Mark Burhop on AI For Engineers

CAD should be the easiest tool in the room, yet it still feels like stepping into a cockpit packed with controls you have to relearn every time. We sit down with Mark, a longtime developer and former Siemens leader, to talk about why AI is racing ahead in some areas while product design and manufacturing still feel stuck and why the biggest blocker is not geometry, it is the interface.

We break down what large language models are genuinely great at today: procedural work, fast research, documentation, and especially coding. Mark explains why “vibe coding” and multi agent workflows are changing software development, pushing value toward architecture, domain experience, and good judgment. We also get real about the messy side: non deterministic outputs, security risks, and the stress of managing agents that never stop running.

Then we move to the physical world. Robots look impressive on stage, but on the factory floor speed, sensing, touch, and reliability matter more than demos. We talk about physical AI, why humanoid robots are both tempting and often impractical, and where near term wins actually live, like AI that helps operators troubleshoot CNC errors instantly or reduces time wasted searching documentation.

Finally, we connect the dots back to CAD, CAM, and CAE: automated drawings, design checking, natural language design, generative design exploration, and the emerging standards like MCP that aim to connect AI to engineering tools. If you care about the future of engineering software, AI for manufacturing, and the next generation of CAD interfaces, this conversation will sharpen your thinking. 

Transcript

Welcome And AI Time

Roopinder 0:07

Hello and welcome to FoDES, the Future of Design and Engineering Software podcast. My name is Roopinder Tara. On the show, we will have guests that will discuss tools and technology that engineers will find interesting and useful.

Mark Burhop 0:21

Hey Ropinder. How are you doing?

Roopinder 0:24

Good to see you again.

Mark Burhop 0:25

Yeah, you too. Been a long time.

Roopinder 0:28

It has been a long time. But uh you measure that in the internet or uh AI time. It's been centuries, right?

Mark Burhop 0:35

So yeah. Yeah, well, I don't even know how AI time works anymore. It's just it's it's seconds that are getting slower or faster and faster. The time between seconds is faster and faster. Something like that.

Roopinder 0:47

We've known each other for a long time. Like I want to say decades. I think we met first at uh probably at Huntsville, right?

Mark Burhop 0:56

When you were with it it could have been. I I uh let me think. So I started way back in at SDRC before that. Uh okay. So so I got part of that when the SDRC and EDS uh or UGS I should say kind of combined.

Roopinder 1:10

You're part of the ideas, ideas group.

Mark Burhop 1:13

Yeah, yeah, but it was it was funny because I didn't really think EDS was quite the company I wanted. So uh because it it had some uh bad had kind of a bad reputation. And uh at that time I I ended up leaving and I left for about 18 months, and you know, the whole

From SDRC To Siemens Research

Mark Burhop 1:30

world changed then. EDS spun it off to private investors, and then Siemens bought them to become uh basically what is Siemens DISW right now, digital industry software.

Roopinder 1:41

Right.

Mark Burhop 1:41

So uh yeah, I I get around a little bit.

Roopinder 1:44

You do, and then and then what? You then okay, then we went to Siemens and then ended up.

Mark Burhop 1:49

Then I went to Siemens. I'd always been a developer up to then, but I really like finite element analysis, and that's what I was doing with SDRC too in in Huntsville, Alabama, um, is where they have the solid edge development uh down there. So I I basically started the the simulation program there, did some did some initially some development with it, and and then later became the the simulation product manager. And you know that that was great. It was a really, really nice, nice thing to do to birth a new product. But after that, yeah, I it it slows down and I always like research. So it's about that time I switched over to uh what's what at the time was Siemens technology or corporate research for all of Siemens. So that that's really interesting because then it connects you to all the Siemens companies, the companies making trains and the companies making, you know, at least back then, you know, energy equipment, uh large gas turbines, just about everything, uh medical equipment, whatever.

Roopinder 2:46

Yeah. Your background, uh your educational background as a as a mechanical engineer, correct? You were trained as a mechanical engineer.

Mark Burhop 2:52

Yeah, so I'm I it's funny. I've always switched between the two. Uh I actually started as a math major, which was kind of funny because if you were a math major, you had to take a language. You know, I tried to take a language and I was I was really, really bad at it. I worked at it really hard. And uh so of course, the first soft, you know, okay, well, you know, engineering doesn't require a language. So the first thing I do is get a job at a French company and have to learn French. So uh it doesn't escape you. So I I did learn if you if you work long enough on it, you can eventually, you know, learn the language. Still have no skill for it, but but at least I you know got something out of it.

