From CAD Chaos to Clarity — with Drafter and Trace.Space

Episode Summary

The episode "From CAD Chaos to Clarity — with Drafter and Trace.Space" delves into the challenges faced by engineering teams in managing design and manufacturing processes, focusing on innovative solutions offered by Drafter and Trace.Space. Chris Barton, co-founder and CTO of Drafter, shares insights on how his company addresses the gap between design and manufacturing through automation and intelligence, drawing from his decade-long experience as a mechanical engineer. Janis Vavère, founder of Trace.Space, discusses her journey in requirements management and engineering, highlighting the transformative potential of AI in this field. Both guests emphasize the importance of integrating advanced technologies to streamline workflows and enhance productivity.

During the conversation, key technical insights emerge, particularly regarding the integration of artificial intelligence into engineering processes. Janis Vavère shares how Trace.Space leverages AI like GPT-3 for requirements autocomplete, significantly reducing repetitive tasks and allowing engineers to focus on more creative aspects of their work. Chris Barton discusses Drafter's approach to CAD design automation, enabling mechanical engineers to achieve epiphany moments where they realize the benefits of using advanced tools over traditional methods. Strategically, both companies underscore the need for collaboration between design and manufacturing teams to drive innovation and efficiency.

For PLM and engineering professionals, the key takeaway is the transformative power of integrating AI into their workflows. By embracing technologies that automate repetitive tasks and enhance creativity, these professionals can streamline their processes, improve product quality, and stay competitive in a rapidly evolving technological landscape.

---

Full Transcript

Michael Finocchiaro

And we're live. Welcome to another edition. I think this is 2728 of AI Across the Product Life Cycle. Very happy today to present Chris Barton of Drafter and Yannis Vavère, if I pronounced that correctly, of Trace.Space. Why don't you guys introduce yourselves before we get into the thing. Go ahead, Chris.

Chris Barton

Thanks, Fino. My name is Chris Barton. I am the co-founder and CTO of Drafter. I spent 10 years as a mechanical engineer and lived the pain firsthand of everything that engineering teams are doing to build high quality, high precision hardware. So Drafter is focused on go to manufacturing, all of the work and surface area between design teams and manufacturing teams and building automation and intelligence forever and there.

Michael Finocchiaro

Awesome, we'll get into a little bit more of that as we go. Janis, do want to tell us about you and Trace Space?

Janis Vavere

Yeah, of course. Hi, Fino. Thanks for hosting us. So as you mentioned, I'm Janis and thanks for being with me here, Chris, as well. Looking forward to this call. But yeah, I'm in a space or a vertical called requirements management or requirements engineering or systems engineering, if you call it. But the way I got into it was actually 10 years ago in 2016, I was the first European.

Michael Finocchiaro

You're welcome.

Janis Vavere

employee for a company called JAMA Software. And I actually did my master's thesis about implementing and doing requirements engineering at a Latvian state government project. And so I've been in love with this field and the people in it for most of my career in tech. And even back then 10 years ago, I was very frustrated with the fact that it felt like the requirements engineering tooling has stopped innovating.

Michael Finocchiaro

yeah. With doors.

Janis Vavere

⁓ Nothing's going on, doors is still de facto the main thing. And I remember telling JAMA executives that we should just do something about it. And then about four years ago, ⁓ I saw an opportunity that with the modern software architectures, you can already build a much better, much more interesting tool with sort of much better API and ability to ⁓ sort of configure it and be flexible. And of course, the AI came out right around the time. And so we built the first ever enterprise AI native agentic ⁓ systems requirements engineering tool. And it's going really well. So I'm really looking forward to this conversation today as well to hear how the whole industry is changing and to talk about it.

Michael Finocchiaro

you Yeah, it's really amazing to see, I mean, I've talked to about 150 founders now. I think my database up to 600. I've got about 50 in the queue. It's just incredible amount of innovation you guys are doing. And I just have so much respect for the amazing work you guys are doing. The first question on the podcast we usually ask is around, and you started answering this already, Janis, but 2022 OpenAI moment. world completely changes, right? We can't think about computers or interfaces or anything the same anymore. ⁓ Are you immediately bullish on the whole thing or do you still have a little skepticism and where do you sit on terms of ⁓ the open-eyed revolution back four years ago in 2022?

Janis Vavere

I love this question because it just brings back all these memories, but we had an interface for where you kind of notion like interface for writing requirements back in the day. And the first thing we did, we took GPT-3 and integrated the API to create something we called requirements autocomplete. It's still on YouTube. You write the first part and then it finishes the rest of the sentence. And then we started building the whole company and...

Michael Finocchiaro

Nice.

