Judy Warner (00:01.4) Hi everyone, it's Judy Won and welcome back to this week's Ecosystem Podcast. Are you listening to all the AI hype around hardware design and wondering if it's ever gonna make a real difference to your work? Well, I'm sort of feeling the same way, but I'm very encouraged by our guest today, Colin Stoner, who's the co-founder of Znode. Colin and his team are baking in their engineering wisdom right into their AI enabled comp- component search tool and making very real strides forward. I think you'll enjoy this conversation and learning about what Znode is doing. Without further ado, let's jump into our conversation with Colin Stoner. those Judy Warner (00:01) Hi Colin, great to see you again. It's been a while. I'm excited to have you on again and learn what kind of progress you've made at Znode. Why don't you start out and introduce yourself and tell us a bit about your startup Znode. Collin Stoner (00:14) Absolutely. Well, thanks for having me on, I always appreciate it. My name's Colin. I'm the co-founder of Znode. I've been an engineering consultant for about 16 years doing electrical engineering on circuit boards. And I started Znode because it's all the tools that I wish I had while I was doing that. Judy Warner (00:30) What exactly is Xenode? Just as an overview, you don't have to dig into the details. get into that. Collin Stoner (00:38) Sure, so Zenon's an electronic component intelligence engine. It's all of the tools for AI to be able to work with your components, understand your designs, and help you automate the tasks that are boring, slow, and monotonous, so you can concentrate more on your design work and less on finding the information you need to do it. Judy Warner (00:59) Yeah, we used to talk about that a lot at Altium about engineers go to do all the interesting work and then they find themselves stuck on spreadsheets and component data. So I know this is definitely an area where the industry appreciates seeing innovation. So over those 15, 16 years that you were designing boards, ⁓ what kind of boards did you design and what kind of applications were you working on? Collin Stoner (01:27) Yeah, so I focused a lot on industrial automation and more industrial type projects, but I've had clients from the US Army and NASA through a bunch of Silicon Valley startups and a lot of quantum computing companies. And then a whole bunch of stuff in between some consumer products, some industrial commercial products, and a lot of different boards for a lot of different purposes. I've done about 250 since 2016. 250 unique designs ⁓ and then I've taken about 150 of them through manufacturing ⁓ in our own facility that I started up. Judy Warner (02:09) Wow, that's amazing. So a lot of designs really if you count them by year one, some only do, you know, three a year or something like that. ⁓ So you've seen it all. You've seen them sort of the easy stuff, the hard stuff and everything in between and then wrestled with manufacturing, which is no small thing. ⁓ So ⁓ tell us about during that time what your frustrations were and the problems. Collin Stoner (02:23) ⁓ yeah. Judy Warner (02:37) And bottlenecks you saw that made you want to start Znode and create something different. Collin Stoner (02:43) Yeah, well, so I'm always doing my design. And I get stuck spending huge amounts of time on tasks that are relatively monotonous, but incredibly necessary. I need to go make the pin out for a new schematic symbol. It's got 99 pins on it. So I'm sitting there copying the names of 99 pins. it's necessary that you get this right and I double and triple check it because if you get it wrong, there goes your board design. But also, this isn't really a design task. It's not. really super fun, ⁓ but it's critical. ⁓ I'm doing an RF design and I'm placing vias in exactly the locations that they need to be. ⁓ But it's RF and they're shielding and we're doing lambda over 20 and so they're every 40 or 50 mils. They're really tight and I'm sitting here placing vias for an hour and a half. Or I'm looking for a part and I know kind of what I need, ⁓ but I don't know who makes what, know, what kind of chip that does that. a client that needed parallel RGB into HDMI. And I figured there has to be a chip that does that, but I don't know who makes it. And I don't know what it's called. Is it a converter I see? Is it a transceiver? You who knows? Anyways, I'm doing all these tasks and they seem to be something that can be automated, but... 10 years ago, five years ago even. They're pretty resistant to that because they're all pretty different. They all require a pretty firm understanding of what you're doing. They're all looking at a document, understanding what it's doing, ⁓ navigating your way around a couple documents, whatever it is. ⁓ So I always wanted to do this. And like most engineers, I want to be lazy and make an automation tool to do the stuff that takes a long time. But they were resistant to that. So when Chat CPT came out, ⁓ it kind of clicked that this can automate those. And now all of these tasks that take me a long time that aren't really part of the design, they're just necessary to