Harnessing AI in PCB design – a conversation with CELUS | Printed Circuit Podcast Episode 32
Hi everyone! Thanks for tuning in to the Printed Circuit podcast where we discuss trends, challenges, and opportunities across the printed circuit engineering industry. I'm your host, Steph Chavez. Everyone in the industry is wondering how AI will impact and potentially improve the PCB design process. We're excited to be collaborating with an AI innovator, CELUS. And in this episode, we'll talk with Antonio Becerra Esteban, Head of Customer Success at CELUS. Thanks for being here, Antonio.
Hi, Steph. Happy to be here. Thanks for having me. Antonio, can you give our audience a brief introduction of yourself and your role at CELUS? Yes.
So my name is Antonio and I am working at CELUS as a head of Customer success. I joined the team almost three years ago. As an FAE. So I joined on the pre-sales side, and then at the time, I had migrated to the Post-Sales side. In the end, what we are doing, what my team is doing, is ensuring that all the users and customers have the necessary content.
The necessary hints, helps, trainings to make sure that they can extract the maximum value from our solution. Let's start off by getting to know CELUS. So the first thing I want to ask is can you tell the audience about CELUS? Yes, gladly. That's what I get paid for. So CELUS is a startup from Munich in Germany. We have a wonderful solution a web platform.
That enables engineers to take requirements, capture them in the form of a block diagram, and then our AI searches for solutions, proposes solutions for all these blocks in the block diagram, taking into account the requirements. And then as a result of all the evaluation by the engineer, schematics are part of the output for different types of software footprint. So in the end it's a startup that helps expedite the process from technical requirements to your first prototype.
Interesting. What inspired the founding of CELUS, what problem in PCB design are you aiming to solve? That is a great question that I've heard one of our founders, Tobias Pohl, talk about quite some times. So all these originates when he's working on his masters. And at the end of the masters, you have to provide essentially like a report, right? Like, you have been doing some research and you provide a whole report out of that. And part of his project was developing some control unit in the automotive industry. So for cars Bavaria is very heavy on in the automotive industry.
And he found that when it comes to code, there are many solutions that allow you to generate code almost without knowing how to code. Right? Actually, I was there last week in a fair in Rome, and I saw the solutions of easy code, code without coding. You describe high level what you need to do, and this program will output the software that you need. This does not exist for hardware.
You have a set of requirements for your PCB. Good luck. Right? You now have to go to Digikey or to the elders of your company if you are lucky and have people with a lot of experience and then start researching, right. Start filtering specs so that you shortlist the hundreds of components you could get, then start reading data sheets to see if you can figure out is this the right component? Is this the component that will solve the problem that I have? And then how do I implement it? Which configuration do I have to take into account and how can I capture? Where do I get seamless footprints and what not.
And this is the problem that he faced essentially, dedicating hours and hours and hours of research to find the components, to find how to capture them on the schematic. And then if you are lucky, they are all available. Right? I raise my hand. I've been guilty of delegating dozens of hours of research to find the components, find the best component, and when I'm going to order it, obsolete, or not recommended for this design. So back at square one.
So that is the back story in the end how all this came to be. I want to talk about the challenges that engineers are facing, not just engineers, but the small businesses as well. So from your experience, what are some of the biggest challenges engineers as well as a small businesses face in PCB design today? There are different challenges.
One very big challenge that I think is affecting us all, is the knowledge gap. A lot of engineers are reaching the age of retirement and they're leaving the market. And younger engineers are entering the market, electronic engineers experience the frustration and the challenges of developing hardware. And many of them think, you know what, if I go more into the code side of this industry than to the hardware side of this industry, so the speed is changing. It's harder and harder to find electronic engineers that have experience in developing hardware, that are aware of all the challenges that appear, supply chain information, implementation problems with mechanical issues that can happen. Right.
All certifications, when you're developing hardware, choosing the wrong layer stack, and then having to go back after the first prototypes because, well, you have to redo everything. How cumbersome it is in the end, because it's very complex to produce a PCB. That adds to all that time consuming, right, that add to all these complexity that honestly, small companies, it's very hard. It's complicated for them to survive in this environment.
If you are a huge company with I want to tell you 10, 15, 20 developers, sure. That's the enterprise level. Exactly, your enterprise level. You have some wiggle room to adapt if things happen.