Janis Vavere

We hired the first really smart engineers that also had actual ⁓ systems engineering background. And then they were telling us that all these requirements are very bad. They were like, yeah, but it just, why, why wouldn't it work? And so we were always bullish on the integrating and trying to build something on top of this technology or with this technology. ⁓ And I remember giving the first task to to the engineering team to build something where I can talk to my requirements. And we tried and we tried and we tried. And I think we probably built the architecture and the backend for AI that if we did that, we would probably now be billionaires because we did all of this agentic stuff before anyone, because there was nothing like that existed. We had to build all the infrastructure. ⁓

Michael Finocchiaro

think we probably will be able to add some extra capacity for the AI. But if we think that, we'll probably not be filling it.

Janis Vavere

Honestly, the AI and the LLMs were just not good enough for engineering back in three, four years ago. And so we had the architecture, but the results were just not good enough. The output was not good enough. ⁓ But we've always been building, trying to see what's the state of the art. And I think now it's coming together, but it was tough three years, to be honest. being bullish on technology that's not producing anything interesting for real engineering.

Michael Finocchiaro

Chris, how was your experience?

Chris Barton

Yeah, I think we had, you know, a healthy level of skepticism to start. You know, we focus on high precision documentation. This is not, you know, words that, you know, describe things. This is engineering drawings that are an agreement, a contract between two different businesses. And so the difference between, you know, 70 % and 100 % is massive. And if you have big guesses or hallucinations or anything else. On these drawings, it's worthless. And so we took the deterministic approach of producing high quality drawings that abide by, you know, the standards that exist out there by engineering best practices. And we know exactly why that approach has happened. Obviously, you know, over the last three years, we've been very inspired by the future of how AI is going to help these different businesses. But at the beginning, you know, it was almost unfathomable to have AI drive this product in a way that it's going to truly add value for our customers. And obviously, it's improved. We have some very exciting things up our sleeves on where this is going to go. ⁓ But we now have a product that is reliable, repeatable, and is able to actually add value without having this layer behind it.

Michael Finocchiaro

Yeah, we'll talk more about that whole probabilistic models on deterministic thing, I think a little later. Because first I'd like to dive into, once you guys both start producing your product, after, well, actually it was back then it was called what? was Codify and there was all these other names for these agents that have now morphed into Claude Code more or less, or Cursor. Although Cursor sounds so 2024 now, right? Doesn't it? So how at a Drafter, how are your developers using AI in terms of software development? How has software development changed at Drafter since they opened the moment?

Chris Barton

Yeah, I mean, here's the reality. know, AI is going to have an impact on every single business that exists. ⁓ And one of the best places where it adds a ton of value today is in development, right? If you are a developer and you are not using these tools, you are going to get left behind. And so, you know, from day one, when somebody works with us, ⁓ it is not a matter of, you know, if it is which tools and how, and how you're going to leverage it. And it's a big focus for us as a team right now. not just on the development side, but how we are going to implement the assistance of this intelligence for us to be better as a team, to deliver higher value to our customers. And then obviously, in the ways that our product is getting used, it has an impact on every single aspect of the business right now.

Michael Finocchiaro

And not just from cruising PRDs, it's in terms of structure and architecture.

Chris Barton

writing test cases, know, doing QA, ⁓ you name it, right? All of the places where you are doing repetitive tasks. Obviously there's human review behind every aspect of this, but the speed with which it can allow you to achieve these things is remarkable.

Michael Finocchiaro

Same experience for you over at Trace.Space, suppose, Janis.

Janis Vavere

Similar but but there was a so we hired ⁓ mainly quite senior software engineers and there was a high level of skepticism all the time like ⁓ as founders and and leaders of the company we always had to Sort of ride the hype train on Twitter and be like well look at this. What's what they are doing? What's look what they are posting and

Michael Finocchiaro

Ha It'd be cheerleaders.

Janis Vavere

And then, and we shared that with the engineering team and they're like, nah, this is like, you know, all hype and, and it's, but, but, ⁓ it's, course it's, it's, it's increased productivity, many, many fold overall, over time, ⁓ because people that couldn't write front end can write front end now because they just go good software engineers and people that didn't know SQL, ⁓ can write. SQL queries and so on. it's only getting better now. ⁓ And this year, especially the last few months, it's incredible. Even our most skeptic software engineers are just incredibly productive right now. And just to give you an example, our ⁓ CEO, co-founder who has finance background is very smart, obviously, but he's never... learned about coding, he's writing extensions on top of Trace.Space API. So it's like the level of like what's possible today. like if our non-technical co-founders can ⁓ provide proof of concepts to pilot customers to what's possible, then imagine what our most senior software engineers can do with the agents. so things are completely changing right now. And we call agents in-house squires, you know, like everyone has their squire.