execute it, they're just dominoes waiting to fall. ⁓ And now we can automate them. So that was something that got me into thinking about what Znode eventually became. Judy Warner (05:13) So when you started the company, how did you envision or what did that workflow look like for you as an engineer, the way that you wished it looked? And I've talked to folks in the industry that talk about wrestling through a 200-page data sheet, trying to find a needle in the haystack. I'm sure that was one of the things. So what was your vision like? Collin Stoner (05:42) Yeah, I mean. Judy Warner (05:42) If you could wave a wand, what would that be like Colin? Collin Stoner (05:45) Exactly, yeah. mean, we started with all kinds of ideas, ⁓ you know, because the sky was the limit, kind of still is. ⁓ But, you know, we wanted to do schematic, you know, stuff. We wanted to automate anything we could get our hands on ⁓ that was part of that design test. But everything we tried to think about, everything came back to, can you understand the component? can you get the information about them? Because before you can make any design decisions, you need to know what you're designing with. ⁓ So no matter what we wanted to do, we were kind of forced to look at the parts first. ⁓ And so we realized everyone else doing something in the space is going to need that. We need it for whatever we want to do. And the ideas were many. ⁓ but we had to do this first and we realized that in and of itself is a pretty large task. And it gets rid of a lot of the stuff that. We don't really want to spend time doing, but we have to do. Searching for data sheets, searching for parts, searching in the data sheets, finding that little footnote at the bottom of a table on page 372 that says, by the way, in this one special situation, you can't do this. And that happens to be exactly what you're doing on that board. That's why your camera's not running. Ask me how I know. Yeah, so you know. Judy Warner (07:09) Yeah, I was going to say that sounds like a personal story. Collin Stoner (07:15) we needed the component intelligence to be able to do any of those things. And that's how Xeno became what it is now. Judy Warner (07:24) So some people call in, have commented on my podcast when I talk to people who are innovating with AI and hardware design and they're like, yeah, yeah, yeah, it's all chat. It's all like ⁓ pie in the sky, but there's really nothing tangible here. How do you, you know, I'm sure you have customers and early adopters and you know, you've gone through the last two years developing this. What do you say to people? I mean, really. You've seen it. You say chat GPT and hardware in the same sentence and people leave the room. Right. So clarify what you mean and what do you say to your skeptics? Collin Stoner (08:06) Yes, so I guess AI is moving at an incredible speed. there were, especially early on, huge problems. Still are issues. But a lot of those are being resolved as the AI becomes more capable of navigating designs. And what we're doing is we're putting together all of the data and all of the specialized framework that it needs to be able to do that intelligently to speak the same language we do as hardware engineers. ⁓ Judy Warner (08:34) Hmm. Collin Stoner (08:36) early on, you know, it's a generalist and not that great of a generalist. But without that infrastructure, we'd tell it what we need and it would interpret it like a lay person and give us something completely worthless. But as it's advanced and as we've built all of the tools around it for Xenode, now when you require something, you know, some spec out of it, I need the maximum voltage. It knows you're not talking about the absolute max. I'm talking about like, I need the max voltage I can put on this without blowing it up. It knows you need the recommended specs. And while that's clear to a human, ⁓ without the context and understanding and capability of a modern AI and all of the tools around it, it might not have known that early on, but that's not the case anymore. ⁓ and they're advancing still. They've got a long way to go, but they've come a very long way. So if you use the chat GPT for hardware design a year ago, two years ago, even a couple of months ago, ⁓ the story is changing, especially with having specialized tools around it. Judy Warner (09:52) So I'm going to play the devil's advocate here a little bit. if you're a hardware engineer and it's getting better, what's unique about Znode instead of me just going as a design engineer and querying chat GPT. Collin Stoner (10:05) Sure. And that's a great question. ⁓ all of the Claude, ChachiPT, all of the big base models are incredible products. ⁓ But they are generalists that understand something about the world. But even the most brilliant person ⁓ or scientist, whoever it is, without information and tools and experience can't really do what we do. ⁓ They might be able to learn and adapt quickly, but without the tools, without the understanding, and most importantly, without the data, what is the actual part you're using? A human engineer would be just as useless without that information. So having an AI