If you are a company that has 1 or 2 engineers, one leaves, 50% of your workforce has left right, and that means 50% of your capacity. If you have one person developing, you are very limited by the complexity of your project. You might be able to output 1 or 2 boards a year. So, mainly because you begin with the research, everything is super complex until you get to prototype, then you start finding you might find some issue.
And every spin is time consuming. This is a trend in the industry, right? We're going through to more automation and more assistance, more information available for engineers to make better and more educated decisions. So the more tools engineers can have at their hand to develop, the easier these challenges will turn. That's where you hear the phrase design resilience or supply chain resilience is intelligence at the point of design. And when you talk about challenges, how do these challenges affect the ability to innovate and stay competitive in the market? Imagine if you have to invest.
I have talked with so many engineers, you have to invest in finding solutions. You finding a microcontroller. I'm going to I'm going to make a very simple example finding a microcontroller that delivers what you want.
Then you are almost get branded with that microcontroller for your life, right? You're used to work with STMicro. That's what you do. If anything is to happen to the line of controllers you work with, you have to start learning again. You have to start adapting again. I was talking with a friend that worked for a company, they had a almost like a PLC, essentially.
Right? Receive analog and digital signals and providing all the outputs. And they were using a microcontroller that was almost decades old. To repair and to produce new units, instead of creating a new board and updating, they were fixed on: "we need to even go to the end of the world to find more of these units," which is a huge time expenditure. You need to invest a lot of time.
So this type of rigidity, hinders companies from adapting coming up with the next iteration. In the end, for every development it can be hardware, can be software. Time is money. If you're a small company, you might not have the time or the money to innovate. You do what you know. You work with what you know.
There might be a better solution out there, but in order to find that, you need to invest maybe three months, half a year, that solution is going to remain even for you. I'm a big proponent for, automation and machine learning AI. So we talk about automation and especially with the evolution of AI, let's move on to the role of AI when you think about PCB design. AI is a hot topic in many industries, so how is AI specifically transforming the field of PCB design? That's a great question. So you have to think that AI, many people perceive it almost as a as a threat.
Will AI take my job, right? I'm sure AI is going to be transformative in all aspects of industry and life. So when we talk PCB development, we can think of AI as a trusted companion, as someone that is going to suggest is going to help you solve the problems. This point that I'm making is very important. It's not going to solve the problems for you.
It's going to help you solve the problems, because in the end, it needs to be a human being behind evaluating the answers, evaluating the proposed solutions, and making the decisions. I would not want to be the engineer that trust an AI to do, I was going to say complex, but it could be an even a simple layout on a PCB, send it to the manufacturing and then when it comes back and the thing doesn't work, I have to tell my boss, well the AI decided that for me, right? Because it is not the AI that is going to have to pay the price for that bad decision. So AI, I think it's here clearly to stay and it's here to boost the efficiency of the industry, boost efficiency of those engineers that adopt it that can help themselves getting answers from a datasheet can help build the models, can help getting suggestions to solve very complex problems, things that you could solve yourself in half a year, or you could use the power of artificial intelligence to solve them in a day or a week. In the end, that's what it's all about. When you think about all the mundane tasks, especially, from the role of a electrical engineer.
There's a lot of research that especially when you talk about reviewing the requirements that come in and how do you translate those requirements into viable solutions. You know, when you think about putting the schematic together and then especially, you know, transforming that schematic into a layout, there's a lot of details. There's a lot of things that have to take place behind the scenes, especially when you think about the role of the EE and all that he has to entail.
There's a lot of mundane tasks, especially regarding scrubbing data sheets. It's a lot, some of these data sheets, you know, a couple several hundred pages. So especially if you think of a microcontroller or a processor, I could tell you a PCB West, AI was a hot topic. And it's a lot, especially when you think about engineers thinking that it's going to take their job.
And like you said, I really love what you said about it's not, you're just going to a human is going to have to control that power and utilize it. So when we think about the tangible effects, what are the tangible benefits engineers can expect when integrating AI into their design process? You talked about mundane tasks, and I think that that is one of the easiest to understand. Everyone has probably read here their LinkedIn or whatever and any post about, "I asked ChatGPT, I had this problem and I asked for a component of recommendation, and I started having a conversation and this is how I came to my solution." Think of precisely this capturing an AI that is oriented to electronics.
