AI-Powered PCB Supply Chain Management with Luminovo's Timon Ruban

Discover how artificial intelligence is revolutionizing PCB supply chain management in this fascinating conversation with Timon Ruban, founder and managing director of Luminovo in this episode of the OnTrack Podcast. From his electrical engineering background at Stanford to building Europe's leading electronic supply chain platform, Timon shares insights on solving complex procurement challenges, automating PCB quoting, and the future of AI in electronics manufacturing.

Learn how Luminovo helps contract manufacturers and OEMs streamline their sourcing processes, manage supply chain risks, and get instant PCB quotes through advanced Gerber file analysis. Timon discusses the evolution from manual Excel-based workflows to AI-powered automation, strategic supplier management, and the exciting roadmap ahead including LLM integration for data ingestion and decision support.

Watch the Episode:

Episode Highlights:

More Resources:

• Timon's LinkedIn
• Luminovo's Site

Transcript:

Zach Peterson 

I sorry Timona Ruban is that how I your name okay and then founder and CEO

Timon Ruban 

Founder and managing director.

Zach Peterson

Managing Director, okay, of Luminolo.

Perfect, okay, let's go ahead and do this intro. Hello everyone and welcome to the Altium On Track podcast. I'm your host, Zach Peterson. Today we're talking with Timon Ruban, founder and managing director of LuminOvo. I recently met with Timon at PCB East and I'm very happy to have him on the podcast today. Timon, thank you so much for being here today.

Timon Ruban 

Thank you, Zach, for having me. Very excited.

Zach Peterson 

Yes, absolutely. So I've talked to some folks at LumaNova before and it was very fortuitous to run into you at PCB East because I got to learn a little bit about the program and meet some folks on your team. What I'm wondering if you could do just to help us all get started is if you can introduce yourself to our audience, tell us about your background.

Timon Ruban 

Yeah, happy to. So my name is Timon. I am German, originally from Munich, though I think I spent around six to seven years of my life in the US in total. I was there in Florida when I was very young and then I did my next exchange year during high school and I studied at Stanford. I guess that's the most important recent stop in the US, my master's degree in electrical engineering. And that's at Stanford is where I met my co-founder Sebastian and the two of us over five years ago.

decided to start LumiNovo and start this electronic supply chain platform that we do today. At LumiNovo, I'm responsible for the whole product and engineering side of things. So I'm basically a little bit more of the nerd compared to my co-founder who does the sales and revenue side of things and I'm responsible for the product vision and what we build. Obviously, I also talk to our customers a lot to understand what their problems are and how we can help solve them.

Zach Peterson

So you said you got your degree in electrical engineering. Did I hear that correctly?

Timon Ruban

That's right. Yes.

Zach Peterson

So how did you feel the need to get into building a software platform? Because I think a lot of EEs probably don't gravitate towards building software unless maybe they do like embedded development.

Timon Ruban

I guess I came to software through machine learning actually. So my undergrad was in Zurich at ETH, which was also electrical engineering, very theoretical. And we did a lot of statistics and I actually really liked the statistics part as linear algebra for signals and systems, but also useful for machine learning topics. And then when I came to Stanford, this was 2015, 10 years ago.

I can barely say before it was cool for the AI thing. It was already pretty hype back then, but nothing compared to the hype that is there now. But because of the small hype in 2015, everyone was talking about neural networks for the first time. This was two, three years after AlexNet won the ImageNet challenge. I'm not sure if you're familiar with that. That's basically classifying cats and dogs much, better than any other computer vision algorithm had done before. And because of that at Stanford, I really dove.

into the whole topic of machine learning, deep learning, neural networks. And that's a little bit to seek into software too, I guess. Then I started scripting and Python and a lot of things to like build these neural networks and build these deep learning models and then wrote more and more code through AI. Then basically ended up in the software side of things.

Zach Peterson

Did you have any kind of vision for like using machine learning within electronics, like, you know, robotics or, you know, intelligent IOT, that kind of thing.

Timon Ruban

Yeah, guess the answer to this question gives a little bit of the history to LUMINOVO because when I graduated from Stanford both me and Zeba, my co-founder, we were Fulbright scholars. So it was great because it paid for Stanford. But it also sucked to some degree because we actually had to leave the US. It comes with a two-year home residency requirement so we couldn't stay and start working in...

the valley like many of our friends. We had to go back to Germany and then we were back in Germany like, damn, what do we do here? And we thought, okay, we don't want to start work for some company. We want to do our own thing and then started looking into what we could do. And the only thing that we really felt we had an edge in, you better than some other people.