Roopinder 3:31

So you ended up learning languages of a different sort, right? The languages

Where LLMs Shine And Slip

Roopinder 3:35

of uh design and and CAD programs, right? Each of those, you know, this is very this ties into what I want to talk about uh with you is AI and LLMs, and LLMs are all are language models, right? Large language models, right? So they uh my beef with uh CAD companies is you know they're you know large language models would have no difficulty at all learning how to use CAD because their language, CAD language is so limited, a few hundred words, and you've got CAD, right? So given that they're given their abilities, profound the great abilities in language, it should be no problem for them to master to make it.

Mark Burhop 4:18

Um you would think so, but there's some limits. So there's you know, there's different types of AI, and and obviously the language large language models is what everybody's talking about right now.

Roopinder 4:29

Right.

Mark Burhop 4:29

Um, and they're great at so many things, writing code, anything procedural, and anything that you can do and then validate at the end. Uh it works well. They don't have a, you know, it's a little bit like uh, you know how some people are are not good at math, but very good at other things. Yeah, like a lawyer does not have to be good at math, they you can do it with other skills. Engineers, you have to have math, but maybe you can get away with some other uh English skills or other things there. Um, I I kind of see LLMs as is something like that. They're very good at language, they're very good at explaining things. If you can put things in terms of language, they're good. Um, the one of the challenges though is mathematics. They, you know, you have so many LLMs that you know, if you're trying to fit two parts together and and one is is 9.2 and the other is 9.11, half the time the LLMs will say, oh well, 9.11 is bigger than 9.2 and and get it wrong. So those kinds of mistakes are are are still common. It it really has a hard time relating between numbers, for example. Numbers don't always get tokenized like words do. So if you've got a long number, maybe pi is tokenized, but other numbers that are just these long numbers, they're broken down into little tokens, and you have to do a lot of things to figure them out. So there's still some issues there that I think companies need to look, you know, especially the the large, large companies uh working in geometry or those areas, need to do a little more work on in those areas still.

Roopinder 6:06

Yeah, sounds like we need instead of a large language model, we need a large math model, a large geometry model.

Mark Burhop 6:12

Yeah, and and it's it's kind of there. So I I I watch a lot of podcasts, and and one of the things that's going on now, symbolic math, it's great at is able to go through proofs and and come up with new things. Um, but just kind of the feel for numbers, you know, what's big, what's small, what's really big. Those that, at least to my mind, that's kind of the one of the challenges there.

Roopinder 6:33

Um is that why isn't that why Jensen? Well, you know, Jensen.

Mark Burhop 6:40

Oh gosh, yeah. Not personally, obviously, but uh yeah, I I see I I probably gobble up everything he does or or what so

Physical AI And The Robot Reality

Mark Burhop 6:50

yeah, so does the yeah, the rest of the world.

Roopinder 6:53

He's like the god of AI right now because all this uh hardware. Uh he said physical AI was going to solve this problem, right? This is physical AI is what we need because it understands physics, it may not understand numbers per se, right? But understands physics or it should. I I I haven't seen a lot of uptake at physical AI. I think he intended for it to be used in robots, so robots could have a sense with a touch, you know, how robots can really bad at touch.

Mark Burhop 7:24

Uh yeah, yeah. I mean, that was one of my research projects that we had was, you know, you can, you know, for 3D printing, you can put filaments in there that conduct electricity. So now all of a sudden, you know, you got a robot. Usually a robot's out there and it's just like kind of blindly reaching around for stuff. If you can add touch to your your robot, and you know, if you bend it, you can pick up uh, you know, how hard you're squeezing things together. If you have uh fields, you know, just from electricity running through, you can detect how close you are to metal and things like that. So I think there's a lot of lot of potential there. I I'm personally I I love I love the robots. I love what every everything that they're moving that way. I don't feel like it's coming as soon as as a lot of people think.

Roopinder 8:09

Yeah.

Mark Burhop 8:10

And part of that is it maybe more feeling that recently when I got from Hanover back from Hanover, Hanover Mese um a few weeks ago. Um, I was really excited to see all the robots there, but so many of those robots are, you know, they're not doing anything useful. They're barely able to stand and walk sometimes.

Roopinder 8:28

Oh, it's it's it's painful to watch.