That helps you capture the technical requirements of your project and offers you fitting solutions. This means that you're going to work way more efficiently on all these tasks of assembling your requirements, trying to find components for them, trying to implement them. Try to distinguish which is actually one of the most time consuming tasks. If the solution that you have in front of your eyes is the right one or not, if it were not, adding information about being able to also base your information on supply chain information, cost optimization, the amount of things that happen actually early on.
I'm not touching layout. I'm not even talking layout, right? The amount of things that happen only on this schematic are huge. All these little tasks.
Finding the models, downloading the models, capturing your schematic, connecting everything. Now if you are able to capture your requirements and AI provides you with solutions, you can check if they match. How do they match? If there are other requirements that maybe you can codify in a different manner, like you might talk to an assistant so that you can validate the solutions. And in the end, when you have found it by the click of a button, you have a generated schematic where everything is connected according to the way you did it on the AI tool.
That is amazing. That saves so much time in checking manual steps, taking information, click here, click there. The picture that I always have on my head is the picture of myself five years ago, with dozens of tabs on my Chrome browser jumping in the end, I have lost which data sheet was I on, was I supplier A or supplier B? That's the key.
That is the biggest power of AI. The requirements, you know, getting them into the schematic is a challenge, especially when you talk about what solutions are available. How do you know, especially if you're a young engineer. You talked about the engineering gap with young engineers coming in. How do you get them to be utilized faster? So there become an instant ROI for return on investment as an employee. How do you get them up to speed faster? And I think with the implementation of AI, it allows that engineer, that young engineer, to be able to adapt quicker and faster and get up to speed and produce a product a lot faster than, than in years past.
So, you know, I want to move on to, CELUS's unique solution. What key features does CELUS design platform offer that addresses the challenges we've discussed? So the one that I would like to highlight, which I love, is CUBOt. CUBOt is our AI assistant.
The journey on the CELUS design platform begins, as I've said, we capture the requirements creating the block diagram. Then we click on Solve or Resolve and the program is going to check for pieces of circuitry, right little modules that are going to fit my requirements. If the requirements are functional requirements, specifications or interfaces that the solution should have, I find it beautiful because it is an iterative process, which means that you can add and update your requirements as you discover more information in a very, very efficient way. So when I have solve, I have to start validating my evaluating the results. I can go inside one of the blocks. I start seeing all the solutions, the solutions we call them CUBOs.
I'll say it's a Spanish word. I might be biased. It means bucket. So, just is in a bucket.
You can put all the necessary information, right, to distinguish, to classify and determine which type of solution that is. And the way I explain it is in two words. A COBO is a digital data sheet that helps our AI distinguish solutions to make it faster for you and easier to understand.
You can now interact with our AI assistant on our platform, CUBOt, to the point that you can ask, hey, I'm searching for a solution that has five volts in 3.3V out and I need 200 milliampere of current output. Or hey, I'm searching for a wireless communication module for my IoT board or for my industrial board. I need to have, you know, Zigbee or this particular solution.
I can be very specific or I can be broader in my intention. And our assistant is going to take your query, your question, it's going to check the solutions that are offered. And it's going to provide a suggestion. Then I as a human can open this suggestion.
I will open the page of the CUBO, I will open the page of this digital data sheet where I'm going to see the application to the full schematic. I'm going to be able to see description, specifications, interfaces. I can see the BoM of this application, even if I go further, if I click on the component, I'm going to be able to to get to the page of the component, download the data sheet as the latest step of validation, which means that through CUBOt, I am going to be able to navigate even faster through the proposed solutions. I'm going to be able to get super quick to the information CUBOt.
On top of that, gives you when it answers it provides it normally gives you also where it comes from on the data sheet, for example, which means that, so, I won't claim that data sheets are obsolete. I don't think that will be the case anytime soon, but you are going to end up reading data sheets only that you're going to read those that really require your attention and not 40 of them, so that you can filter out your results. And for this, CUBOt it's great. It's wonderful. I really enjoy talking to our assistant. I seen the demo that you gave on how this flows and I was so excited.
But at the same time, I was a bit disappointed because I would have loved to have had this 20 years ago in my career. You know, 30 years ago when I was starting and working with EEs, because a lot of times I was asked to go research data sheets because the EE had a stack of data sheets so, we were both trying to collaborate, and we spent a lot of our upfront engineering in doing that, and I could see where this is a huge time saver and to have it at your fingertips. We as humans, there's that factor that you have to factor in, which is the human error of missing something. Whereas AI, you know, it's a computer, it's it's computer operated and there's so much power there.