It wasn't per se lecture engineering because we had no experience working or designing PCBs or much, but we had been at the cutting edge of deep learning and machine learning research while I was in Stanford. So I TA'd the machine learning class there from Andrew Ng, who was the founder of Google Brain. I actually worked at Google and one of their deep learning teams. And so we came back and thought, let's do something with AI, which is a terrible way to start a business because you have this hammer and you don't know.

what where the nail is that you want to hit it on. And because of that, we actually didn't find a product idea or problem. And what we ended up doing instead of is some consulting. So for two years afterwards, the first three years right out of graduate school, basically, we did AI consulting. And at the time without an electronics focus. So we

consulted for lots of local German companies like Volkswagen or Infinien, but also insurances like Munich Re or local TV show companies like ProSieben, you probably don't know. It's popular trash TV channel here in Germany and helped them understand how they can use AI. Where they cannot use AI, a lot of it was about clearing up people's misconceptions of how or how it cannot be useful.

Timon Ruban

And we did that for two years and after two years, we got a little sick of the consulting model again. That was really exciting in the beginning because we hired people for the first time and we first time sold something to a big company like Volkswagen or Infinion. But we were doing consulting and so we thought originally we had wanted to build a product.

And then we went back to the drawing board. And at this point, my co-founder is also an electrical engineer, even though he does sales. He's gone even further away, I guess, from his studies. But we both had this affinity for thinking that electronics is just cool shit. It's like fun to work on little prototypes and things. so we ended up talking with a local contract manufacturer at the time, just outside of Munich, the CEO who had just joined and who was trying to...

improve their supply chain processes and they showed us how they're run and a lot of them were run on Excel. Getting Excel bombs from their customers and sending out Excel sheets to their suppliers to get quotes for all the parts which are available and the prices and more Excels to do the calculation of the labor for the assembly and it was just like one very long arduous process. We're talking and this is 2020 now so five years ago and this is how the idea slowly started.

to bubble up that there's a lot of kind of, to some degree, low hanging fruit of Excel based processes that we could improve upon, but also a very high ceiling because obviously the electronic supply chain is probably the most important and definitely most complex supply chain in the world. And so that got us really excited to look into the topic of electronics more deeply. At that point,

AI didn't play a big role anymore. we were actually like after two years of being, it has to be something with AI. At this point, we were like, actually doesn't have to be AI first. Let's find a real problem that we care about, a way we can find customers that we can help.

Zach Peterson

Okay, so that's kind of the background of how you got into building LuminOvo. Maybe if you could just tell us a bit about what the platform does.

Timon Ruban

Yes. So our mission is to become the go-to tool for procurement and quoting of electronics. So basically anything that has a printed circuit board at its heart. I would say today we have around 250 customers today. Around 80 % of those are contract manufacturers because that's where we started. So one of the main use cases that they use Luminovo for is exactly the process I had just mentioned.

when they get basically a new request for quotation, a new project from one of their either potential customers or existing customers with some product, with a bomb, with the Gerber files and PDFs about how to assemble it.

and they need to answer questions about their supply chain. Like how much is it going to cost if I want to make a prototype next week with 50 boards and how much is it going to cost and is everything going to be available and what are the risks in my supply chain in terms of life cycle and compliance when I want to go into serious production next year and produce 50,000 of these boards.

And actually if I do 50,000, then maybe I don't just want the list prices that I can get from Digi-Key or Mouser. I actually want to run a negotiation with some of the larger manufacturers or suppliers for the specific project. And then I have to run through that. And this whole process end-to-end, basically sourcing.

both for the material costing and the labor costing end to end to estimate all the costs and risks in your supply chain for a PCDA is what we do.

Timon Ruban

at Lumion today. And more recently, we've also been starting more and more to work together with OEMs, obviously the customers of many of our customers, because if you are on a procurement or supply chain team at a larger OEM that has many different products where they work together with different EMS, you often ask yourself these same exact questions about what are the costs and the risks in the electronic supply chain.