Mark Burhop 8:31

Yeah, so and and so you're saying, oh well, we're gonna have robots next year. And it's like, where are they? Because they're not at this conference for manufacturing.

Roopinder 8:38

No, they weren't at I went to CES uh second year. This time I was a judge for their innovation uh innovation awards, and uh I saw the same thing. I was robots, robot after robot after robot, and they were doing so many things so simple and so slow. It I think it just gave robots a bad name. They were supposed to be impressive, and sure enough, the crowds gathered, but watching them, a couple of them are folding laundry, uh laundry uh shirts. And yes, it's amazing they can do that. It's also everybody I heard somebody say, Well, that person worked in the laundry, that robot would be if that robot worked in that factory, that robot would be fired in the first hour because it's pretty damn slow, right? So it doesn't help their image, I don't think, to parade these robots out.

Mark Burhop 9:30

Yeah, and that it's funny, that was one of my comments too was uh on the Hanover Mess. Um the I I felt like they took the robots from CES, which you know, CES is more let's show cool things, and and people that don't do manufacturing can look at the robots and say, oh, that was so amazing. But bringing robots into a manufacturing conference, you better have your A game there. People engineers and people from factories know exactly what people need to do on the factory, and they they that you know, seeing a robot stumble around trying to carry one part like six feet to a table is just uh uh it's it's terribly disappointing sometimes.

Roopinder 10:08

One company had for the second time in a row, second year in a row, they had a robot carrying um uh bin uh from where the parts came off a line to another place, another part to another line. And again, it was moving so slowly. Again, for the second year, it didn't hadn't picked up any speed in the second year. It was still wowing the crowds because these guys haven't seen a robot yet. And they may be just happy that, oh, I could go home and tell people I saw a robot at a factory. But to people like us, like it's not really practical yet. It's there is very limited use in uh humanoid robots. I think they're I don't know, it could easily be the uh the first bubble bursting because humanoids are shouldn't even be a thing. I like it like if I designed a human, I don't think it would it would be like us, right? So consequently, if I wanted to have somebody somebody weld things on an assembly line, I would have an industrial robot with some good at welding. Sure. I can have a humanoid robot, right? Because there's so many things on a humanoid robot that don't lend themselves to specific tasks like welding or right? Are we going too far? Like trying to pretend that a human form is the best thing for a robot.

Mark Burhop 11:24

Yeah, well, I I think there's there's a couple reasons for that, though, too. So one is you, you know, so many uh the the environment, the factories, the moving around, it's all built for humans, right? So the door handle is always at a certain level, the door size is always for a humanoid robot, the stations that you know, if you're gonna replace a human doing something with a robot, that station was probably set up for a humanoid robot, too. Um, so I so I I understand the need to do that, the humanoid robots, and also the need not to. There's lots of cases that doesn't matter, right? Picking up pallets and carrying them around. These little floor robots are 10 times better than that. Um, so yeah, I think there's it just depends on what you're doing. I mean, the the goal was always you want factories that you can reconfigure. So if you have something configured for just doing one job and the new factory doesn't require that anymore, you know, it's it'd be a shame to lose that investment, right? You'd want to be able to move it to uh to the next task. Um but the is the cost. I don't I don't know that the advantage of cost is there yet, is is the thing. So I I you know, so when it comes to AI, I I I think the interesting things are probably more on the computer side right now. Um the real advantages of

Vibe Coding And Multi Agent Development

Mark Burhop 12:40

things you can do are there, and then and I'm sure there's some good give me an example of that. Coding. If you're gonna write code, and I could talk about this all day because I've been deep into AI-based development, you know, even if you're even if you're a mechanical engineer that took a class of Fortran, you know, maybe going way back, and maybe you write a script every once in a while to maybe automate your simulation or automate something on your Windows machine or Linux, vibe coding is just awesome. You can just go to your your AI, ask what you tell what you want for your script, and it will create one in a fraction. And you know, if you don't like it, you just talk to it a little bit and you're fine. So I think that's uh uh a huge things, uh huge thing that especially engineers that that know what programming is and maybe use it occasionally, I think they're gonna be super happy with with what's going on here. Um now it's not that I I don't think the IT people are gonna be happy about that. All of a sudden the engineers are writing code that's very useful for them, but probably have security holes all over the place and all kinds of other challenges too.

