Masters of Technology Happy Hour
Conversations with masters of technology, those who produce it or those who use it.
Masters of Technology Happy Hour
S1 Ep10: John McEleney's CAD Revolutions from Desktop to Cloud
When technological platforms shift, opportunities for disruption emerge. Few people understand this principle better than John McEleney, who has twice been at the center of CAD industry revolutions – first as SolidWorks CEO during the 2D-to-3D transition, and later as Onshape co-founder when cloud computing transformed design collaboration.
In this wide-ranging conversation, McEleney takes us back to the mid-1990s when he created the famous "plume chart" that visualized how 3D CAD would expand from specialized applications into the mainstream market. He reveals SolidWorks' strategic genius: rather than targeting committed Pro/ENGINEER users or complete CAD novices, they focused on the sweet spot – companies with "two seats of 3D knowledge and eight seats of 2D." This approach minimized both the resistance of entrenched users and the educational burden of complete beginners.
The discussion shifts to McEleney's early recognition of cloud computing's potential. After attending an AWS event around 2008, he became convinced that cloud technology would "totally change things completely," eventually leading him to co-found Onshape. Though the engineering community initially resisted cloud-based design, McEleney explains how security concerns have reversed – cloud systems can now instantly patch vulnerabilities across all users, while traditional installed software might take "years" to update everyone.
Looking toward the future, McEleney offers a nuanced view of AI's role in engineering design. Rather than AI systems that attempt to "design a car" from scratch, he envisions AI as a "river guide" helping engineers navigate complex design decisions by connecting unstructured data and revealing the ripple effects of changes. "I'm hugely optimistic about AI impacting and helping engineers," he explains, "not replacing them, but improving their reasoning and visibility."
Perhaps most fascinating is McEleney's insight into innovation itself. He shares how Onshape's revolutionary FeatureScript language emerged from a gut feeling about its potential power, even though the team couldn't fully articulate its applications at the time. This story highlights a crucial lesson: sometimes the most transformative innovations require faith in technology's possibilities before all use cases are clear.
Ready to explore how platform shifts create opportunities and how AI might transform engineering? Listen now and gain insights from one of CAD's most influential innovators.
Hello everyone and welcome to the Masters of Technology happy hour, where once a week, I have a drink with someone I meet in the course of business, but someone I'd like to get to know better as a person. Today, we have with us as a guest John McEleney. John, I'm going to read this bio of you, ChatGPT, so prepare yourself. This is the first time I've done this. I did it to myself. It was quite flattering. I got to say I don't know if yours is quite as flattering. Okay, here it goes.
Roopinder:John is a pioneering figure in the CAD industry, known for his leadership at SolidWorks and as co-founder of Onshape. He joined SolidWorks in its early days and rose to CEO in 2001, guiding the company through rapid growth and a successful acquisition by Dassault Systèmes. Under his tenure, SolidWorks became a dominant force in mechanical CAD. In 2012, co-founded Onshape, with fellow SolidWorks veterans, including John Hirschtick, who will be a guest on the show, launched the first full cloud platform. His work has reshaped how engineers design products, emphasizing collaboration, mobility and innovation. Mcelhinney remains an influential voice in the evolution of engineering software. I'd say most of that is accurate.
John:It was done with a lot of help along the way and there was a couple of one minor omission, which was a company I was part of called CloudSwitch, and the reason I think that's important is that it was really seizing on the opportunity of the transformation of the cloud, of helping to move applications, and along the way I met some really wonderful people that ended up joining on Shapen and in fact, part of my strategy of getting on board was having them help me find a talent like them and in the end I asked them to interview many of the people and they're still here today, john.
Roopinder:The premise of this show is Masters of Technology, which you certainly do qualify. It's also called Happy Hour and I always encourage people to have a drink with me. Although here's one right now, it doesn't have anything. It looks like you're still at work and you have work to do. I am. I have a drink, but it's sparkling. But sparkling water I won't have mine. This is just water. Actually, it's a decoy. One day we'll have to do this over drinks. So I was trying to remember. We've known each other a long time, I'd say 30 years, john. I think 95 is when I first met you guys, 30 years ago. Yeah, I remember what that moment was. Do you remember? I want to tell that story if you don't.
John:Yeah, I do. I tried to recruit you to come help join us at SolidWorks in the early days, remember that day? Yes, I do, yes, of course, and unfortunately that did not come to fruition, but we met along that journey and then, through the years, of course, I've stayed in touch, both personally and professionally 30-something years ago.