Zach Peterson

So when you mention risk, to me, I think this means obsolescence or inventory risk. I think obsolescence is the one that a lot of people focus on. And if you read a lot of guides and application notes or whatever online, they certainly talk about obsolescence a lot. But I think inventory is kind of the other part of that. How often does inventory get replenished, or does it get

replenished at the level you need so that you'll always have sufficient inventory of that particular component. Are those the two areas where you're talking about for risk?

Timon Ruban

Yeah, there's a third one that also fall, I would put them kind of under a risk. I guess they're even more go or no go, but it's compliance. I guess you wouldn't have to put that on a risk. We often put it on a cost and risk because it's nice to have two words. And in the end, you have to make sure that for in the industry, if it's a medical device, at some point you have to check that all your parts are actually compliant with the regulations. Most common one being ROHS and REACH.

But there's also an automotive, the different types, AECQ, and different types of compliance thing that you have to watch out for, that you need to check just like you should be checking obsolescence and the life cycle of all the parts that are on your board. So that's the third next to inventory obsolescence compliance.

Zach Peterson

So when you say check, you mean like literally test different components.

Timon Ruban

Exactly. And look up basically for every last part that is on your bomb, which if they are ROHAS compliant, if they are generally been available to order, if they are still active or already end of life and what the last product change notification is that you got about that part. And obviously doing that manually becomes a big hassle to a degree that people sometimes don't do it.

And basically lose risk in the end. Risk always comes back to cost, right? So if you don't have the risk thing under control, it'll just lead to like crazy cost explosion at some point. The worst, worst case would be you need a redesign because that part, there's no part available. And then slightly less bad case, but still pretty bad. You have to do some risk by where you pay a hundred X the price for some part on your bomb, because there's just no other way to get it. And you don't want your whole car from BMW not

be able to be manufactured because there's some resistor missing. That's usually super cheap.

Zach Peterson

I think there's another element there, which is risk of recalls, which we've talked about in the past, either because parts were mislabeled or maybe they did come from Digikey, but they were a return and somebody bamboozled Digikey and got some fake parts into Digikey and then they sent them to you. So yeah, I think there's a lot of ways that you could end up in any of those situations where you have parts that don't function the way you thought they would or...

Timon Ruban

Yeah. Yeah.

Timon Ruban 

Yeah.

Timon Ruban 

Yeah.

Zach Peterson 

aren't going to be available at the quantity you need.

Timon Ruban 

There is maybe one, think, think this is interesting also, I mean, for the audience of this podcast, right? Cause we are first and foremost a supply chain platform. And so our main target group of the people that we usually work with will be people on the procurement or supply chain teams of an OEM or an EMS.

And not usually directly the engineer who designed the PCB. But, and this is a big but because obviously we compare ourselves to a lot of other human and supply chain tools for other categories that are not just electronics. And there's lots of tools that try to do sourcing and risk management on a very category agnostic way of like, doesn't matter if it's soap or food or metal or whatever it is you buy.

The big difference and why we thought there is a place for an electronics specific supply chain platform that is different from these horizontal tools is actually the collaboration with engineers. And the thing that makes electronics different and makes the supply chain so much more complex compared to buying some material that any supplier can give to you and manufacturer is that alternatives are so hard.

to find and choose. So as a procurement guy, you're often left out to dry in electronics. You can't just be like, here's my specification. Do you supply it? Like figure it out. The specification is this part. And then if that part's not available, the EMS will come back to you and will be like, the supplier will go, well, we don't have this one. There's an alternative. Maybe this works.

But now you're often already lost on the procurement side. And you have to go talk to your engineer who designed your board to understand if this is actually a valid alternative. Will it really fit in our design? What are the important characteristics here again? And it's this complexity that's really at the heart of what makes procurement for electronics so much more complicated. And there's so many things that then derive from this, the whole setup with.

Timon Ruban

global chip manufacturers and distributors and everything. That's very different than if you're a procurement person for glass, where you have some bottle and needs to look like this and anyone who can make glass bottles could probably make it. And then if these guys go bankrupt or they have too high a reprise, well, you go to the four others that can and you don't really need to talk to engineer anymore because they can all quote it for you. And that's really what makes electronics much more complex and more fun, obviously.

more intricate, more problems to solve.