Michael Finocchiaro

So is the Squire suing up for scrums now instead of the developer if he's out sick?

Janis Vavere

Yes.

Michael Finocchiaro

They are. The agents are showing up at their place? No. In terms of the actual products, terms of Drafter and trace.space, I'm imagining that you probably have features of the software. The AI is really visible to the end user through a copilot or a chatbot or something, but there's probably also some of the plumbing.

Chris Barton

Yes.

Michael Finocchiaro

of how the thing was developed and the models maybe you're using underneath. So where exactly is AI implemented inside your stacks?

Janis Vavere

⁓ We implemented everywhere. And what I mean, and we call ourselves AI native for that reason, is that there are, we implement also mix of like NLP, like deterministic models for things that like to check quality of something in a deterministic way, for example, and surface that information to the large language model, to the AI. But the way we we do things is that we, in general, for anything that is ⁓ critical, the AI generated result has to go through a human review. So the engineer is the one, the user is the one approving its results and putting them into the permanent database. But there are also low impact things like surfacing potential ⁓ matches or links to other requirements or When you create a new thing, it generates an item name for you. So it's just basically things that just take a long time and we just implement them without asking anyone whether they would like to generate this thing specifically. We just present to the user and they just can say accept or decline. But those are things that just make users individually way more productive. ⁓ because they just don't have to think about these things. They don't know that they exist. ⁓ And then on top of that, have, of course, the agents who can give them tasks and chat with them and they understand the whole context and they understand the rules and how the system has been set up and all the data model and examples and so on. So it's all over everywhere where we can. think through the prism of AI first and then we...

Michael Finocchiaro

So it's all over anywhere that we can, which means there's a place in the AI first, and then

Janis Vavere

think about how to implement it in a way that's reliable for engineering.

Michael Finocchiaro

we think about how to implement it in a way that's reliable for the environment. How about you, Chris?

Chris Barton

Yeah, I mean, we've been very focused on being deterministic, right? I can guarantee that we will produce a high quality drawing on our initial generation.

Michael Finocchiaro

And it'll be the same drawing the next time you ask, right?

Chris Barton

Exactly. And why that's important, right, is you don't want to have best guesses, right? We focus on design intent. Why the heck are you making this thing the way that it is? What is important about that? And we've been able to build that infrastructure layer of actually tangibly adding value. Why that's so exciting is we can now begin to put our tendrils of intelligence into this. And that's where we're connecting the dots, right? And the beauty of this is it's going to allow us to accelerate so quickly. This is not just a chat bot. that gives you a little bit of information that you could find anywhere else. This is exactly what your design intent is in the application of what is actually tangibly adding value. And so we've been very bullish on our approach of why we're building it the way that it is. And it's going to set us up in a great way to now bring in that level of intelligence directly into the application of what actually is important to our customers.

Michael Finocchiaro

So in your case, is it more putting guardrails around the stuff you're doing to ensure that there's deterministic results?

Chris Barton

It is ensuring that every single ⁓ moment where we're recommending something, bringing knowledge in, we can present that in a way that says, hey, we made these assumptions, and we did this math behind it for you, and allowing users to actually see that. It is not, here's a number, best of luck. It is tangibly saying, here's exactly how we got to this. And if you don't agree with that logic, power to you. These are the recommendations that we're adding. But especially with engineers, right, we're serving customers who are the most intelligent, most skeptical humans on earth. ⁓ As soon as you lose that goodwill, as soon as you lose that kind of moment of trust, it's all downhill. And so we've been really focused on building that from day one, first interaction.

Michael Finocchiaro

And are you guys building your own foundational models underneath your knowledge of design intent and requirements management? ⁓ I'm just wondering, are you thinking of building your own models and then you're using inference on top of that with maybe the customer could bring their own LLM because they've got their own open AI key? ⁓ I'm just wondering how you're doing that. level of inference you're using and how you're leveraging the commercial LLMs if you are or not.

Janis Vavere

We have very little proprietary stuff that we use, but... That's one of the reasons is because most of our customers, as Chris mentioned, are very conservative, very pragmatic people that ⁓ do high precision work. And on top of that, engineering is their core IP, the holy grail of what the IP they create, the products they create is. And so they have company approved models. Sometimes they are company fine tuned.

Michael Finocchiaro

Hmm.

Janis Vavere

aligned or just going through their enterprise subscriptions of OpenAI. We just have to make our ⁓ platform ⁓ able to connect to their models and their AI. We have some stuff that we do in-house, but that's the real inference is happening either through a contractual agreement with with the with true customers AI or through their own AI or through the big LLM providers and they are also just getting so good right now that it would be very hard to compete with trying to build something of our own for what we do.