That is where experience plays a role. Right then human errors tend to you could you could use your experience to overcome the typical human errors with you learn over time. Now with the power of AI you can help.
I won't tell you fully substitute 30 years of PCB development, right? But as you mentioned before, you are going to be able to use the power of AI to come to speed sooner. So the ROI of a new, hire, it's going to be much higher. Exactly.
And I think it also will allow not only to be able to engineer faster, design faster, but it'll allow you to remove a lot of the mundane tasks so that you could focus on truly the pure essence of design architecture. When you think about designing an entire electronics system. I mentioned the demo, but I want to know, can you share any success stories or examples of how your solutions have positively impacted users? Yes, we have actually quite some users that have, have texted us, thanking us.
And it's not like for a specific solution, but for, for these and peace of mind that they have a way of finding circuits there where they don't really have much idea, because it could be that you are used to working with, say, microchip, you're used to working with microchip microcontrollers, right? You know, the all the microchip environment, that's what you work with. But now you have to find some special sensor or some power filter in some, some other type of solution that you're not so familiar with. You will be happy with a solution. You only need to solve a certain problem.
And we have had quite some users come to us actually, we were last weekend at the fair in Rome, and the amount of feedback that we got of people being amazed that with the platform, that's for them, I was very positively surprised. I was very pumped. It's always great, right, to hear from everyone.
All in all, there is not only that they you are speed up the process, but it's also. I mean, this is going to sound too fluffy, but it's almost like a peace of mind that you can have some help in a part of the PCB development that up until recently you were on your own. I agree, I think, you know, you mentioned earlier, when you think about the young engineers starting or small business unit, but I am looking at it, you know, for me, I've 35 years in the industry how I can utilize and harness this power with my experience and use it to my advantage.
But the key here is be willing to trust and use that automation or that AI solution that's available, and that's I think that's the challenge we think about, today's engineering teams and how they embrace automation, especially with the evolution of AI. It's going to be interesting to see who will embrace it and run with it and who will remain status quo. And then you've got to think about what is the cost of doing nothing.
Let's move on and talk about partnerships with Siemens EDA. How will integrating CELUS AI powered front end with Siemens PCB design tools benefit users, particularly those in the SMB market? I would say the benefit is huge because now you don't only have a very powerful solution for PCB development, as it is the Siemens EDA set of solutions, but it's also you have then what comes before, right, when do you enter the realm of layout, PCB development, I think it is easy to admit that the Siemens EDA solutions are top. But before, before you get to the schematic now what you can do with this integration is get to the schematic much faster. the boost in efficiency, you get, open Siemens connect, create a new project and then start capturing the requirements. You are offered solutions.
You are offered pieces of circuitry that are connected together according to what you're describing. And as the last step, once you have validated, you are now on your EDA tool from Siemens. You have your schematic, that has to be generated based on your input, based on your validation. Now you can take it from there. You go from your idea, from your prototype, in your Siemens environment, in hours, say, in a day. It's amazing.
It's huge. Like I said, I seen the demo and I'm crazy excited about this. But let me ask you, what excites you most about this partnership and its potential to influence the industry? Where do I start? Having a partnership with Siemens, It's it's huge. This means that such an actor in the industry as it is, Siemens, has seen and recognized the value of our tool, the value proposition of these things of this startup in Munich. Right.
This means that we are going to be able to learn and develop so much better, because in the end, what we want is to offer value, to offer good solutions to our users. And this partnership, it's going to help us gather very good input and develop our tool in the direction of what the industry needs. That is marvelous. I couldn't agree with you more.
Now let's focus on the long tail and SMB market. Why is important to make advanced PCB design tools accessible to small business and independent engineers? I think it all boils down to what you mentioned before is resilience. If you don't have advanced tools, you get stuck in the past. Nowadays, if you don't jump on this train and start adopting advanced tools that boost your efficiency, that bring more information to you, to make better decisions that help you create better quality PCBs, help you check for emissions, right? All these are going to cost all these small businesses time and money. And this is the thing that most businesses have the least of, right. So it is strategically a key for small businesses to embrace the power, the multiplying factor that advanced PCB development tools can provide for them, not only to be resilient, but to thrive in the end.
I agree with you, when I think about our evolution in this partnership and in how the engineers are evolving, I think those that embrace it will definitely excel and lead the pack. Those that don't, Frankly, they're going to get left behind. And when you think about the competitiveness of getting your product to market, if you're okay with coming second, third or fourth in the industry, then continue what you're doing. Those that want to be the first to market or want to be successful. You know, moving forward, AI is definitely embracing this solution is the key. Absolutely. It's evolution.