Zach Peterson

Yeah, I would agree with you 100%. It relies so much more on collaboration with the engineer because if you think about some areas of the industry, like for example, consumer electronics, there's always a focus on cost. And so I think the procurement folks are, if they are going to suggest alternates, they probably just look at the cost. And if it's a throwaway consumer thing, maybe it's a toy or whatever, probably doesn't matter, right? Go with the cheapest stuff. if you start talking like aerospace,

and you have to do procurement. Swapping out a resistor can be a big problem because they could have tin or cadmium. They could have high vapor pressure metals, that type of thing. And it's a big reliability problem. And yeah, it's not just so simple as, well, this part is out of stock, but we saw this on the recommended part for Digi-Key. Can you qualify this?

Timon Ruban 

Exactly. Yeah. Exactly.

Zach Peterson 

Yeah, yeah. So you also mentioned that, or I believe you mentioned that people can also work on the entire assembly, which is not just the components, but it's also the PCB. So I was just going to ask. So your system can essentially look at the PCB and help a customer get a quote for the PCB?

Timon Ruban 

Mm-hmm. It's basically when I come.

Timon Ruban 

So our system, when you want to talk about basically PCBA, Printed Circuit Board Assemblies, and we help you understand the costs and risks. The risks are obviously mostly in the components. What we just talked about, compliance, lifecycle, it's all got those components, understand the risk in that supply chain. The cost, however, comes from three different sources. One is the components. The PCB often makes up around 10 to 20 % of the costs.

Zach Peterson 

you

Timon Ruban 

and the labor. So we do all things. We help you estimate the costs, like source all the different parts from DGKey and Arrow and directly from manufacturer. We help you estimate the labor costs and we also help you get instant prices for the board itself, which I do think is one of the unique, now your proposition is out of our software. The way it works is that you can just upload either we support Gerber files or ODB++. Let see the design files of your board. And then we have...

a Gerber engine that will automatically extract the geometrically relevant characteristics of the board that you need to give to a PCB supplier so they could tell you the price. it's things like simple things like the width and the height of your PCB that's pretty easy to get. But much more intricate is getting things like the minimum clearance or trace width.

And a bunch of other, think we have around like 60 different specifications in the end that you can specify. And since we also directly collaborate with many different PCB suppliers who will tell us their capabilities, like the machine that they have set up of like, what is the smallest trace width that they can do at which price. And they give us, we work together with them to put into our system, their pricing engines and their formulas of like given these different specifications.

This is the cost of such a PCB. We can actually, in many cases, an instant estimate, in some cases even binding price from some of these PCB suppliers after you upload your Gerber files. Since that is one of the important processes, if you want to turn out a quote quickly, if all the components are done in a second, but you have to wait a week to get that manual quote from a PCB supplier, you also haven't won that much in terms of...

throughput speed for turning around a quote.

Zach Peterson

Okay, so working with PCB suppliers, where are they located? mean, are they located in Europe, all over the world, Southeast Asia? Familiar names we might know?

Timon Ruban 

Yeah, so I guess that's good seek into where we're at as a company right now. So we started, as I mentioned before, we moved back to Munich, we started here in Germany and very quickly built up a team all over Europe and also customer base all over Europe. But I would say pretty much...

We dabbled last, I was at IPC Apex last year for the first time, but I was a little bit of a board dabbling and not making big bounds and leaps, but it's really, I would say since the beginning of this year and also this IPC Apex was our big kickoff starting in the US.

Now have a Luminovo Inc. We just got access to the government cloud on Azure to be able to soon also provide ITAR compliant software in the cloud. And we've hired a core team in sales and customer success in the US now, where we're now rolling out more and more in the US. So in Europe, I would say we're already to some degree the leading software provider.

electronic supply chain platform definitely when it comes to contract manufacturers and in Europe many of the largest PCB suppliers that was your question and like zuehrt I think it would be the more English pronunciation usually when I say the German one people don't know who I'm talking about but that's actually they're the largest PCB manufacturer in Europe actually they are one of the PCB suppliers that are directly integrated there

Zach Peterson 

I was just about to say.

Timon Ruban 

direct PCB shop when you up go on their website, ours. It's actually powered by our engine to help you extract all of the things to get enterprise. And here there's many other like safe PCB and others that we work with. And how this usually works, it worked in Europe the same way it is now step by step working in the US, is that we usually work with contract manufacturers and OEMs that we work with.

And when we onboard them, we ask them, who are your PCB suppliers? The three that you usually source from and then have them may help us get intros with these PCB suppliers and onboard them to the platform. So as we get more and more clients in the US as well, this process is slowly starting to happen now that we are onboarding more and more of like local American PCB suppliers as well. Kind of the same process that happened in Europe over the last couple of years.