Roopinder:I've enjoyed all that collaboration, professionally and personally. It's been great. I want to say, though, that when you try to recruit me, I look back on that and I just wonder if I chose the right option. I chose to go with another CAD called IMSI makers of TurboCAD, and I did it for very practical reasons. It was right in my neighborhood, I didn't have to move right, but when I look back on it, it was not the first time I bet on the wrong horse.
John:We all have a shoulda, coulda, woulda, but I do think part of the reason it was interesting, besides your dynamic personality and the way you could add it to the team At that time. If I go back in the Wayback Machine, the market really was in its infancy from the CAD perspective relative to 3D, and a lion's share of the market, of course, was going to be a new entrance into the 3D world and you had come from that 2D world in your previous life. Your domain experience on the product side was going to be something that we thought would be extremely valuable really to influence our thinking relative to how to connect with those 2D users. They were going to make the move to 3D, so that was part of the motivation of getting you on board, if you remember that discussions I do, I think my role that I would have fulfilled at SolidWorks was more than amply taken by.
John:Correct me if I'm wrong. Was it Greg Jankowski who filled that role? I don't really remember, but it was around that same time. I had him out as a guest.
Roopinder:Actually I had him, recorded him yesterday and he told me how he actually went on to write that book. That's correct SolidWorks for AutoCAD users. I don't think I could have done that really I'd never written a book and that seems like a monumental effort he created.
John:SolidWorks for Dummies If you remember the four dummies series, which was really popular at that time, and SolidWorks for Dummies became a number one seller. It was number one for a little while in the Amazon store, but yeah, he was really influential in helping drive a lot of the educational, awareness and learning programs for people to move from 2D to 3D
John:See didn't need me. After all, you got somebody better.
John:Goes to show you need lots of great people, not necessarily just one the mythical lone entrepreneur sitting in the garage by themselves. It's exactly that a myth.
Roopinder:I don't expect you to remember this, but I remember in around it was in 95, I came on in Cadence. It was actually before that before that we were going to, you had offered me the position. That was when, okay, you took me aside. We were at a trade fair. Might have been NDES, maybe it was Autofact. You remember those Autofact or?
John:NDES. Maybe it was Autofact. You remember those Autofact, or NDES?
Roopinder:I do, yeah.
John:It's really true. Pre-internet, the only way you can find out information about different sort of products and offering services were trade shows. There's many other benefits of trade shows, but that was one of them.
Roopinder:I was working for Cadence magazine. I was their editor and you took me aside and sat me down and showed me. I wish I would have saved this. You sketched out the mainstream CAD graph, right oh?
John:Sure I can describe it to people, if you'd like, please do.
John:It was really simple. The horizontal axis was time and the vertical axis really was functionality, if you will, of the CAD system or the system in general, about complexity, simple to more complex, and the idea was that we called it the plume chart. It was almost like an exhaust plume. That was a parabola that was turned on its side and it was this expanding parabola. At the bottom, there was 2D, if you will, and at the top, in terms of required functionality to address more complex problems, were 3D systems and that at the upper end of the spectrum of that plume, if you will, was Dassault Systèmes and Siemens, which was really focused on very complex process type of companies planes, trains and automobiles, if you will and at the bottom was a lot of 2D that was relatively jigs, fixtures, if you will.
John:It was 2D drafting and then high-end 3D surface and solid modeling, and what we said in that plume chart was that it was going to be the emergence of this mainstream 3D solid modeling and it dovetails also with our go-to-market thoughts at the time, which was that you had many people that really wanted Pro/ENGINEER, which is now Creo, was really the first 3D CAD solution that metaphorically had a spreadsheet attached so you could make changes and things would update. Before that there was many explicit solid modelers, but if you wanted to make changes it was quite complex and costly to go ahead and do it. You had Pro Engineer. That was really crushing the market and as we built SolidWorks we really tried to target those users that had their noses up against the glass. They wanted what Pro? pro-engineer had, but it was too hard to learn, too hard to use and too expensive.
John:And as a result that was our raise in debt for SolidWorks. If I remember the chart correctly and sit down with you, the very bottom part of the problem was just starting. Small slice of it if you think about it like slicing bread was where we were saying the market was and that it was going to emerge into a very wide parabola where it would take market share from a lot of the market leaders as well as the 2D users. We deliberately tried to direct our marketing and sales team to focus on a user base that really was not 100% pro-engineer users and not 100% 2D users. The ideal customer profile at that time really was it was a 10-seat sort of potential account.
John:We didn't go after somebody that had 10 seats of pro-engineer. We didn't go after somebody that had 10 seats of AutoCAD. And the reason we didn't go after 10 seats of pro-engineer inside of there was it would take too much activation energy to get them off of pro-engineer. Conversely, educating all those users would have taken so much energy to move them to 3D. So we call that the modeling saturation index. So we wanted ones that had two seats of 3d knowledge and eight seats of 2D, because that was our raison d'être. It was too hard to learn, too hard to use, too expensive.
