycles that can work together, right? That's going to be really, really exciting. you to to learn more, you can certainly go to our website or Toby's website, Arcelis' website. we just did a press announcement at DAC about what Siemens is doing around AI. That covers the entire EDA platform. So I'm here talking specifically about board, but in that same way that we talked about multi-domain talking together. Siemens has an agentic platform that covers our IC, Questor tools, and our PCB and packaging tools. And so I think we're right on the cusp of some of these really interesting agents that are going to really Really do what you said, take out some of that work that people don't want to do and allow them to focus on the things that we do want to do. we continue to focus on the design process, right? And just making that more efficient and more pleasing and easier to use. And whether that's using AI or not using AI, we just want to make engineers be able to have a a a greater experience in the tool. And for us, it's all about scalability within that. If you're doing a simple design. It's the same tool as if you're doing one of the the very complex designs that the leading edge customers are doing. and so you you can certainly learn more about our entire scalable solution as well. be happy to answer any questions any of your listeners have. but like I said, and and and and really like you said, Judy, we've been in this a long time and it continues to be really exciting. And so I'm really enthused about what this new level of technology, how we can leverage AI. Judy Warner (42:58.466) It does. Matt (43:04.24) and what that's going to bring to the customers and engineers who are in this every day. Judy Warner (43:09.933) Well said. Toby, final thoughts for you? Tobias Pohl (43:10.926) Yeah. I can definitely echo what Matt said and and for us we've we we push more updates to our platform in a short amount of time than we've ever done before. So there is l lots of new things coming. big one really around like Matt said as well around the agentic flow of connecting to many places of the ecosystem. I think that's one of the most exciting updates that we're bringing, but also we together with the the broader ecosystem bring. And how to to learn more and really to get started with it. In our case, extremely straightforward. Anyone can just start building designs on our platform, finding the the link on our website or on app.zellos.io and can just get going and would always see the the latest version of our platform live. and thank you Judy for for giving me this the opportunity to present that today. wa was a lot of fun. Thanks for that. Judy Warner (44:08.239) Well, it's been my privilege to host you both. And well I well I like platforming folks like you, I always learned a lot and I find that very exciting to speak to executives like yourself and sort of s get a sneak peek into what's coming. So thank you for all you've shared so openly and the expertise you bring and I wish you well on continued progress on on your collaboration and and work together. So for our listeners, we're I'll make sure and I will put websites for you below and Matt offered questions. I'll put some LinkedIn links below so you can go explore further. thank you once again to Matt and Toby. We'll see you next time. Until then, remember to always stay connected to the ecosystem. Matt (44:58.556) Thanks, Judy. Tobias Pohl (44:59.971) Thank you. Judy Warner (45:00.27) We are done.