combined with something like Xenode, that's a reservoir of all that data. all of that understanding and all of our expertise as the founders of Zeno, me and my co-founder and all the people that work for us, we're distilling everything we've learned and putting it into tools that allow the AI to be helpful to you. So then you can use it to accelerate your design. So it's not about challenging or making a better model than chat GPT or quad. It's about giving them the tools and infrastructure and context they need to be able to be useful in an application that's very specific to our field. Judy Warner (11:34) There's been a lot of people, mostly my old guy colleagues are like, it'll never happen. But I am encouraged by what Znode's doing, Quilt or CircuitMind, there's so many, and I see all of you ⁓ just making us old folks look silly because you are making really meaningful progress. So tell us a bit more about how you're training Znode for context and then if you have any use cases you're talked about the camera one with you which was the old workflow. Do you have any use cases either for your team personally or perhaps customer use cases where you can show the kind of precision and capability of Znode? Collin Stoner (12:25) Yeah, a lot of the, I mean, we've got some basic applications. So, you know, I need to find a part, but I don't know exactly what it is. Just describe it. You don't have to go dig around on a catalog page somewhere and hope you find it. You can ask for it and let the AI dig for you. ⁓ And that's kind of the basis of a lot of the applications we have. ⁓ We're not trying to replace engineers. ⁓ We are engineers. We like engineering. ⁓ What we don't like is spending a lot of time doing things that are engineering adjacent that take away from actually doing the engineering. So, you know, if I need to go find a part. describing it, telling it what I mean, giving it the specs that I require and letting it go find it for me, let it go dig through the ket log and find something that's viable to give me four or five possibilities that are really pretty close and then I can make the final choice. That's kind of the basic, that's what you use Znode for. But more generally, if you've got a task that requires digging around and doing a whole bunch of this kind of work that's design adjacent but not necessarily critical design that's what we want to use Znode for. So a power budget. I've made, can't count how many power budgets, but you know you're making a battery powered device and you've got 25 different chips on there. I need to know how much current I'm drawing in every state of the device operation so I can figure out how long it's gonna stay alive for with this battery. always a long and boring task. Let me go dig through every single data sheet that I got and find where it says what the quiescent power draw is, what the active power draw is, what sleep mode does. You end up with a million numbers in a spreadsheet and it takes half a day or a day to put together. ⁓ The idea with Znode is that's a task in AI to just send it out to go and do all that, collect me all of the information from every one of these data sheets for every part in this bomb. ⁓ Now you're probably gonna need to go check and make sure it got the one that you want, because some of the chips have pretty complicated low power modes. Which low power mode did you mean? But it jumps you straight to where the data is, shows you exactly where it got it from. And it did it for 25 chips, and it did it in a minute or two. So if you need to go and dig in a little bit on something, you've taken a nap that might have taken you all afternoon, down to a few minutes. And that's the kind of thing we want you to be able to use it for. Judy Warner (14:49) Mm-hmm. Collin Stoner (14:57) not to replace doing the design work or replacing the need for an engineer, replacing those tasks that don't really help with the design. They're just necessary to complete it. Judy Warner (15:15) That's a powerful example, pardon the pun, of something where it's about workflow and time, right? And like you said, it's not like, ⁓ shop this out to AI and then trust it and go build a physical product. It's just speeding up the research time and the grind and getting you to the things you need to pay attention to critically faster, which seems like a... Collin Stoner (15:19) you Judy Warner (15:44) huge win to me. Any feedback from your customers or any someone's come back to you and said, my gosh, we did this thing. And it was amazing. And it saved us this much time, money or like your camera example. It caught me before I spun another board or anything like that. Colin? Collin Stoner (16:09) So we've had a number of aerospace folks key into this kind of product immediately because they have a lot of data that is not easily searchable. And so they're kind of the first people that have looked at this seriously and said, you know, this really could change the way we work. We've got all of these disparate data sheets that are, ⁓ you know, internal to our org. ⁓ They're not easily searchable. ⁓ but also very complicated and very expensive. So we don't want to screw