Roopinder 13:47

And yeah, you know. This also seems like if you're if okay, I'm not a coder and I'd be very bad if I tried to code things right now. My last core computer course was basic and Fortran, and that was decades ago. I would just love this AI that can code for me. I would love it because it can do it can write near 99% good code, right? And much faster than I could take a course over again and then do it. I could write Python code, I could write script for CAD. It's wonderful to me. I'm easily wound by this, right? Yeah, of course, a developer is not because a developer, a real a senior developer, a high-level coder would see, like, oh, that's not really elegant code, that's not very good.

Mark Burhop 14:34

No, but here's here's the problem. And if if you said this nine months ago, I'd say, yeah, you're you're you're right. If you had said this in December, uh, and I think that's where there was a big change, is it's all changing. And I think that's something people are having a hard time with. AI is changing so quickly. And what I mean by that, if I saw someone writing code today for professional software, uh I'd I'd be very worried about them right now because there's no reason to do that. And they're and and you could say, oh, well, it's quality, but the code I'm writing with AI right now has a far higher quality, it's far better architected um than my own code would have been. And I've been doing this for many years, right? And you can write a code. Yeah, because I'm you know, I it much of my life has been creating production code and creating good architecture so that it works there.

Roopinder 15:29

Oh, so it could impress it could impress you even that you're a very good thing.

Mark Burhop 15:33

Yeah, and I it I think that would be true of most most developers that I think, you know, I I had this this, you know, I was actually on LinkedIn a little while ago, and there's there's some discussion within more on the manufacturing side about, well, you know, maybe we don't have to write code anymore. But if you go to software development side, if you go to a senior developer at any any company, especially the fan companies, but the other ones too, uh, I I would really doubt there's a senior developer still writing code there unless it's some really niche application. I I would spec that all senior developers are using AI now to create their code. And it's even gotten to another level where they're not writing the code, they're sending out multiple agents to write the code in these these big sandboxes. So if you're, you know, I have a lot of friends that you know NVIDIA or uh you know other other high-tech companies, and you know, that's what they do. They have tons of tokens. They say, okay, we'll write this code here, and you maybe we'll even have three of them write the same code, we'll pick the best code of the three. Um you have a lot of of new possibilities now that are coming to mind. So if you're uh if you're an independent coder not using AI or just using it for code complete or merry minor things, you're not gonna be able to keep up. It's gonna be a real hard thing, and it's getting worse every year for that. If not every year, every day, every month, every week, it's getting better.

Roopinder 16:58

Okay, it's getting worse for them.

Mark Burhop 17:00

You mean like the people that are gonna be go because Yeah, I I don't think uh a lone coder without AI can anywhere near keep up with the what's going on with developers that are writing the code. And it and it's not uh it's not that they don't have any value, the value is just moved up the chain. Yeah, so senior developers that can use AI are super valuable right now because they have both the experience and the the capability to use AI. There's challenges sometimes for the young generation, the people that can get out of school, because yeah, they can they can maybe use AI, but they've never used it in a production environment. They may not have any experience, let's say in CAD or manufacturing to to really be able to apply it nicely. And so that's that's a little more challenging there.

Roopinder 17:51

So this has made this has actually made big news in the tech world because of the the wave after wave of layoffs that are happening in tech. A lot of them are being applied to uh uh uh being uh applied negatively to the coding staff, the developers. Uh I think who is it? Okay, so I'm writing about this right now. This is a company that made up 30,000 people, right?

Mark Burhop 18:24

Oracle Oracle, yeah.

Roopinder 18:26

Oracle just made up 30,000 people, and uh 12,000 of them are in India, and a lot of them I'm sure are developers, right?

Mark Burhop 18:34

A lot of developers they made such a big bet in uh in AI that they need to now clear some of their uh operating costs, and I think they're gonna be developers that are gonna be yeah, it's it's uh you know, it's there's a lot of concern about jobs right now. My my personal thing is is that you know that it's not gonna be what people think it is, but I do think there's gonna be a lot of changing of jobs, and that's still gonna be a big impact. So even if you have, you know, let's say we're we're laying off people in one section, there's hiring, probably equal hiring. You know, that's that's the debate. Is it equal? Is it less hiring, more hiring, or equal hiring? Uh a lot of people think it'd be less hiring, but I think there's new jobs being created at the same time. But that's that's hard, right? You don't it's it's like going back to the time when when horses were replaced by automobiles. You have lots of lots of people in New York that shoot horses and didn't you know supported the horses and fed the horses and all those other things.