Zach Peterson 

OK, that's really impressive, actually. And I think it's a great way to grow into some of these bigger PCB suppliers. So when someone's preparing a quote based on their design, do they see a menu of options from different suppliers? They can see where it comes from, what's the lead time, what's the cost per unit, those types of things.

Timon Ruban 

Yeah.

Timon Ruban 

Exactly, you have a little menu or there's a little like green, red, yellow system that is basically first just tells you, given your specification, which of these manufacturers can manufacture it, given the capabilities that they have, which one do we then actually have an offer, which one can do it, but you can't get an instant offer. So sometimes their pricing engines are just configured that if it gets really in the extreme cases, they want to have a manual look over it instead of giving out an instant price.

That's kind red, can't do it, yellow, can do it, but no instant price and green, can do it, here's your price. And right away.

And yeah, as you said, after you upload the Gerber, as you see a visualization of what it looks like, you can see how we classified all the different layers, which is a really important step, actually. And there's a menu where you see all the automatically extracted specifications. We are currently working, we'll probably launch that very soon, also on using AI to extract all the remaining specifications that are not in the Gerber, but usually in some PDF with a stack up.

or just some PDF that has a table where it says what the surface finish should be and what the solar mass color is and all of those things. Because that was as of today, it still like the last remaining manual step. You put the curbs and then you have to manually click through the things that are in your PDF. And we're about to get rid of that manual step as well by automatically extracting that from the PDFs. But once that's all configured.

Zach Peterson 

So that's where AI comes back in, right?

Timon Ruban

And this is actually where it comes back in. So we have a very much back to the roots moment at Lufthansa right now, because five years ago when we started building this platform for the electronic supply chain, Seba and I, told ourselves.

No more AI. We made that mistake when we wanted to use or not looked around for how we can use AI and then it kind of got us sidetracked and we ended up just doing this consulting business. Let's just solve real problems. And if AI happens to be the way to solve it, great. But if it not, then that's fine. And we're just going to build the software to actually solve problems.

And it pretty much, I would say it took like four to five years to now be back at a time where I'm actually very excited about the problems that AI can solve, not just about AI as a solution, but really the problems that it can help solve. there's actually many, many different use cases where currently either in almost about to ship stage or in exploring stage, I think there are as many opportunities now to actually get AI back into our software.

Zach Peterson 

Excellent, excellent. So one question I have, you've kind of, I think, addressed this unintentionally, but how complex does a PCB need to be before the system can't process it? Or is it just solely based on the vendor capabilities?

Timon Ruban 

Yeah. It is actually mostly the vendor capabilities. So we have seen some pretty insane 16-layer boards that can all be processed in our Gerber engine. Sometimes the really complex ones just take a little longer. The short ones will be done in like 10 seconds of processing and running through it. And the larger ones can load up to a minute. But not like an hour or something. So it's all still pretty quick. And...

Since the algorithms for extracting the min trace width actually do not change much, no matter how simple or complex the board is. That isn't really the bottleneck. The bottleneck is to some degree that vendor by vendor, they usually have for kind of their core business where they get to see a bunch of PCBs that just like this. They're usually very...

secure in the price that they give out. So they have a clear pricing engine themselves where they're like, yeah, well, if it's these specifications, it's going to cost us much. Here's your instant offer. Whereas if it's very complex PCBs, where just the type of specification is of a sort that they haven't seen very often, those are usually the ones where they still want to have that human check. They'll still probably be internally then for them a price suggested, but they just want to double check that to make sure that their pricing engine for these very complex edge cases that they haven't tested a lot.

are not going haywire and giving out a totally wrong price. So usually it's more level of complexity that basically depends on the confidence of our PCP vendors in their price, whether they're willing to give it out on the click of a button or after a human looked at it and said, yeah, this is fine, you can send it to LumiNovo and give them this price to show.

Zach Peterson 

So what's the user experience after somebody gets their PCB quote and their parts quote, and they click, OK, I'm going to buy it from this vendor? What does that experience look like? Is there a communication platform, like somewhere that the vendor can communicate with the customer, and you guys are kind of the intermediaries? I think that would give an opportunity maybe for feedback to come back to the PCB designer.