Roopinder:Yeah, it was. It was marketing genius, I gotta say. And I had a conversation with Scott Harris a couple of weeks ago. He was on the show and I said so was it the 80-20 rule? Did you want to have 80% of the functionality? It's like Autodesk said this For their success. They said, oh, we give you 80% of the functionality for 20% of the cost. I said, Scott, is that what you guys did when you first came out? And he almost laughed. He said, oh no, we don't think we had 80% of I'm paraphrasing 80% of Pro/E. We're lucky to get 30%, but it was the 80% that mattered.
John:Right, it was like it got the parts that mattered, you know the Pareto principle certainly applies in many things, but I think Scott and Jon and everybody else would agree that you're starting to build a product In some respects.
John:Our customer base grew with us and I don't just say that just to be kind. They actually in many cases, as they were learning about 3D, didn't even know that we didn't have or what they needed. And we had a development cycle at that time, early on. That was a six-month development cycle and when they realized they needed a new functionality, oftentimes we would basically come up with the next release and have that new functionality and with that cadence on a regular basis. Culturally and positioning-wise in the market, we created this sort of fait accompli mindset with many people that if they didn't have the functionality they'll have it soon, and that was kind of really a very deliberate effort on behalf of our messaging and marketing at the time to get it in that direction and it was successful in the sense that customers would identify things and we had a punch list of things we needed to do.
Roopinder:It was very clear. Yeah, I'm looking behind me. I saved my notebooks. But I doubt if I saved notebooks from that long ago. I should have saved that. I would have been good in the CAD museum. I would think that graph. You probably do it a hundred times.
John:That's one way to get in. The museum is drawing that diagram, but and people would try to get a little bit too specific on it but where are we in that slope? And it was really a conceptual model about how to think about the market and how it was evolving.
Roopinder:You were head of marketing at the time. You started in sales, correct.
John:And this might be a message for people as they think about their careers go sales correct. And this might be a message for people as they think about their careers go solve problems. Don't worry necessarily about your title, because I joined at some ambiguous title of business development and at the time we needed to build a partner program because we had our core application but we needed integrated applications. So I helped build the solution partner program and then ultimately hired a guy who's pretty well known in the industry. He's a great guy, Bob McGill. I think. He's since retired.
John:Once I got that started, we ended up hiring Bob to run that program full time and then I was getting our whole Asian distribution channel launched. So I went and recruited a lot of our distribution partners over there and got them up to speed and trained and then, once we did that, we ended up finding somebody that was living in Singapore to go ahead and run Asia Pacific and I think I moved into product marketing and different various roles in sales and from there multiple executive roles and then took over as CEO. The one.
Roopinder:Yeah, that's what the bio says. Bio got that right. Yeah, I'm looking for a magazine in which we actually feature SolidWorks.
John:Autodesk and AutoCAD.
Roopinder:Yeah, I think that's all Like 90% of what we covered, what we're trying to get into. We're trying to get into. There's one fateful issue, though I don't know if you can see it Mechanical Desktop. Sure, that's another wrong horse that I bet on there. So this was all about mechanical desktop and I want to bring this up when I talk to Hirschtick, because this was not quite favorable to SolidWorks, because I was coming at it from the AutoCAD side and my feeling was this to SolidWorks, because I was coming at it from the AutoCAD site and my feeling was this is going to be a lot easier for AutoCAD users to get used to mechanical desktop than it would be for SolidWorks. But anyhow, first I got an issue with that one.
John:Let's talk about something else. They had lots of natural advantages of having the AutoCAD users. The challenge, of course, sometimes is when there's a shift in the curve in terms of what's possible. Lots of times, the great opportunities happen for disruption when there's a platform shift and in the case of Windows and PCs, that created a whole new platform shift that allowed us to change how people interact with the system, how we sold to them, how we supported them, not unlike the platform shift with CloudWeb Mobile, with [], it allows you to really transform how you interact with customers, how they learn about you, how they learn about the software, how they learn about getting support, how they learn about their own problems of ways of addressing and the challenges they're facing.
John:New platforms are typically when new opportunities get created. It's happening right now in the AI world. It really is a new platform which is redefining what search means, which is why there's lots of questions about how search will look in the future. If you're trying to apply the same rules and just extend out when there's a new platform, you inevitably leave the opportunity for somebody to leapfrog you, that's true.