up. So that's been kind of an early use case that we've seen, but we're just starting to get a lot of these tools out into the world. And so we're very excited to see how other people start using it rather than just me. ⁓ I've used it to do the power analysis example to find specific items. Judy Warner (16:44) Right. Collin Stoner (17:07) about a very specific feature, like the camera example, you know, trying to get it work for a board. And there's some footnote that says, hey, by the way, in this mode, you know, you can't turn it on like this, you need to do it a different way. You know, you need to go through I square C and make a change. Judy Warner (17:11) Mm-hmm. Collin Stoner (17:27) finding those little things, take the debug time down from, you you're looking at it for a day or two or maybe even longer to, let me just do a quick search before I spend, you know, the rest of the day chasing my tail, ⁓ just to see if there's anything that mentions this. Describe your symptoms, you know, ask it, find me anything on the data sheet related to any of these issues. Judy Warner (17:42) Mm-hmm. Collin Stoner (17:54) So we're just now releasing a lot of this into the world and we're really excited to hear how some people are using it. Judy Warner (18:01) Yeah, sounds exciting. Well, this makes me want to ask you about some of the features. before I do a follow up on that was When you're using Znode, does it intuitively help you maybe catch blind spots? So say if you've only been laying out boards for five years or something, will it help you catch that footnote or like what's the limitation of the query or how it kind of watches your back, so to speak? Collin Stoner (18:30) Yeah, so it's able to go through all the data sheets and information for you. So depending on how you ask it, it will guide you. So if you ask for guidance, it will give it to you. ⁓ but we've kind of defaulted it as a professional tool. So it's hard to give you the most concise answers that contain all of the information without being excessively wordy. ⁓ So you could certainly put a workflow together, you hey, I'm trying to do this. Do you have any feedback for me? ⁓ And it'll give it to you. But by and large, we've built it so that you can say, I need information about X, and Z, go and have it. Judy Warner (19:01) Mm. Collin Stoner (19:11) give you here's the results, here's the register table, here's the spot in the data sheet for the pinout that tells you exactly what you need to do. ⁓ So that way you can get that information to move on. Judy Warner (19:22) Good. Okay, let's talk specifically about some of the features that you referenced. ⁓ It's been a while since you and I've talked and I know you've done a lot in the last year or more. So what are some of the features and the progress you've made so far and things you're particularly excited about or that are maybe coming out soon that you think our listeners would be interested in? Collin Stoner (19:49) Yes, we've got a bunch of new features, a lot of them launching in the next month or so. ⁓ The biggest change we've made since we talked last is we've implemented something we're calling Guided Discovery. So instead of just type a search in and it gives you a catalog of results, ⁓ what it does is it gives that to an AI, which will go then and search all the places that could be. And it will give you some guidance as to, here's what I found, here's what's available. ⁓ And if you want to see some of those results, go here. But also, if you're looking to narrow this down, here are the things you need to care about. ⁓ If you are a board designer that's only been going for a few years, or you're new electrical engineer trying to understand what's out there, you can say, I need one of those rectangular connectors that I saw on the computer when I was building it. The ATX connector. What is that? Go. Judy Warner (20:23) Mm. Collin Stoner (20:39) And it'll go and find connectors. It might ask you a little bit more about what you need, but it'll show you. It's like, OK, well, we've got some stuff from Molex here. Here's the KK series connector that everyone uses. But you probably don't know it's the KK series until you've worked with it for a little while. But also, there's this other style connector from Amphenol. And here's some JST connectors. Not sure which one of these you need, but here's some options. And here's the way to get to search. So that's Guide to Discovery. ⁓ That's out right now and we're really excited about it and seeing a lot of improvement in the way people are searching with it. But we're about to launch an ALT tool. ⁓ So... I'm sure everyone who's been around for a little while remembers 2020 and 2021. I feel like I spent half of my time finding component alternates back then, and I'm worried about to have another event that's similar. So finding component alternates is usually something you need to do, and when you need to do it, your hair's on fire for it. yeah, that $25,000 Judy Warner (21:29) Mm-hmm. Precisely. Collin Stoner (21:46) five chip, you know, that's always a big problem. So we've made a tool that kind of distills a lot of the things that I had to