Timon Ruban 

you

Timon Ruban 

Yeah, we actually as of today only to a minor degree do the actual last step of purchasing and especially if you're looking at it from the EMS or OEM perspective where you're someone you have a board and you want to understand the cost and risk for it.

It is something that people often ask for. It depends a little bit. seems like the more smaller customers usually ask for it. The larger ones have very complex ERP setups where they do all of their purchasing and inventory management. They're usually not that interested in triggering the purchases from LUMNOVA, but rather do it from their existing setup. So as of today, the journey stops there. Usually we have via APIs integrations towards our customers' ERPs where they can pull in basically all the offers that they decided on, that they chose in this now

the pump and then trigger the actual purchasing from the ERP. Our focus also currently is still more on the strategic side. So rather than moving from this quoting stage towards more operational procurement and purchasing, what a large part of our focus has been and still is, is on getting even more strategic. So as opposed to

just having a nice user workflow where you can quote one board, having more strategic dashboards that give you an understanding of your portfolio of hundreds of boards, understanding where the risks in the supply chain lie across all your demands for these hundred boards and how that bubbles down to the actual parts that might be shared between some of these designs and how the price and lead time and stock.

history has developed over time, which are all the types of information that you usually want if you're more of a strategic procurement, a strategic buyer who maybe runs large annual negotiations of multi-million volumes with your suppliers one or two or three or four times a year.

Timon Ruban 

That's currently been a big focus of ours and helping these more strategic, also monitoring workflows. guess this monitoring is important keyword. That's a big new thing that we're currently working on is going from just helping with the operational, like got a new bomb. I want to understand the price. I to understand that risk right now to a more.

proactive in terms of our software monitoring of well across your 200 assemblies here this part now the price the stock totally plummeted so you might want to look at that to catch that risk early before it's completely gone or here the price trend has been trending upwards maybe you want to buy now or here it's been trending downwards maybe you want to renegotiate with your supplier these types of strategic workflows

Zach Peterson 

So speaking of strategic workflows, I think one thing that companies are doing is they're looking at more geographically segregated, let's say, or something like this, production of a product where they can produce it closer to their end customer. That way they don't have to produce everything in Taiwan and then ship it back to the United States where it gets tariffed and then sits in a warehouse.

Timon Ruban 

Mm-hmm.

Zach Peterson 

and then ship to 10 other countries and then it gets tariffed there too. Instead, they can just produce it closer to where it needs to be and then ship it directly to where it needs to go. Can your customers implement that type of strategic sourcing?

Timon Ruban 

I think the closest thing that comes to mind of like a use case that recently has been, we've discussed with many of our customers now as we're expanding, as I mentioned before, also more towards the OEM procurement side, as well as the contract manufacturer on EMS is creating that transparency across multiple production sites from an OEM and EMS perspective. So we have more and more cases now where we have

customers on the OEM side who are working with two to three contract manufacturers in different countries in Europe and Asia and the US. And that are because of this kind of even more global production as they shift things from one country to another and transparency around the stock levels at the contract manufacturers, which is often a stock paid for by the OEM.

How does it compare to the demand of what I'm produce where and which country? And so kind of this demand versus stock transparency across all of your contract manufacturers is something we've heard a lot from many of our OEM customers, but even from the EMS side, people sometimes talk about how they need to share this with their customers and if we could help make that more transparent.

And I think that's a topic we could be uniquely well suited to address because we have this large customer base on the contract manufacturing side and we often integrate in the ERPs. know the stock levels. And so the way we think about it is that we want to give these contract manufacturers the ability to choose which of these, this data they want to share with their customers. I think it touches to a, sorry if I'm going on a tangent here, but it touches to a topic that's actually very...

important to many of our customers is this like, controlling your data, right? Because we do help them monitor, work with their supply chain data lot. And so to us, that's very, important that we don't want to be this like data spider tentacle getting everything and sharing with whoever needs it, but that we, people who own their data, really are in control. And in this case, it's a big win-win actually, because we have

Timon Ruban 

These OEMs that are our customers that work with two, three, EMS that are all our customers that want to strengthen their ties together. where if both sides give the double opt-in of saying, yes, you can share my demand forecast with the EMS automatically. You can share your stock level directly. We can connect these links in the supply chain that help people manage. guess then it comes down to risk of like demands versus stock levels. How much do I need to buy?

much better. And this is the reason actually we call ourselves a platform. It's a confusing word in the software world sometimes and different people mean different things with this word platform. But the reason why we decided to call ourselves the electronic supply chain platform is because making these point to point links that both sides benefit from is very much at the heart of what we want to enable to have less and less friction in this global supply chain that is the electronics industry.