Roopinder:So you've been in the forefront of two revolutions of CAD platform changes with design software. Do you see a third one coming where the platform is AI-based rather than C++-based or cloud-based? Well, I think it could still be cloud-based, but yeah.
John:I think the biggest shift since the early days of SolidWorks, of course, was cloud web and mobile. That was the latest platform shift. I think we're in the infancy. In early days of SolidWorks, of course, was cloud web, mobile. That was the latest platform shift. I think we're in the infancy and early days of AI.
John:I think AI is a platform, but I don't think it's a compute platform in the sense that it changes how and what you do relative to discovery, connections, sharing intelligence, et cetera. But I think the compute still is going to be done on a cloud-based system. There's no AI system. That's a device that you're holding in your hand. It's using the infrastructure that's been built upon. So I don't think it's necessarily a platform from a compute standpoint, like we've done from mainframe to PCs and DOS to Windows to cloud.
John:Having said that, I do think it's going to have profound impacts, not just one, multiple impacts in terms of our industry, both in terms of finding information machine learning in terms of our industry, both in terms of finding information machine learning, really changing scenario analysis, and I have some very definitive views. I don't know if they're necessarily right, but I have some ideas of how it can really transform what people do, because people don't just sit behind a CAD system. That's not their job. Engineers are designing, building, solving problems related to products, and that could be getting that product built. There's all kinds of problems to getting it manufactured, to get it distributed, to get it supported, to reliability, to quality. There's all kinds of issues that they deal with, including suppliers and procurement, and it's a potpourri of problems, and I think that AI is going to have some profound implications on how really unstructured data gets connected to help them resolve problems easier, faster and quicker.
Roopinder:I totally agree. I think engineers do solve problems. That's what they really should be doing. I want to have a thought here. I want you to comment on it. It's a thought that I've expressed impatiently at CAD vendors, including PTC and Onshape and everyone else who even bothered listening to me. I don't think CAD ever helped us solve a problem. I think CAD helped us document or help me put that on geometric sense, but it never helped me solve the problem. I solved the problem in my head and I tried to document it on a CAD system. I would say the computer-aided design has been oversold. I don't even think it's accurate.
John:Let me say this about that. Let's be clear. I don't think it's been a magic black box that solves all the problems. It hasn't done that. In fact, it's created many problems for people. Sure, it's done a good job at documenting what a design is and we've automated that part of it. I think engineering will always be about taking inputs and synthesizing them ourselves to ultimately making the right decisions, or hopefully the right decisions, so that we're not going to abdicate that to a system. They all the systems are going to be support tools for engineers to make the decisions. I believe that. But I do think that cad system and design systems and analysis systems have solved many problems, everything from being able to will this fit, will this work, spatial orientation, can I, can I repair this thing if I build it? And I got to get a tool in there, I got to arrange a motion.
Roopinder:I would agree, but motion analysis yes. You can't solve basic CAD. It's just making shapes. It's making shapes.
John:But it's making shapes with intent. What kind of fit do I want to have in this thing? Do I want to press fit? Do I want to have it be threaded? Do joint? How do these things fit? How do these things go together? How do they get assembled? How do they get manufactured? There's plenty of problems it does solve. Still, you've got to directly solve the problem using the tools, and I think that AI is going to help us synthesize other tools, to help us identify possible solutions. But in the end, the solution is ultimately still coming from the engineer's head.
John:Usually, what happens with innovation? It tends to be very linear thinking. For ChatGPT and say Make, let me take me a new design of this. Is that the real problem that people have? I would argue there's some very unsexy but very important problems that sort of machine learning and real deep compute can help us with. For example, the amount of wasted heat loss due to the friction of the design process, of people trying to figure out where the hell things are and how many things are similar but slightly different. I would venture to say and it's a total wild ass guess on my part 10 to 15% of inventory is probably of different things that are similar but different and they've redesigned because they couldn't find it. Similar but different and they've redesigned and because they couldn't find it. And so you think about just the hidden cost, the carrying cost in their balance sheet of inventory.
John:What's a realistic problem that most engineers, most of them, have is a supplier getting out of that business and you need a new supplier, and that new supplier may not have the same exact part, it might be different materials. Then you got to try and figure out the implications to other parts of your system and we rely on the engineer that has a view of it to say where does this get impacted throughout my whole design and my whole system and my subsystems, whatever? And then once I go and figure out what might be an alternative solution it might be different materials, different parts or a redesign Then I've got to go and say what are the implications on cost, what are the implications on reliability, what are the implications on environmental impact? What are the implications on schedule?