do back then into a single tool. Here's my part. This is what I'm doing with it. I need some alternates. Go. And it'll go and it'll find potential candidates for you. And then it'll look into each one of the data sheets for those and say, all right, so your original part has a pinout like this. This one has a matching pinout. This one has a pinout that's close. Here's what they are. Here's where they differ. Here's how their specs are different or the same. If you have any special features that you care about, hey, I'm using this chip, it's a 24-volt chip, but I'm only using it for 12, then we take that into account. We'll show you chips that are 18 volts instead of going all the way to 24. Maybe that opens the field up a little bit, and that gives you some more options. Or you can say, hey, I'm using this chip, but I'm not using the enable feature. We can go find chips that don't have an enable feature, but still do everything else. So that way, you can find something when that chip and everything in its family is out of stock, and you need to swap it with something. Now you've got a choice and a tool that will help you a little faster. So that should be coming out in the next month. So we're really excited about that one. Judy Warner (23:00) Yeah, I bet. Yeah, I think when we were preparing for this, you were talking something about a bomb tool too. Am I remembering right? Collin Stoner (23:10) Yeah, yeah, so we're putting together some tools for working with your BOM or a list of parts that allow you to apply Znode against many parts at one time. That's really the use case that we want to see in the futures. Both search, but I've got a whole bunch of parts and I need to do this for all of them. So as a contract manufacturer, when we build boards, know, what kind of part it is influences how we build it. ⁓ Is this service mount or through hole? That's a pretty straightforward question, but it's time consuming when you need to do it for 100 line items in the BOP. ⁓ Or, you know, something maybe a little more critical that's a diamond in the rough kind of problem. ⁓ can I wash this part? And I need an answer for every part. The answer for 99 % of parts is yes, you can wash it just fine. But for some MEM sensors, you can't. ⁓ And if you don't look at that on your BOM beforehand and search specifically for that, you might be in a world of trouble after you build 50 boards and then can't wash them, but they have acidic solder on them, you know, for cleaning processes. ⁓ Judy Warner (24:18) Mm-hmm. Collin Stoner (24:21) So we're making a tool that lets you upload a bomb and then ask a question against every single one of those parts, potentially many questions. So you can find that information and you can do it quickly. that. Judy Warner (24:34) I love that you guys are building in that ⁓ because you have a background with the manufacturing. think that's so incredible. And I know they come from really mistakes or, you know, missteps you've had along the way, but that you're really taking that wisdom you've developed over more than a decade and baking it in, which is just so powerful because they're just not enough. time in the day and it's so easy to overlook something like how you clean flux off a board or the way a connector connects or you know all of that. So I think that's something really unique from the people I'm talking to that you're you're baking in that wisdom. So I think that's really incredible. So regarding the... Did I interrupt you there? Colin, you were starting to say something. felt like I talked over you. Collin Stoner (25:31) Or you let me, I'll go on all day. ⁓ Ask me how I know how painful it is when you run a board with clean flux on it and you can't wash it. Been there, done that. And it was an expensive rework. So ⁓ yeah, we're trying to bake all of that in. we're... Judy Warner (25:47) Yeah. Collin Stoner (25:52) We're electrical engineers. Me and my co-founder are both electrical engineers. I've been in manufacturing for the last seven years now and design for 15. So I've seen a lot of these problems. know what's tripped, know, trips us up as engineers because I've stumbled over all of these stumbling blocks myself. ⁓ Judy Warner (26:10) which is so critical. You know, we've talked about this idea of DFM forever, but there is a clear delineation and there has been for a long time between what you're designing and what's manufacturable. And there's so many little nuances in that. And so if you can bake in some of those lessons, I just think, wow, what a gift really to engineers. Collin Stoner (26:38) And that's what the ALTS tool kind of is the first version of that, because the guided discovery is sort of a basic search tool, which is very useful. But ⁓ for the ALTS tool, a lot of it is, this is how I searched for ALTS. ⁓ And I've baked it into a process that. guides the AI that's helping you along the right process. Here's what you should be looking for. Here's what we need to find. And all of that's baked in. But it's still flexible enough that you can tune it to your own situation. And if you have some expertise on either the specific part you're looking for or in general, you