Zach Peterson 

Okay, okay. So you've been doing this for a while. You're now getting back to your roots of bringing some AI back into Luminolo. I wanna know what's on the roadmap for new features in the product.

Timon Ruban 

So I think there's three topics I'm most excited about. One we've already touched upon and this is becoming more strategic in terms of monitoring costs and risks. So that's a big topic that we're working on and collecting, A, the right data, the right data partnerships. There's many, just recently closed, that we'll be happy to soon for these costs and risk topics.

to just enable this more strategic approach of monitoring across all your portfolio of PCBAs, the cost trends, there are also lessons and product change notifications. So this monitoring topic across your portfolio, that's topic number one, I think that I'm very excited about. Topic number two is supplier relationship management.

And so that's something that's not at all done by any of our customers with Luminolo today. So it's supplier onboarding and management. Most of our customers use us to do sourcing, right? So they have a list of approved suppliers that they want to work with. Maybe some of them are preferred. There's some that are blocked. And then with in our tool, they can give us these preferences. These are the proof suppliers. And then you can work with who are sourced from them, both automatically via API, as well as through more manual.

quote request workflows where you send out an email and they submit it in a supplier portal. But where did that status ever come from in the first place, whether the suppliers approved or not? And so there's this topic of onboarding, collecting compliance information, doing evaluations of your suppliers. And that's something that all of our customers do. Everyone in procurement and supply these supplies have to deal with these topics.

It's very closely related to sourcing, obviously, because you want to have this actually feedback of your onboarding process feed directly into which suppliers that you can source from. so we just this month actually kicked off a new team. We have a few, two, three design partners from our customers that are working very closely with us to help us shape this new module. And so that's, guess, second topic I'm generally very excited about. It will probably be a few months before this goes into a better.

Timon Ruban 

release to be available to more of our customers. And I'll have to go into more detail here. And the third topic, we've also somewhat rushed on that. It's been a personal favorite of mine because it is back to the roots, is bringing large language models back into the workflows of LumiCoco.

And so we're starting mostly with the topic of data ingestion. So this basically means today, lots of data ingestion workflows at LumiNovo. get a PDF from a supplier who made you an offer and they didn't want to fill out the Excel you sent them, but they send you a PDF. And then you have to get it into the system to have in one source of truth, all of your quotes and costs and prices. And so this, get that PDF into LumiNovo is one that large language models are really, really good at.

Something will launch very soon. Different PDF topic we talked about before is the PCB specifications. That's another data ingestion. Get this PDF with the PCB specification, the stack up drawing, extract the relevant information and get it into a new host so that you can get your instant prices. But another big topic that's at the very heart of our work today is the topic of bomb importing. So you have these axles that can be incredibly messy.

Every company has a different process, lots of human errors in how these exels are filled out and sent around. And that's another topic that's within this realm of data ingestion that we're currently...

Looking to make much more agentic. We already before had some AI but it was more expert systems a lot of heuristics and rule-based in this bomb importing But now making it truly agentic with like smart large language models that can iterate based on feedback from the user to be much much quicker and importing these bill of materials and Yeah, as another topic

Timon Ruban 

I'm really excited for, and it's really just the start. So we'll start with data ingestion. But I think once you're done with data ingestion and those things work really well, the next frontier for these large language models is decision making. And then we're back to the monitoring thing of like, okay, across all of this data, what are smart recommendations of which things you should be looking at?

trading off two part alternatives and maybe actually having the LLM make a recommendation of how this could fit or not fit, which issues are the most important ones. There's a lot of this decision making support that I think these tools can be very helpful for. And here we're in the beginning exploration stage, but I would say in the largest grand scheme of things is the third topic I'm very excited for.

Zach Peterson 

What about something like DFM or DFA? Because if someone uploads their PCBs, you can see everything in the PCB. So there could be an opportunity for AI to look at the PCB. Maybe there's 10 common DFM errors that the system can look for and notify the user before they send it off to get quoted. And then, of course, it gets quoted with the manufacturer. But the manufacturer

notices and comes back and says, you know, hey, you've got this problem. We need you to fix it. know, nobody likes that.