John:The ripple effect of a change can be very significant and today we rely on the insight and knowledge of gray hair and experience to try and say where's the node in the network that I got to go next to see where this thing gets impacted, and I think a systemic view of taking what I would say is a ton of unstructured data could help us, you know, be better at our jobs and be more efficient at our jobs, to be able to understand these are the areas you need to go look and then be able to say if I'm going to use this material. What other things do I need to be concerned about and have things propagate that might be first or second or third level problems that don't require us to be so deliberate to try and think through all the implications.
John:A river guide help us navigate the rapids, so I think that would be huge in terms of the engineering, art, if you will.
Roopinder:I want to get back to AI and CAD, because I seem to detect that you think it might be maybe somewhat useful, but not totally.
John:I think it's going to be useful. It's not going to be somewhat. It will be useful and I think we'll have plenty of false starts, but it will be useful in the long run.
Roopinder:Okay, I'm reminded of years ago. Two years ago, it seems like forever ago, but it was only two years ago. You might have seen this because the picture went viral two years ago. You might have seen this because the picture went viral. It was a Belgian advertising company and they had put a sign on the side of a building. It said a real big sign, and two sides of the building it said "hey, chat gpt, finish this building. And it was a joke because they were obviously advertising. They were advertising for construction workers, but it was a joke because they were obviously advertising. They were advertising for construction workers, but it was a joke because AI wasn't going to replace those people. Chatgpt will never do the human's jobs in that regard. And since then, though, I think some CAD companies I won't say which one, I might as well, because I've written about it Autodesk showed at AU essentially make me a car.
Roopinder:There was actually a software that they had run out and bought and it actually created a car. It was really ridiculous. It wasn't a car at all, it was like the shape of a car, and the most you could do with it would be aerodynamic simulation perhaps of that shape, but it was skin deep. There was nothing to it, any internal stuff, and maybe you could explore concepts, do some design exploration, as you mentioned. It will never make a shape for you, but wouldn't you say that generative design that uses topology optimization, which is in Onshape?
John:You say that's an attempt to make it.
Roopinder:I don't think it's Onshape, but it's in Creo. It's in Onshape too, okay.
John:Say that's an attempt to make a design. I don't know if I said that it's not going to create geometric design. I think there's going to be plenty of things that will make geometric design. I think you're going to find ways we're going to have the advantages of configuration. Say, I make ramps or anything that might have structural elements. You might say make it 26 feet long versus 15 feet, and how many support structures might I need underneath it, and I would automatically figure things out. I think there's gonna be plenty of things that will be automated. That will be great. But this idea that just says just make me this, make me an oven, there'll be some things that we can refine using ChatGPT, like interfaces, might prove interesting and anything else would happen in big innovation fields. When you think of walking out of your building in wide open terrain, there's going to be some things that succeed and some things that don't, and if we knew only the things that were going to succeed, we'd only do those.
Roopinder:But we don't.
John:We got to iterate and test the idea that people may try some creative things and they fail miserably. I don't think we should necessarily throw stones at it. I think we've got to look at it and say they're trying to figure out some things. That's interesting in and of itself, right In terms of generative design, for sure it's valuable the idea of taking generative and saying let's do it, but do it with an eye towards optimizing this, but with an eye towards prismatic or normal subtractive methods. We're optimizing the objective function of minimizing the weight within the constraint of making it something that's subtracting material or adding material. Then you're doing organic shape. So, yes, it can create things, and I think we're still nibbling around the edges. I think our industry finally, for heaven's sake, is finally coming around to not being afraid of the cloud. It's a very conservative industry.
John:That took forever huh, A lot longer than everybody thought. But now that flywheel's in motion big time and that's a good thing, but it's a conservative industry. So on the AI side, if you believe that technology does get adopted a little faster each time with a new platform, maybe it won't take 10 years for AI, Maybe it'll only take five. But things will come out and there'll be a lot of point solutions. There'll be agents that help people do things and it will evolve in an interesting way. It won't necessarily be completely predictable. I'm hugely optimistic about the idea of it impacting and helping engineers. I don't think it replaces them. I think it helps to improve their reasoning, improve their visibility.
Roopinder:I remember after you left SolidWorks and I don't think you'd started Onshape yet, but I was covering SolidWorks and Jeff Ray was the CEO, and he made the mistake of saying that SolidWorks was going to be cloud-based and I don't know if he used these exact words, but he was taken to say that SolidWorks was going to go away. He meant SolidWorks as we know it. He used these exact words, but he was taken to say that SOLIDWORKS was going to go away. He meant SolidWorks as we know. It was going to go away and it was going to be replaced by cloud-based SolidWorks anyhow. That caused a huge furor among SOLIDWORKS users because people didn't trust the cloud whatsoever at that point. Is it taken since then, a couple of years back, where the cloud became known for its advantages? ?