It's evolution right? It's almost a matter of evolution. If you get behind how behind is enough right. How behind can I get to survive? That's a luxury I don't think almost anyone can afford. Right. So getting your product to market as quick as possible is key in such a competitive environment as it is electronics development.
From my experience especially, you know, starting my career off at a small business and then in my evolution to the large enterprises, I can now only just see just a small business. But as well as the enterprise level, when you think about level load and resources and how to allocate that, and you get the most out of what limited resources you can with the shrinking budgets, the tight timelines, and I can see this solution and this collaboration between CELUS and Siemens significantly being a positive impact to the industry. And I want to look ahead to March to 2025. I know there are limits on what can you can share with us, but what can our listeners look forward to with the up and coming, fully integrated product? They can look forward at an explosion in efficiency. They can look forward to way less frustration when developing their PCBs.
They can look forward to having a confluence page with requirements for the PCB, and getting from there to the first routed PCB. The first route of the layout between the combination of the CELUS design platform, giving the symbols that the schematics and the footprints, and Siemens EDA providing the tools to route and define the PCB in record time. So, they can look forward to talking to their solutions through CUBOt, for example. Right. They can look forward to getting the products to market in an absolute Record-Breaking time.
I couldn't agree more. You know, how do you envision this product changing the daily workflows of engineers and designers? The way I envision it, it's actually very, very positive. What I envision, what I see is, is a lot of happiness and way less frustration on their lives. Because if you don't need to spend your days reading data sheets, testing boards, realizing that you read something wrong, that you find you know all the time that is invested in solving something with your analog brain, but using AI as a helper to make your daily life more comfortable today, make your daily life, you can focus more on what I believe, probably all of us, entered this world, Right? Nobody entered this world because you like reading so much. You've read every book in there in the world.
Now you have to go to the deal. The real deal. You have to read data sheets. Right. So you entered this world well, because you are, in one way or another, a creative person. You like solving problems. And that is what I see, what I envision in the future, a revolution that will allow engineers to be creative and focus on what they really like.
It's definitely a challenge, when you think about hashing through all the data sheets and then how do you interpret them when you read it? The could be a little cloudy or a little hazy, or technically it's maybe a little hard to understand, but I think with the AI, I think it will definitely help solve that problem. It an amazing solution in an amazing shorter amount of time, and time is money and that's the key. Before we wrap up, do you have any final thoughts or advice for engineers looking for a place to start when they're thinking of looking to do their first PCB design? I would say as a fresh engineer starting plenty of content out there for you to learn and grow on.
I would say explore. Don't be afraid of making mistakes. You have this life power of being young, having a lot of time ahead of you. Embrace the new solutions. Don't be afraid.
Learn to use the new technologies. Use the AI, master it. Use it to create your PCBs.
Use it to create PCBs where you're going to make mistakes and going to burn them. That's also learning, right? But don't be afraid. Don't be afraid of using the resources you have at hand, because that is what is going to make them stick out in this market.
I love that that phrase. Don't be afraid to use the resources at hand. And that is the key, is this is a huge resource. Use it to your advantage. And when I think about this opportunity, I think about establishing a better work life balance.
You can design faster, less time. And like I said, time is money. And time is important. And in how you can gain back that. That time is key. And I think this is a great opportunity. Hundred percent.
If I have had a better work life balance, I wouldn't have these. Great. You and me both. Yeah. You I got the silver on the side.
You know had we had these before, maybe I wouldn't even have such gray hair. Yeah, I know we're doing demos today for integrated product, but what are you most excited about when we think about that, that production release in 2025? No, I'm super excited about seeing this. I mean, I won't see it on people's screens, right? But seeing these live people clicking people, people realizing the value of the solution, people really enjoy in all the benefits of our collaboration, producing real PCBs. With all the boost that AI can, can provide them that I'm really looking forward to. I'm really looking forward to people using Siemens EDA saying I created this board thanks to the collaboration with the between CELUS and Siemens.
Awesome, awesome. Well, I think that's all we have time for today. So thanks for sharing your thoughts and insights on CELUS. To our audience,
Continue to tune in for more trends, challenges and opportunities across the printed circuit engineering industry. Thank you. Bye bye.
2025-01-26 18:02