Chris Barton

We're in a similar boat. It doesn't necessarily make sense to build our own models, especially with how we are focusing on the information. And of course, we take data security incredibly seriously with our customers. So ⁓ I would say there are going to be opportunities with data sets that exist in our public that we could have kind of a hybrid approach. Right now, it's really bringing the intelligence onto the infrastructure later that we already have existing.

Michael Finocchiaro

It makes me think too that since you guys are both playing on the edges of ⁓ where the big three are working, right? The Zemus, PTC and ⁓ Daso. How does that work in terms of the workflow of the engineer that's using your tool, right? They mean, they're probably using Creo or Katia or SolidWorks, Or NX. And then they're going to use your tool. And in the case of Yannis, they're probably using TeamCenter or 3DX or ⁓ or a windshield for the PLM. How do you guys deal with that? Because obviously you want to compete with them. You've got to partner with them, but you want to obviously replace the things that they can't do because they're so far behind in AI. I each of those companies is two, three years easily behind where you guys are. ⁓ How do you coexist and actually displace maybe sometimes the big three?

Chris Barton

I mean, I'll go first here. We are very focused on go to manufacturing. That is the gray area where CAD systems stop and all of the manual work and human review and jewels of human energy go into getting things right. If you look at the workflows that exist there, is, you know, emails and literally printed out PDFs and red lines. Yeah, it's crazy, right? And so this is our opportunity. Yes, you can make a drawing in these CAD systems. We have integrations that

Michael Finocchiaro

Mm-hmm. Excel's and Excel's and Excel's and Excel's.

Chris Barton

allow us to output our drawings directly into your CAD systems. It allows you to work in your PLMs, right? I can get a team onboarded in 60 minutes. We've been very intentional about that approach because design, you know, is this monolithic beast. We want people to continue to use those design tools. They're great for a lot of things. When it comes to making drawings, I can guarantee that we are much better than those CAD systems. And we allow you to be very intentional about what you're doing in bringing that design intent. into the workflow. Obviously when it comes to everything else that exists in that space, that's where we get to do our magic, right? We're connecting the dots that have never existed. That is where the intelligence of your decisions in the past and your historical, you know, records can all get connected so that when you're making decisions, you're certain that they are made with all of the intelligence and knowledge there.

Michael Finocchiaro

And in your case, you're doing primarily 2D. So do you have your own 2D kernel or are leveraging somebody else's?

Chris Barton

We have built our own 2D drawing generation. Obviously, ⁓ that was a bigger challenge than most people thought it would be. ⁓ But the beauty of that is I can work with SOLIDWORKS customers today, with NX customers today. It means that we can continue to serve all these different types of folks. ⁓ And more importantly, we have the digital output that as soon as you want a model-based definition file, all you have to do is hit a button. We're starting to have conversations with customers about that.

Michael Finocchiaro

Awesome.

Chris Barton

And so a lot of the big name customers that want to move into that digital era, really excited at the prospect of being able to be data rich with a 2D drawing and a model based definition that are digitally connected. We can help you do that.

Michael Finocchiaro

Okay, well, it's just interesting that, you know, we went from all these awesome UIs to just one line in the middle of the screen when we did Google, then we did it again with OpenAI and you're going from all these awesome 3D titles. actually, all I care about is this beautiful 2D exact deterministic one. That's really cool. Oh, sorry, Yanni.

Chris Barton

Yeah. And most people think, right, we, know, 2D drawings are the thing of the past. The reality is, is 95 % of the things made on earth, leverage 2D drawings. It doesn't mean that I don't want to move the industry into a digital era, but we need to meet people at the lowest common denominator of what exactly is done today. And that's the product that we have built. So I want to work with people to move into the future and we can literally lift all boats with this rising tide.

Michael Finocchiaro

It's Nice. So Janis, you've got some input coming from Chris in terms of 2D and to your requirements. ⁓

Janis Vavere

Well, definitely there are interfaces possible and I think that iteration, even in hardware engineering, will go through all kinds of agentic prompting of, we call those trade studies, but you want to change requirements and then you update the drawings and see what's possible. So I think there's going to be a lot of interesting things like there are things people like physics techs that I hear that are doing really well right now already. that some of our customers use and they're very interested to hear in ability to change requirements and see updated models and simulations and then get that feedback and go back into specification. So engineering is changing faster than ever, especially considering how slow and conservative it has been. But I love Chris's approach because I always say that we play the game on the hard mode. So we have an industry where the de facto main tool is IBM doors from the late eighties and nineties. And then the last innovation was JAMA and Polarion 20 years ago. And they haven't really changed since like 10 years or so. so we are going, well, CodeWimmer is kind of the same era. And so we are going head to head.