can give it that and it will take it into account. So that's what we're trying to do with Zeno is we have all this experience and understanding. We're trying to build the data, the context, the tools that let you exploit that experience. ⁓ Even if one, you haven't. spent a long time in the field so you don't have it yet. Or if you have spent a long time in the field and you don't want to, you you have a job to do, you need to get it done, and you don't want to spend the time we've spent explaining that in general AI. That's what it seems like. Judy Warner (27:48) Right. Right. That's incredible. I think that's so empowering and useful. ⁓ So I assume that on Guided Discovery, your alt tool and even the bomb tool, you've been doing beta and you have some maybe customers you're partnering with on that. Collin Stoner (28:12) Yeah, so we've had some of them in beta and we're still working with people to get kind of the kinks worked out and everything going. So those will be released kind of on a rolling schedule over the next month, maybe two. ⁓ But we're always making feature updates and always looking for people to help us ⁓ try it out, understand how everyone else might use it. Because I know how I might use it and that's how I design the initial one. But how does it actually get used? ⁓ That's always super interesting to It's very fascinating to see the way that other people will approach the same problem. So yeah, we've been doing some betas and we're getting ready to release a lot of them shortly. Judy Warner (28:44) Mm-hmm. In that way then you're sort of not just baking in your own wisdom, but you're baking in sort of a more global wisdom of other people's workflows too, which is very cool. ⁓ From those beta cycles that you're going through now and you've gone through in the past, what are some things you've learned along the way or tweet as a result of the feedback you've gotten? Collin Stoner (29:18) So I think understanding how other people search through things has been interesting for me, because that's not something you normally see unless you're standing over someone else's shoulder. So that's influenced how we've built the tools and all of the context for the AI doing that search. So it kind of takes into account, how do I look for things? What do I start with? ⁓ Where should it be going in the data sheet to find these kinds of things? ⁓ Judy Warner (29:28) Mm-hmm. Collin Stoner (29:47) And so that's, I can't talk too much about how specific customers are doing things, but understanding the different ways you might get to the same answer has been very helpful because then we can have the AI follow all of them up ⁓ and check its answer. So it doesn't just do it the one way and say, well, good enough. ⁓ will be able to, all right, so here's the answer I got. think it's right. Judy Warner (29:52) Right, right, we're right. Right. Collin Stoner (30:13) let me try another route and make sure I arrive at the same answer. And that's been helpful to, one, find errors, but also find misunderstandings in what you actually typed in. We actually found something with this recently. The question was, for an accelerometer, what is the lowest power draw? What's the lowest current? Judy Warner (30:17) Mm-hmm. Collin Stoner (30:39) but the accelerometer's got a couple modes. It's got a completely off, or well, not completely off, a full suspend. You're not doing anything. It's just on the board. And that's one answer. And then there's the accelerometer's still running. You can get data out of it, but it's in the lowest power version of that. And then there's a few more above that. So the way the question was worded was, what is the lowest power mode in general? ⁓ And it wasn't quite what the user was expecting because they were thinking lowest power where I can still use it. Not lowest power, I'm in sleep mode, don't talk to me kind of mode. And it gave an answer for one, but it also gave the context around it and how it arrived there. So, you know, that way you could understand, this is, I asked the question wrong and like, technically the answer wasn't wrong. You asked for the lowest power mode and gave you the lowest power mode. What you meant was the lowest power mode while I'm using it. And so given the answer and the context around it, we were able to. Judy Warner (31:35) Right? Collin Stoner (31:48) they saw that and then were able to go through and, no, no, no, I meant this and got the correct answer they were looking for. But giving that additional context was something we learned early on when we were working with users because, you know. I ask a question a certain way, and especially as the developer, I'm giving it as much context as I can, ⁓ which is a key for everyone else using it, as AI in general. Give it as much context as you can, you'll get much better answers. But having the AI give you context back, so that way you know it's the same thing, that was something we noticed ⁓ and tried to build in. Judy Warner (32:19) Mm-hmm. Right? You're reminding me a while, within the last year, I heard Elon Musk say something about when he was talking about Grok, he says, right now Grok spends like 