John:I left in 2007. Okay, and I took about a year off and my journey to CloudSwitch was I had gone to an early AWS event in Boston and Werner Vogels was there, the CTO of Amazon, and there were maybe 50 people in the room and I remember, coming home, my wife said what did you think? I said I'm not sure I understood everything I saw, but if even half of it's right, this is going to totally change things completely. And I said so I'm going to spend all my time only focused on this and that's what I did. It was proclaiming cloud. That early Cloud was early. Let's be really clear. But I think the idea of Jeff leaning in with a vision that sort of says, hey, things are going to evolve and things are going to change, that's his job as a CEO to look forward to future impacts, of what it means to the core business and how it can help customers.
Roopinder:Oh, that's true. I just don't know if I should have announced it to the population. Yeah, I wasn't there and I don't know the situation.
John:And I'm not trying to just shy away from it. I just don't know all the details.
John:But the idea of building something internally which I know people were doing and trying to make it cloud-based and platform agnostic and take advantage of many of the things that we can experience today clearly the right way. But yet look at our industry. It's a conservative industry. The irony is everybody looks at their laptop and thinks it's going to be the most secure device and in probably another five years. We said this a while ago. If you were going to look back at it and say, how did we ever think we're going to make those endpoints secure?
John:Let me just give you a simple example. Today with Onshape, if there's right now, if there's a system requirement to do a patch for any people that are doing traditional install-based software and they need to do a patch because there's a problem, how do they get everybody to go and install that patch and update? It'll take years, especially so many people don't trust necessarily the current release of a software. There'll be two releases back, so it'll take them years to move people forward for something that is in their own interest, perhaps a security patch. Compare that with Onshape and I'm using Onshape as there's no installed software. That is pure cloud-based, native cloud.
John:If there's something that comes out tomorrow that somebody says there's an exploit and you got to be aware of it, we can literally take everybody off the system, point them to the new version that has that fix in it and instantaneously everybody's up on the latest version. What does that mean? It means it's a hell of a lot more secure because all it takes is one one device, if you will connected to a network that has that security flaw, and everybody's got that security flaw. Look at Stuxnet, iran right now with the nuclear, with the centrifuges I don't think it's ever actually been admitted, but that's when Israel had done with the Stuxnet virus. One laptop connected to the internet and it spun all their centrifuges up to 11 instead of 10. They self-destructed. So, yeah, having the ability to respond is really an advantage, and so people are going to look back at their desktop solutions in the future and say how did we ever think we were going to keep those endpoints secure?
Roopinder:John, I want to put this in perspective of time, just because people may be listening to it later, but you remarked Iranian-Israeli conflict and just now, just yesterday, they just stopped talking about the US strike against Iran's Fordow facility. That's where the centrifuges were if they hadn't moved them, but yeah, so it's a great example. For a long time, I was on that boat with their users that I don't know trust the cloud and I made the mistake of saying I trust things in my desktop more than I trust them on some server somewhere. I think I made a mistake of saying that to Hirschtick and he said oh yeah, what makes you think your desk, your office, is so secure? I could break into your office faster than I could break into a data center? And I said oh yep, you're right Once again.
Roopinder:Maybe another time I've been on the wrong horse. I've got a history of this. Speaking of your history, John, you made sure SolidWorks was a success. You went on to another platform and made that a success. That was a platform change for the cloud. You think there's going to be a third one here where there's going to be like a totally AI-driven cat. I'll put you on the spot.
John:I can tell you I will not be developing it. Yeah. I think, AI is going to impact it. Like I said, I don't know that there's going to be pure AI CAD-based type systems, but I think it's going to have a huge impact in our industry for sure.
Roopinder:I've said this. We should have a debate really about this. I want to debate a lot of people about this because, like I said, I'm impatiently wanting AI in CAD because I don't know how much longer can I do this, but I think the time is right. I don't want to make the mistake of saying, make me a car right or make me a plane. I don't think that's a plane I would ever fly on, first of all, and I just trust engineers for what they're good at. But I want to be freed from the grudge work of CAD. I want to be free of having to learn CAD.
Roopinder:I've learned a lot of CAD systems more than one handful of CAD systems, I've got to say each time I've got to spend a lot of time learning the CAD system. If not, design me a car, I want to say, hey, design me my bike as an example. I've done this a lot. Make me a bike, but tell me where I want to put the joints for the optimal design that I want it to do. I don't want it to start with blobs. I don't want it to have generative design type stuff where it starts making a blob because it'll never get to a round tube.