Roopinder 19:36

That's an industry.

Mark Burhop 19:37

It's an industry, yeah. And so that what happens when that whole industry goes away replaced with the automotive industry. So now you got lots of new jobs, mechanics, whatever. But you you better be not better be sure, but you would like to see that there's a way to go from the guy shoeing horses to becoming a mechanic on something, so there's not this huge displacement of people.

CAD Jobs Drawings And Automation

Roopinder 20:00

Right. So we'll see what happens. I'm noticing it uh starting to notice some trend like that building in our industry, talking about a CAD cam and CAE world. I'm noticing, and I don't I can't tie this into AI directly. I'm seeing more and more mechanical designers, not engineers, but designers, were having difficulty getting jobs, right? They are, and I know that AI is making some inroads into CAD where the drawing is automatically generated, like feature control blocks are automatically generated, you know, dimensions are automatically put in. Lessening the need for pure mechanical designers who only make drawings, right? So I think there's going to be a not too distant world where the mechanical, the drafter, the CAD drafter working on a CAD machine is just not necessary. The engineer can just practically push a button and generate that drawing.

Mark Burhop 20:58

Yeah. And and so that's that's maybe a good example because in my mind, you you the the thing that should be happening is you should be looking for places that you can where you can make the designer's life easier. And and look at drawings, for example. So drawings, you you're doing your solid model. Sometimes drawings are kind of an afterthought, but you're going to send it to a CNC shop. Uh, other places it's more rigorous, uh, in aerospace, for example. Now the AI can go through and look, did all did you put all your dimensions in the right place according to standards, right? Because there's there's drawing checkers still out there that do that. Um uh the drawing checker job is is probably gone because AI can probably do a better job going forward. Um you know, much like radiology is has improved a lot with AI. So there's other uh, you know, other things with the the the you know, making sure you have all the right dimensions, making sure you you know put it all out there correctly, make sure there's anywhere. So that that's awfully useful that the the AI can just go out and say, hey, you missed a dimension, put It in here rather than you know sending it to drawing checker or whatever and seeing what happens. But even that, I said, you know, I said drawing checkers are probably going away, but uh you compared it to radiology, but radiology is booming. Uh everybody thought radiology would be the first job to go, and it's it's actually doing really, really, really well.

Roopinder 22:19

So you're you've been very active on LinkedIn. I've got to say, I was one of your big fans. Uh oh yeah, yeah. The reason I was a little bit late on our talk today was I was reading your recent report about the uh interfaces, the UIs. UIs again. Yeah, yeah, yeah. Well done. Well done.

Mark Burhop 22:37

Oh yeah, yeah. I uh so I've kind of consolidated all my, you know, I used to do, you know, I'm still on Twitter or still on X uh and other places. I um the I I like long form content better. I don't like these little short, you don't really understand what it says. So I've been writing these really long essays now and putting that on on substacks. I started out there. The only negative part is they're super long and no one wants to read them. Um but I but I I kind of well good, thank you. So yeah, my mom reads it too. So here's here's the here's the trick my mom does. And she sent me a text this morning. Um on that one.

Roopinder 23:14

She says, Well, don't tell me she sends it to AI. It says, What is my son trying to say here?

Mark Burhop 23:20

Yeah, yeah. So so she'll uh I wonder if I've got the text here. I'll just see what it, you know. So you read the you just read the text on uh on the interfaces. So she said, Oh, so she said, oh heck, so my we're on a thread, my wife and my wife and my mother and I, I chat a lot. So I'm looking to see if I can find it. But uh oh yeah, so Mark, I checked out your chat GPT, I checked with chat GPT about your latest essay.

unknown 23:47

Okay.

Mark Burhop 23:47

And and the summary of my eight pages was AI is like a smart helper sitting in the room, but no one knows how to talk to it easily yet. So that's the sum of my paper. Now you don't have to read the paper. It's it's AI is sitting in a room and it hasn't learned to really communicate well with engineers yet.

Roopinder 24:06

So your mom, that's funny. Your mom has to use a chat GPT to figure it out. What are you using, by the way? Are you using Claude? Or are you all of them?