Michael Finocchiaro

Thank Yep. Well, what about code beamer though? This code beamer, right? Yeah.

Janis Vavere

against ⁓ tools that have been built for 30 years. AI is a big thing for us, but we are also just doing much better interfaces in every possible aspect, like better APIs, better UI. So there are so much data, especially in the AI world being generated and also in the physical world, physical engineering, where the systems engineering with... with these legacy tools, they just can't keep up. They can't do 10 clicks to make one link between two objects. And so we are also rethinking all those interfaces in the context of how much data and how fast the data is being generated and how fast it changes. So not only the AI has to help, the API has to help, and the interface has to help the engineers keep up.

Michael Finocchiaro

Hmm. you

Janis Vavere

And so we are going head to head, but people are very frustrated with legacy tools right now. They are integrated into their development. The processes have been built around them. But I hear from the people, the customers we speak to, they're very upset with the state of the art in this industry.

Michael Finocchiaro

Yeah, one of my customers, he can't even go beyond Creo 10 because of all the customization. just, it's kind of crazy. ⁓ Well, we're already four years into this crazy revolution, right? And we've seen things come and go. mean, like I said, sorry, Codex and cursor seems so 2024 now. It just seems so far away after the cloud code thing. And then whatever the hell that open cloud thing was, you know, and Malt book and all that crazy stuff. ⁓ It's hard to imagine how the pace of change, even like four years ago, you couldn't even imagine things changing as fast as they're changing now. ⁓ Do you guys ⁓ ask skeptical and as bullish as you were four years ago?

Chris Barton

I'm probably more bullish now. Mechanical engineering will be dramatically different for engineers next year than it was when I was literally doing design work, you know, 24 months ago. ⁓ The tools that are at your disposal, the processes of connecting different discrete data points and data sets is just coming. And that's going to be the tip of the iceberg, right? ⁓ You can look at all of the skeptics that came when... The internet happened or early AI, like we are now going to be able to dramatically change the way that hardware teams are developing the speed that they're working. And similar to how it's been happening for software developers of smaller teams, writing more code, having higher outputs. In the next year, few years, we're going to see smaller and smaller mechanical engineering teams achieving more and more and more. Right. And that is with higher velocity from their manufacturing partners with less people connecting the dots on procurement and supply chain and everything else in between. It is simply, you know, digital AI manufacturing engineers connecting mechanical engineering teams and their partners downstream.

Janis Vavere

Honestly, we were quite skeptical about AI until this year. It was getting better for building software, but it was just kind of not good enough for what we do. And then it feels like now we actually just released something we call Space Agent. It's like an agent orchestrator that you can talk to about your data and about what tasks you give it. And we see with our customers, it's making a systems engineer 10 times more productive. They have like a body to talk to and analyze the data and see the missing gaps and fill those gaps. And I believe next year we're going to have autonomous agents running all the time, surfacing issues and analyzing impact. And so things are going to change rapidly now. I believe there's no way any of the... companies that want to dramatically improve efficiency of their engineering teams will have to look at AI first tools. Otherwise, you just get left behind.

Michael Finocchiaro

Yeah, this is seems to me too that we're gonna see the collapsing of a lot of roles into this mechanical engineering role because the design for manufacturing, you're talking about Chris, and design for simulation. I mean, I don't think a mechanical engineer is gonna be allowed to just work only on design. They're gonna have to be thinking about all these other things. There's actually a question in the chat from Lawrence Cook, who's the co-founder and chief of product at Generative Engineering. Good question. He says, when engineering moves to a world where trade studies, simulation driven exploration and optimization becomes the norm rather than the exception. What is the role of requirements? Do we even need them anymore? What do they evolve into? So I think it's a really interesting question. Obviously more focused to you, Yannis, but I mean, and Chris as well. Let's go ahead.

Janis Vavere

Well, we see teams generate requirements from what they have in code and the diagrams they have. And then as they update the code, they update the requirements. Well, what we see is that the requirements is something that ⁓ you actually have to make sure that your product is well and alive and to be able to evolve it. And actually the agents use the specification to evolve the product. That's where your... your parameters live, where your decisions live. It's also where you get capture what's unique about your product, what your customers actually want. And you use that as a reference. Otherwise, if we just let the autonomous agents run everything, I think it would all approximate to the mean, to the... to the sort of typical, the slope of like whatever is in the model, right? So I think that the requirements or the specifications or writing things down, documenting your decisions, it will exist and maybe agents will write most of it, but the place to do that will be inside Trace.Space, for example. I truly believe that.

Michael Finocchiaro

it will assist and maybe agents will write most of it for the players to do that.