Roopinder:It just won't get there. I want it to use round tubes because I know as an engineer a round tube is really good in compression and tension and can't be beat in torsion. Engineers we know this shapes that are optimal for their use. Sure, I don't want it to start with blobs. So how about a CAD system, John, that says, hey, I know Roopinder loves bikes, I know he loves round tubes. Know, Roopinder loves bikes, I know he loves round tubes. I'll just when he makes a line. Why don't we?
John:have that, make me a tube structure, or there's ways you can automate systems today. You can do that inside of on shape today to do that, that instead of making a cross section a square, make it a tube and whatever else. But the idea I think you just pointed to a great use case optimize where those connection points are, whatever, whatever it might be. But like anything else for Roopinder, I think there's certain enabling capabilities that happen with some technology, in this case AI. But I think the way to start is not necessarily where do we go apply this? I would spend the time on what are some of the problems you're trying to solve, what's the problem domain that people have, and then how can these tools help people with solving those problems? And I think you have a much greater chance of adding value back to the customer and to the person than if you just there's some core platform capabilities and many of those, many of those things are what people are thinking about.
John:But you don't want necessarily technology trying to look for the problem. You want to understand the problem, technology trying to look for the problem. You want to understand the problem and where you can apply the technology. If you start there and I think in our industry we're still looking for a killer use case of where AI can help and I think the VR AR world suffered a little bit from that which is this amazing technology that everybody in their gut says, geez, this really should be able to help many parts of the product design, product maintenance experience, and there still is yet to be the kind of killer use case that sort of drives and fulfills and pulls the demand, and I think that's going to happen.
John:Here too, you're going to have lots of one-offs, but there will be some use cases that are killer, that people say ah and once people have that, it'll help create manifest more use cases and I think that that's how network effects start to think over exactly when I remember this.
Roopinder:I think AR/VR had high hopes for that too. I remember one show, and there was Jim Heppelmann. He said Roopinder , I gotta show you something. Takes me over to a booth and he's got these VR goggles right and he says and sure enough, you could see the part that I was going to help. I think it was some factory or some motorcycle. I could lift this part off a motorcycle. It was giving me instructions as I was gonna do that in my VR headset and I could see the part. I could see the instructions and I thought, oh, this is going to take over the world. It hasn't. It didn't. Apple is the last big company to introduce a VR headset, their Vision.
John:They pretty much pulled the production of it.
John:Just haven't found necessarily killer use case, but there are some verticalized use cases and there's a great example of Volvo truck Volvo truck. Volvo truck has famously in some of their engines this a million permutations of different things that are going to be assembled and the combination of VR on the instruction side but IoT in terms of figuring out exact configurations to ensure that configuration was assembled properly. That's an interesting use case, but there's just been no generalized broad kind of killer app that drives the demand and sometimes we can bang our head against the wall for a long period of time and then they appear and sometimes they don't. That's just market realities. I think on the AI side you're going to see that emerge with agents and with specific use cases that then will start to be generalized. Maybe somebody will go out and create a whole AI-based CAD system, but I just feel like there's a high bar of functionality required for a CAD system. As we experienced with the creation of Onshape. It was a what was considered a high jump really was a pole vault in terms of functionality. I always liked your analogies. We knew it was not going to be a speed bump or even a hurdle. We knew it was going to be a high jump. It turns out it was really more of a pole vault and and thank God, we created and this is an example where you have sometimes you have to have faith in technology.
John:We started Onshape. We had built this prototype and then brilliant young man, Ilya Berman, created this idea of feature scripting. He says, hey, we should look at this, guys, take a look at this. And we ended up at the time deciding to go and we all knew deep in our guts that feature script could have huge potential of in the future. We just didn't know exactly how, but we knew it was like really profound, but it was very hard to articulate what it could do and we were starting something and we're going to move forward with that prototype. And we took a pause and said we had to develop the language, which was much like JavaScript, if you will, but it was going to be an embedded language and it's a day zero decision. You can't retrofit it.
John:And so we put a pause on that prototype and we went and developed FeatureScript and actually said let's re-implement the prototype using FeatureScript. And then we saw how we could automate it, which would allow for customization and the assurity to be able to move things, and the profound example I'll give you is configurations, how FeatureS cript works, and we didn't know this at the time, we just knew in our guts. And this is sometime where, as a team, you're paying, you're leveraging the experience you have for insight and intuition. And an example of where FeatureS cript is great is if you want to build configurations. In many respects, when you're using Onshape, what you're actually doing is you're graphically interacting with the system, but behind the scenes you're actually creating feature script code. Okay, and feature scripts is an interpretive language and the interpreter is actually Onshape. It's executing that code that you're doing.