Mark Burhop 24:15

I I I can't, you know. So that's you know, one one of the the the secrets of coding right now is they're giving um a lot of times if you buy Claude Code or you you buy you get a through Anthropic or you get an account through Google or one through OpenAI, they tend to give you a better rates with tokens on the coding UIs. And so what I end up doing is for uh over a year I got by on $20 a month subscriptions because they were getting everybody to go. And typically I'd use all the models depending on what's best. You know, for a long time it was Cloud Code. Cloud Opus was the great greatest lately. Open AI, uh, their latest, uh you know, their latest GPT is really good at coding. Google will come out with another one next week, I hear. That's the rumor at least. And that one's supposed to be great. So I think there's a whole lot of developers right now that are jumping between models, whoever's offering the best functionality at the time will do that. But there's a lot of free models out there too that are really good. And and there's a lot of tasks that you can spin off on those. So that's where I'm at now. It's like, okay, well, this job's a planning task. So let you know, let's give this to you know, uh Claude Opus, and this one's just an implementation thing. So let's just give this to a fast Gemini model and get the code done.

Roopinder 25:36

So I I by the way, this I just I want to say it's genius, and I don't just say it's uh to call people geniuses who agree with me. But I I have to say I do agree with quite a bit of what you're saying, that the UIs, UIs are where AI could make the biggest difference. You know, you're you're a you know CAD like the back of your hand. Uh, you know, and I've studied many CAD programs and reviewed even more. Uh, you know, I was an expert in some for a while, but uh anymore. If I get back into a CAD program, let's say I want to make something in my workshop, right? Or or uh for my deck,

The UI Bottleneck In Engineering Software

Roopinder 26:15

I want to make a deck or shelf. I won't use CAD. I gotta, I have admitted publicly. I'm sorry, uh it's just too difficult for me to use anymore, right? And and it and all of that is because the interface. I mean, I love that it makes perfect geometry and things parts will fit, but that interface kills me because that's that's three or four words I have to learn all over again, right?

Mark Burhop 26:38

It's yeah, it's like it's like stepping into a Boeing 747 with gadgets all over and buttons all over, and and you can do everything with it, but uh gee, it's it's a lot to learn. So do you think like this AI promise the natural language interface that chat GPT gave us and uh and other I don't think it's I don't think it's that I think all right, so if you look at CAD now, so you know again it's like this whole big huge interface of of UI. And what a lot of companies are doing is like, oh well, here's a few more buttons to use AI. So you're you're kind of making it worse in some ways, right? And and AI can do a lot more. So I I think if AI is removing uh obstacles, that's the good thing. So if it's removing confusion about the UI, um if it's making it easier for uh, well, if you're doing uh manufacturing, if it makes it easier to write your G code or your information for the control.

Roopinder 27:36

That'd be perfect for writing G code.

Mark Burhop 27:38

Oh yeah, it's great. Yeah, so it you know that and there's still a lot of people that do that. So that you know, it's it's it you you're you're essentially using the same technology the coders are using now. And uh it can it you can just do some wonderful things.

Roopinder 27:52

Yeah, but yeah, I think yeah, uh go ahead, go ahead. No, no, no. Do you think CAN will have a uh one day will have a natural language interface or no, right?

Mark Burhop 28:02

Well, yeah, no, I think it will. Um I don't I don't see how you cannot get there. The if you want one, if you if you want a natural interface that says, hey, just go create me a uh here's my deck, here's some measurements, you know. Give me a give me a deck that kind of matches with that.

Roopinder 28:17

I think that will probably be because that's that's kind of my dream, right? I'm not a developer, of course, so I could dream and I can have unrealistic dreams. Yeah, and I kind of want to say, like, I want to next time I make a shelf, I want it to say, hey, CAD program, whatever, whichever, make me a shelf. And you know, and and I know it'll suck at it at first, right? It won't, right? But eventually I want it to learn what kind of shelves I have made, what kind of materials I use, right? What kind of shop I have, what tools I have. Take all that into account, and you know, at least say, okay, well, here's some ideas, right? I want it to interact with me. That's where I dream about CAD being, right?

Mark Burhop 28:58

And that's that's the right, yeah. So take that. Yeah, I guess there are two parts to say here. One is so what you're doing, what you want there is essentially what I, you know, the the homolog or the the the same thing in coding is the vibe coding. So now you can vibe code it something and get something good, okay, usable. Uh, and I think that what you're going for right there is is easy. The harder part now is I want to design a new airplane. That's not something you can you can, you know, one shot a prompt for. Um but what you can do is all it could, but it could work with me, right?

Roopinder 29:32

Okay, let's not.