Chris Barton

You mean, I'll take a swing at this, which is as an engineer, what are you doing? You're solving a problem and you start first principles thinking, saying what is possible and how the hell am I going to get there? And that is requirements, physics, mass. What are the things that you need to achieve and all of the work, all of the iterations that happen today through human effort to get to a finished product, to get to a manufacturable, repeatable product is just effort saying. Do we meet the requirements? Do we change the requirements with the things that we have designed? And you iterate and you iterate and you iterate. And I don't believe that requirements are going to go out the door. You still need to know what problem you're solving. You still need to know why you are taking a deep approach on that. But when you get to the iterations that happen to get there, that is the time that's going to be compressed. You still have your levers of what am I trying to achieve? What am I trying to get there with? But all of the work and repetitive work and communication and discrete Data point connecting is the place where I think time is going to dramatically shrink.

Michael Finocchiaro

I don't know if you guys saw that, but there was this pretty awesome webinar between Cisco N-TOP and Astari Digital about ⁓ how they were working together, solving some problems. what's incredible is that the design space has become more or less infinite. You used to be able to test a handful of variables because that's all I could handle in my calculator. it's pretty awesome that now, thanks to AI and thanks to all this stuff, we can actually explore more as infinite design spaces, which is just kind of sort of mind boggling. Another thing I'd like to ask about, the answers I've gotten in the past are pretty interesting, is when I look at the demographics of what people watch the podcast, I do actually get a younger audience sometimes. There's about 30 % of entry level people. And I'm sure there's two parts to it. One is like, What should they be focusing on so they don't get AI'd out of a job because everybody's kind of worried about AI's going to steal my job. But even more importantly, shouldn't they be going into work for you guys instead of a boring job at Meta or just a corporate job at Oracle? Shouldn't they be going to, to draft or into Trace.Space because that's where you guys are doing the cool stuff, right?

Chris Barton

Listen, like, there are going to be different jobs out there for different folks with different risk appetites. ⁓ I would rather be on the cutting edge, you know, thinking and dreaming side that is going to change the way the processes exist today. I have worked, you know, in a cubicle farm where you own one little gear and that was all you did every single day. And, you know, that, that works for some folks, but that process to me was antiquated. It was old and. You know, the number one thing that I would encourage younger people to do is go out and get your hands dirty. Go experience building something. Try to build something on your own because you are going to start seeing the cracks in that process today. Find somebody who, you know, needs help, has a vision. Go work for them. Go work for them for free if it really means so. Go spend 30 days trying to help someone get there. And you are going to learn so much about that experience, about the gaps in connection and tools and everything else in between, and then take a step back and say, what do I want to do now? What did I learn from that experience and move into it the next time? And obviously, you know, AI is going to change a lot of things, but there are going to be tons of human jobs that need to happen to be creative, to solve problems, to see what is going on in the world and what the opportunities are. ⁓ I'm not worried about engineering jobs disappearing, but there's plenty of things that somebody could be passionate about to go solve.

Michael Finocchiaro

Thanks.

Janis Vavere

Yeah, well, for a, for a, just hired our first founding account executive in US and Amanda and she's from a marketing world. But what she realized was that how long I'm going to be selling marketing software to marketers that sell to other marketers, you know, and or market to other marketers in B2B SaaS and all that. And so there are people that just want to have some, some bigger meaning.

Michael Finocchiaro

Amanda here. Ha ha.

Janis Vavere

in their lives than just building B2B SaaS that's being sold to other B2B SaaS companies in Silicon Valley. And that might even all go away quite soon. And so finding people like that and people that actually want to have some ⁓ impact. Because in reality, both Chris and I, we're building generational engineering software. If we succeed, it's decades of engineers that are going to be using something like this, our software or something like that. And so ⁓ In general, ⁓ know, we say that San Francisco is the boom and bust town. And you can see the boom right now. And so it attracts all kinds of people, but the people with mission, drive to change something, to leave an impact, they're always around. So just need to find them.

Michael Finocchiaro

There's another question here from Christian Meisner. Probably more for you Chris. Does clear intent require an explicit functional specification of the geometric features?

Chris Barton

Hey, Christian. Good to hear from you, It's a complicated question because design intent can mean a lot of things. You can boil it down to form, fit, and function. You can boil it down to ⁓ what the actual tangible features are on a product. The reality is it is a little more complicated than that, and it requires context. So that is the direction that we are focusing our product, which is context. What is the entire picture that needs to be assumed here? And how you boil that down into every single call out that is in a discrete moment and its impact on the overall system? And so ⁓ if you have full context and full design intent, you don't have to do any of the work downstream. And that's the direction that we're very much headed.