95 % of his time figuring out what the heck you want. You know, and when I listen to you, that really resonates. And what I hear you saying is that with Zeno, you're building in that context and you're also baking in. ⁓ the broadness of the way different ways would query and give ⁓ the AI that you're using in Znode that intuition, right, over a time which is incredible. And I think that is ⁓ the way these tools are going to become truly useful and powerful. Collin Stoner (33:16) We're, yeah. Yeah, and we're doing as much as we can with what users give us. ⁓ But if I can say something, general AI use ⁓ Znode and everything else, the more you give it, the better it'll be. So if you give it a three-word request ⁓ or you give it a three-word request with, also, this is the circuit I'm trying to build and this is what I plan to do with it, you'll get a much better result with the second one. Judy Warner (33:52) Great. Well, Colin, congratulations. You guys have come so far since the first time I've talked to you. And ⁓ I really appreciate what you and your team are doing really to improve the workflow and use AI in a meaningful way that impacts their workflow. I had just one more question for you. You mentioned early on automotive, and you were saying they had a lot of internal documentations and workflows that they had to bake in. I was thinking about the compliance issues around automotive. Like how do you make sure that given those specific applications like automotive or aerospace and defense, how does that play in regards to your tool or does it? Collin Stoner (34:42) Yeah, so we're starting to work with some of those folks specifically for some of those problems right now. So a lot of aerospace has this this kind of trouble because the parts are either controlled or they're the kind of parts that they sell three or four of in a year because they cost $100,000 in there for a satellite. ⁓ And this is just now becoming a much bigger industry. I think in the last five years, like the number of space startups has just exploded. ⁓ So the industry is still trying to adapt to that. So we're we're trying to build tools that cater to that. and working to create solutions that can handle the... special certifications and special requirements around those documents. So if you're in aerospace and this is a problem for you, know, trying to navigate the DLA site and find all of the parts that are available and, you know, you've got to call the manufacturers to get a data sheet, ⁓ come talk to us. We're working on solutions for that right now. ⁓ Solutions that may even be on-prem depending on what you're building. But... ⁓ Yeah, those are active challenges we're trying to solve. ⁓ For automotive, ⁓ lots of certifications are required. That's a great use for, here's my bomb. ⁓ Are there automotive versions of these chips? The ones that we don't have. Judy Warner (36:04) Mm. Collin Stoner (36:05) And for the ones that we do have in this bond, can you help me find a second source? Because most automotive stuff requires multiple sourcing. Those are time consuming tasks. So that's the kind of thing we build Cnode for to be able to do this deep dive into a data sheet. It's like, yep, I see you've got the commercial version of this part. Here's the automotive one. It's AAC 100 qualified, whatever it might be. Here's the alternate part number. And then here's a second source. And so we can update the bomb with two columns for each one of those. Is it qualified? What is it qualified to? Whatever the question might be, feed that bomb into ZNode and see what you can get. Judy Warner (36:48) It's incredible. You got your work cut out for you, buddy, but you've come so far. And again, congratulations to you and your team. Before I let you go, Colin, how can our listeners ⁓ reach you and learn more about you? And what would be next step recommendation if this conversation has resonated with them? Collin Stoner (37:09) Yeah, absolutely. ⁓ So check us out at xeno.ai. We're running a promotion for all of your listeners. ⁓ We'll put a QR code on the screen or in the description with a free month promotion. So you can try xeno.ai yourself. ⁓ But you can go to xeno.ai right now and sign up and get started right away. Judy Warner (37:32) Thank you for doing that. It was so nice. I love when our guests do that for us. So I really appreciate you doing that. Well, Colin, thanks so much for your time. I know you're busy and your time is a premium. So I really appreciate you sharing with me and our audience all the work that you're doing. Thanks for joining us and let's do this again soon. I'm sure we'll have more to talk about later this year. Collin Stoner (37:37) Absolutely. Absolutely. Thanks so much for having me on. Judy Warner (37:55) My pleasure. For our audience, make sure you go check out those links and hopefully I'll put a QR code for you on the screen once Callineyeve connected so you can go get a free month and just test drive it yourself and see how that might help you in your workflow. We appreciate you joining us this week on the podcast. We'll see you next time. Until then, remember to always stay connected to the ecosystem.