Timon Ruban 

Yeah. Yeah. Yeah, I fully agree with you. Yeah. That's also a topic. think to some degree, we discussed this with some of our customers often because it's a very obvious, almost like next thing lying next, right next to our Gerber engine processor of extracting one sort of things from the Gerber files.

And so far, because we are a supply chain platform more than we are like a design tool, our focus has mostly been on how can we make this engine better to get more instant prices? So like, how can we make it more reliable to get the right specification out of things that we can then double check against the capability of the PCB manufacturers? And this is almost the event, right? To some degree, you're checking your characteristics for pricing against the capabilities.

Zach Peterson 

Yeah.

Timon Ruban 

And so I do think that's also something we will continue to invest in. We have one team that actually works just on this whole topic of PCB engine. think they're right now very busy with actually onboarding new suppliers in the US. So that's why right now their capacity is little bit limited to tackling huge new topics like the DfM thing as we are picking up speed in North America. But I do think we'll get back to that.

Eventually there is, mean, one, it's not really a topic of like, this is a thing we'll be working on, but a general trend that is also related to LLM in a very different way that I see. And I'm very excited about as a software company. Let me know what is that with this whole AI assisted programming with tools like cursor and Claude co, I'm sure if you're familiar with them.

There's a lot of trends that are making software development much easier and faster. And so we're also exploring a lot of like how we can get that integrated. We're today around 30 people in the product and engineering team in LumiNova. Like 30 people nonstop working on writing code.

talking to our customers to improve the software. And if we can use these AI assisted coding workflows for ourselves to write as much code and as good of code as 60 engineers would do, then we can do twice as many things. And we can do PTFE at the FM as well as onboarding new.

PCB suppliers in North America. And I do think that this will not be unique to LumiNoble. But I do think that's a trend that we'll see that will be easier and faster to build software. Hopefully that will mean for us that we can solve many, many more problems for our customers. Because it's both the biggest blessing and the biggest curse of being this electronic supply chain platform is that the surface area.

Timon Ruban 

is so wide, like the supply chain topics you could tackle, which falls under the supply chain hat. And there's so many topics of real problems that our customers approach us with, from DFM to you talked about purchasing before, where you could do more automation, to better integrations actually towards the ECATs and the PLMs, towards new modules that I'm excited about, the supplier onboarding and supplier management. There's so many workflows.

within this thing that we are usually constrained not by interesting or important problems to solve, but by capacity. And I do hope that these tailwinds that are coming with AI assisted software development that we'll be able to deliver more value faster to our customers over the next one to two years. But knock on wood, we'll see how it goes.

Zach Peterson 

Yeah, yeah. Well, what I'm hearing is you're growing, you're focusing on your core competencies, and you're focusing on, well, there's not a shortage of problems to solve. So I think that's excellent. And I want to say congratulations on all the success. One last question. Can people contact you to get a demo of the platform?

Timon Ruban 

Yeah.

Timon Ruban 

Yes, absolutely. I think the easiest way is just to go on our website. It's luminovo.com. Not to be confused with Lenovo. luminovo.com. And there should be a big blue button that says reach out or request a demo. And if you put in your email...

Zach Peterson 

You

Timon Ruban 

someone from our sales or customer success team will reach out as fast as they can to schedule a convenient time to show you all the things we just talked about in this podcast. You can, if you want, always reach out to me personally. My email is timon at luminovo.com. I'm also always very happy when people reach out to me directly so I can hear the voices directly from our customers or potential customers.

Zach Peterson

Excellent, excellent. Well, Timon, thank you so much for being here today. Again, congratulations on all your success. I'm excited to see where you guys go next. And of course, we would love to have you come back on in the future to talk about LuminOvo again as you develop more of these features in the platform.

Timon Ruban 

Yeah, amazing. should check in again in a year and see how many of these things have come to fruition that I just got really excited about. So I'll be curious myself to see how fast we can move.

Zach Peterson 

Let's do it. Let's do it. We'll put it on the calendar. Thank you so much. To everyone that's out there listening, we've been talking with Timon Ruban, founder and managing director of LumiNovo. If you're watching on YouTube, make sure to hit the subscribe button and hit the like button. You'll be able to keep up with all of our podcast episodes and tutorials as they come out. And last but not least, don't stop learning, stay on track, and we'll see you next time. Thanks everybody.

Timon Ruban 

Thanks,