Michael Finocchiaro

So do you guys, before I move into the last section, do you guys feel that we're, if I have to ask a question, we're not there. We haven't had the open AI moment in engineering yet, but are we close? Are we like six months away, two weeks away, two minutes away? mean, do think that we're on the cusp of the open AI moment where everything's just gonna change overnight and nothing is, because people are still using Creo, right? They're still using CATIA V5. ⁓ At what point will everything change? Do you think we're just there or it's still a ways off? Or maybe we're engineers, it'll be forever. We'll be stuck in the way things work. What do you guys think?

Chris Barton

We're getting close. ⁓ And there's not, you know, similar to chat GPT, there's not going to be one moment where it changes overnight because the broad applicability of it ⁓ isn't there, but there are going to be a few different moments where it happens. And, you know, I think in the next 12 months, we're going to start seeing those moments. We're going to see those iteration times starting to compress and people who experienced that moment will never want to design without that moment ever again. And that's going to be the moment that we know. And of course, it's going to happen in these different areas and the teams that are able to connect those dots and connect the tools that are focusing on different areas and bring it together. They're going to be the ones that start, you read headlines about, right? They're just moving faster. They're achieving more. They're getting, you know, more motivated people to come work there. And so that's what I'm excited for, right? Draftr is really focused on getting to that moment and expanding that moment and continuing to have a higher and higher impact. I know a lot of other really intelligent, really inspirational people that are focused on different areas where they're going to have a similar thing, but I'll think to your question, it's going to happen in the next year.

Michael Finocchiaro

All right, that's pretty aggressive. Janis, you agree with that or you're a little less aggressive on the timeline?

Janis Vavere

say I agree in spirit that there is a

Michael Finocchiaro

Yeah

Janis Vavere

change happening faster and faster and that change is also coming to our world. And some teams that we work with are moving incredibly fast and they are incredibly frustrated that the AI native tools are not available for every part of the workflow that they want. And so that's a good opportunity for us because they already see acceleration in other parts of the workflow and then they go to their existing vendors and they ask

Michael Finocchiaro

Mmm.

Janis Vavere

Where's the AI and is just a sidebar, side, it's on the roadmap. ⁓ But from the sort of, from the other perspective, it is hard to imagine a sort ⁓ of fully autonomous engineering sort of the way we see in some parts of software engineering.

Michael Finocchiaro

It's on the roadmap. It's on the roadmap.

Janis Vavere

And one of our team has recently said something very, very cool that I really liked was that he said that, you know, it's easy to create software, but the real magic and the real pain is maintaining software. And that's what the wipe coders will learn soon. And to a degree, creating even an actually internally coherent engineering. specification or drawing, think, is pretty hard and then probably even harder to maintain it and then actually make it into a real thing and then build the next version of it. So it's hard to... If there's an open AI model for engineering that's so good, then maybe there's no point in doing anything anymore. And so we should just go into services or something like that. But I believe the real engineering creativity but also the precision of engineering will not disappear anytime soon. That will be always needed ⁓ to just make sure that we actually build safe, ⁓ reliable and unique and better product. Just the speed of how fast we can do it will change dramatically in the next couple of years.

Michael Finocchiaro

Let's talk a minute then about your customers. So when I look at customers and I think about digital maturity, I ⁓ think about a range from like one to five where one is Excel email, ⁓ For almost everything ends up being Excel email when you're collaborating to a five, which would be fully autonomous, agentic ⁓ digital twins, right? I don't think anybody's at five. think very few people, few people at four, maybe not very many at three either. What's your experience? Your customers, are they closer to one closer to three?

Chris Barton

I mean, I'll go first. Our primary output is a 2D PDF. ⁓ I think teams, you know, are trying to move up the ladder, but it, it is very hard to have that be across your entire stack. And so the, the, the ubiquitous thing that you can use in this industry is a 2d piece of paper that you can print out and draw on. We are meeting people where they work today. So I would say that as a one, and I want to lift everybody out of one. ⁓ if we get every single person from one to two, man, we've made a huge impact. But.

Michael Finocchiaro

It's solid one, right?

Chris Barton

that's not going to be good enough. And so it's taking other aspects of that business and moving it to three, moving it to four. And there are some teams that are pushing us every day. I love that. But the reality is, is there are different aspects on part of these workflows and we just need to move into the digital era so that every aspect of this is captured digitally formatted and structured data.

Michael Finocchiaro

How about for you, Yannis?

Janis Vavere

Yeah, it's a good question overall. I like it a lot because we see teams that build complex products with a lot of software on it, where they see, I think they're moving from three to five in the software engineering part. ⁓ Or at least they see the potential and ambition to get to five. But then almost everything else they do is one.

Michael Finocchiaro

Mm-hmm.