John:How is that profound? If you're doing something, let's say that you're doing. How is that profound? If you're doing something? Let's say that you wanted to have something that had multiple configurations. I want wheels that are two inches in diameter, four inches, eight inches, and I want to allow the customer to configure that with two inches, four inches or eight inches, but let's say that the shaft that's going to hold that there. I want to give them the ability to have an infinite number of permutations. I want it to be able to be 2.99 inches, 3.15 inches, 3.175 inches, 3.08 inches, so you want it to basically be infinite With most systems today you have to other than non-shape, you have to pre-calculate your configuration.
John:So you can only say today you have to other than non-shape, you have to pre-calculate your configuration. So you can only say two inches, four inches and six inch wheels and four inches, eight inches and 12 inch shaft, and you pre-calculate those number of permutations and essentially you're picking out row three, column six, and that's your defined, pre-calculated configuration. You can imagine when you get an N by M matrix, it starts becoming unwieldy. When it's thousands of permutations, just select something in a configuration that is literally defined by them, not a preset configuration. You would have an infinite number. It would be impossible to manage it. But feature script, it's only one configuration because it's what you're specifying and the interpretive language is creating it real time.
Roopinder:It's hugely powerful, which?
John:Means the maintainability of it, the maintainability of the system from release to release. It happens every three weeks. It's just not a problem because it's inherently the code that we use to build a launch.
Roopinder:That I found to be the most unusual thing. It's actually the code that the program is written in, that the other features are actually scripted in.
John:And now the user you're giving the user the keys to the castle. In effect, that code is actually yeah, and and the meta point here is not about feature script per se, but it's about the idea that early on you sometimes don't necessarily know exactly how it's going to play out, what it's going to be, but you have enough insight or intuition to say we know this can be extremely powerful and we had conviction to go and take the pain actually of a several month delay to go and test it before we went, because we knew it was a day zero decision.
John:And I think those types of things are going to happen as we look at AI getting implemented alongside CAT and impacting how people work. It's exciting.
Roopinder:John, this has been great. I would keep you forever, but maybe you would just promise me you'd be back on the show some more. Maybe next year we can discuss what actually did happen.
John:Put it on the calendar.
John:Sounds like a great thing Before I let you go though you still seem to be to help out, whether it's technologically licensing agreements for the team or helping with some large accounts, or to go spend time mentoring the sales team. I got rid of my Irish Catholic guilt by having them reduce my pay. That allows me to not feel guilty when I'm not in here. I'm involved in some boards, including a public board, and I'm still pretty active physically and traveling, and and so it's a great balance of staying connected, because it's an industry I love and I really want to see this, see Onshape be continuing its success and accelerating its success, and because I think it's really a platform that that's going to have huge implications long term.
Roopinder:I can see that you really still are very much a champion of it. You're going to make sure it does. I hope it does well too. I got to say I do worry about it a little bit, since Hatwoman left Not just Onshape but all told at PTC. We don't need to get into that.
John:I will say this Neil and the team.
John:They're doing some of the most important work they can do, which is they get outside of the walls here and they get in front of customers and there's a solid team there and solid leadership, and Neil continues to recruit great people, and so I don't share that concern. I actually I think Jim was a great leader and did some great things. I think Neil's a different leader, but he's equally impressive and customer focused and he's building the team around him to continue the growth and success of the company. I don't share that fear. I'm actually excited about things.
Roopinder:Fair enough. Let me ask you I hope this gives you more time to spend outside of work and enjoy life. Are you able to?
John:Yeah, my wife just had a significant milestone birthday, so we went with the kids and did some traveling and it was wonderful. Then they came back and we went to paris and we happened to be there with the french open and saw the historic women's final with coco and alcarez, with the five and a half hour men's final, which was just incredible. It was awesome and as usual, the french put on an amazing show and a wonderful city and wonderful memories from there.
Roopinder:So you always had really good vacation stories. I remember you going to whitewater rapids in chile. You got to stay young.
John:My college roommate started something we call the adventure club and we yeah, we've done. Back when we all turned 40, we climbed mount kilimanjaro, went on a safari, we went down to patagonia and chile, did class five whitewater rafting, went to alaska and flew out on a glacier and did ice climbing, and then it took us in the tundra and we hiked for six or seven days and riding horseback in outer Mongolia. So every two years we try to take experiential wild adventures and it gives you young.
Roopinder:Does tennis replaced your love of golf?
John:No, my wife is a tennis person. It was her birthday. If it was up to me, we would have gone to the masters or something.
Roopinder:All right, John, it's been wonderful.