Janis Vavere

And that creates this frustration there, like how is this possible that like we have this amazing workflow that we didn't have six months ago where we see such improvements in productivity and satisfaction and excitement and the speed. And then everything else is just sort of slow and boring and the same as it was 10 years ago. And those are our customers right now. ⁓ Teams that see, that have made the transition with agents in some parts of their workflow, but then they want to see the replicate that across everything. ⁓ But then, you know, they will start out competing the people that have not went to the agent first world and those will come and the laggards will die or will have to change. So it's all fine. It's exciting overall. And really the people that we spoke to, they are the most amazing people overall from the because they've adopted AI in some parts and they're excited about it and they're looking for more.

Michael Finocchiaro

So the other that leads to sort of my last question, which is like when, someone implements Drafter or traced out space is there, mean, you, you both are sort of lose to, but is there sort of a, an epiphany where the customer says, my God, you know, I, didn't I do this before? And why don't I fix some of these other problems in terms of digital maturity? Like, in other words, adopting a startup, which is solving one problem very well. ⁓ using AI stuff and using latest technology? Does that become a catalyst for digital transformation for the entire company? Is there sort of a realization like, my God, we actually could do things differently and thank God we were using Drafter or trace space so we wouldn't have seen this before? Good, Johannes.

Janis Vavere

I would say yes. For us, ⁓ some of the enterprises, this is their first useful AI implementation. And they are like, well, welcome to the AI world. Because many of the teams, the best thing that they have access to is a Microsoft Copilot that has been purchased through the central IT. And of course, they all use ⁓ Cursor and Chagypti and Claude on the side.

Michael Finocchiaro

⁓ god.

Janis Vavere

that they can talk about it, but it's not a company approved AI that they can use on daily basis and it's integrated in a safe way. ⁓ yes, so the excitement that people have when they don't have to do the repetitive mundane annoying work and that they can actually finally focus on sort of creativity and engineering again, that drives a lot of interest and a lot of expansion as well.

Chris Barton

Yeah, I I think we're similar. ⁓ Anytime a mechanical engineer makes a drawing with Drafter, they will never want to make it the existing way. And we can have that epiphany moment. And the typical thing that we hear from customers is, hey, it's great that you do drawings. Can you also do this? Can you help us with this? We get pulled kind of into the adjacency, into the places where that data goes. And, you know, that's exactly what I want to hear. Can you help us? solve these problems that exist that are related. And that's what we're focusing on right now.

Michael Finocchiaro

That's awesome. Well, I had a great time. I appreciate you guys joining the call. At this point, I also ask where we can see you. know that Janice, you're going to be with me at Threaded and at ACE down in Miami in two weeks. I'll be at CDFAM next week. So I actually meet PhysXX, you mentioned them earlier. I finally meet those guys. I haven't talked to them yet. ⁓ And other friends of mine too. Where can we see you guys in a couple of months? Where can people meet Chris and Janice other than Affirmation Miami. Go ahead, Chris. Where can we? Where are you honest?

Janis Vavere

Yeah, from my side, we'll do Incozy in Yokohama, in Japan. And we do almost a weekly webinar and we'll do a ⁓ dinner in Boston and a couple of events, one in El Segundo and one in San Francisco in April and May. So follow our ⁓ events page or our LinkedIn and we'll be there.

Michael Finocchiaro

nice. The Elsa Gunner, is that the one within top or is it a different one? That's a different one? Okay.

Janis Vavere

I just thought we'd do our own engineering dinners to get people together,

Michael Finocchiaro

Very nice. How about you, Chris?

Chris Barton

I'll be speaking at Kinetic in San Francisco the second week of May. Would love to see people there. ⁓ Happy to kind of dive in deeper on what we are. We're going to be going to a bunch of different conferences, but I would encourage people to find me on LinkedIn. We post a ton of free mechanical engineering knowledge and resources. And if you want to DM me, would love to hear either places that we can add more value or ways that we can connect.

Michael Finocchiaro

Nice. Awesome. And this podcast was sponsored by AWS. At one point when I this, there'll be sort of a URL you can click on to download the white paper to keep the sponsorship going. It's been awesome having you guys. I can't wait to meet you in person. The two of you, Janice, obviously before Chris. And thanks for being so inspiring. It's just awesome to see what you guys have accomplished. And it's always amazing to talk to founders. Thank you.

Janis Vavere

Thank you, Fino, I appreciate it. And thanks for time, Chris, as well. Very nice speaking with you.

Chris Barton

Thank you, Fina. Yeah, you too. Really appreciate it. Thanks for putting these on, Fino. It's always fun and inspiring talking, hearing you talk to ⁓ other founders in this space. So keep it up.

Michael Finocchiaro

That's very kind, thank you. All right, guys, we'll see you in the next episode in a couple of weeks after all this travel. Bye-bye.