Mark Burhop 29:33

Oh, yeah, but see, it can do all the stuff before. So you okay, I I want an airplane, but let's start collecting some requirements. How big is the airplane need to be? What are all the things? Have you thought about regulations? You know it's going to take this long to get it approved, and and and all those things that it can work with you on. Right. The whole development process is is is wonderful because it can do this research very quickly, it can update documents in your PDM system.

Roopinder 29:56

And it can understand, it can I say I'm doing this at Boeing or Airbus, right? It understands what aircraft have been before, right? And I want so it can, it doesn't seem to be an unreasonable request, although I'm sure sci-fi-ish, but make me an airplane like this, like the 787, only I want it to go not as far, uh, and and I want it to be more aerodynamic. I can I can make those kind of requests because those are modifications. Oh, yeah.

Mark Burhop 30:23

And those those are more research type requests too, because you know, if you were if you were a person to go do that, right? You would go look, you become an expert on a lot of the things that the old design has, and you'd be thinking, okay, where's the areas that I have to work on to modify this? Uh, you know, how did we have processes before to do that? It would know all your processes, it would, you know, set everything up for you. It could schedule the meetings for you, you could say this is the expert on this person, these are the people we need to talk to. Exactly. All the all those things. Here's here's some uh concept diagrams, here's some even some concept UIs that we wanna you want to do. So all that is totally accessible to AI. And so that's why it's a really hard thing with AI. There's uh, you know, one hand, I I keep looking at what companies are doing, and it's like, oh man, why that's never gonna work. And then you see some of these easier things like what we're talking about here, and and uh those should be trivial to do, but not enough people are focusing on them. So I think that's one of the the challenges in the cat industry, is there's a lot of diversity in where they're investing in AI. Uh, I don't think they've found the best ones yet. I think a lot of I think as a whole, the industry's throwing darts

Startups Generative Design And Exploration

Mark Burhop 31:30

at the at the dartboard and seeing which ones actually work well. Yeah.

Roopinder 31:34

So who is throwing these darts? Because I see one of my colleagues on the internet, let's say, Michael Fino, Finocchiaro. He does he's amassed a study of uh hundreds, literally hundreds of companies that are have AI CAD or CAM or CAE programs. The only startups there's a literally hundreds of them, right? I thought, wow, this is amazing. The CAD companies themselves, as you point out in your report, are also have AI filtering in now. I would say very slowly.

Mark Burhop 32:10

Yeah, I it's it's that was kind of my uh repress impression too. It's I mean, it had some good things. I've seen some nice things coming out. Um I I saw the Seau presentation on generative um design that I thought was pretty good. Uh I saw uh uh you know from Siemens doing uh it wasn't G code, but the language for their controllers using AI for that. That those are nice things, yeah. Um but yeah, it's too conservative, and a lot of times these companies tend to watch what others do first, and they they're they can be fast followers. So I think if you want to see what's really going on, you have to kind of watch the startups out there and see what they're doing.

Roopinder 32:49

Do you give them any hope? Uh I'll tell you why I'm saying this because I don't see any of them. So there's a few that are getting gaining more visibility, but I don't see them gaining a lot of traction or penetration in the market. I think most people are sort of happy with their programs, as it were. I don't think they're ready to jump on the ship of like a startup to do handle their design work or even give it time of day. Uh so I think the CAD companies sensing this are probably are taking a slow road. Like that's not the right term, slow walking this. Yeah, right. Right. They they can afford to. And uh I I'm reminded of a conversation I had a couple of years ago, and I may even mention this last time we talked about AI, it was a couple of years ago. And uh he said, sure, you could have AI make Chat GPT do a prompt, you know, after a prompt designed a bracket for you, but the time that it would take you to tell it exactly what that bracket looked like, or you know, how thick it was, where the holes were, you might as well have made it yourself. And I think that's that's what cat operators also think. It's like I'm not gonna use AI to make prompt me. I don't need it to do natural language interface because I can whip it out. I can whip it out.

Mark Burhop 34:03

Yeah, so so so here's the the answer to that now is is yeah, totally right. So I I got you know, uh uh say a dozen of these trivial kind of tasks that I could just whip out in no time. Yeah, so rather than me doing it, I set up my agent, I give it, okay, these are the tasks I got to get done. They're all trivial. It takes it three times longer than it takes me, but but that's okay. I go to bed and I wake up and all the tasks are done. So sometimes the it's not the speed of that you can do it, but it's the throughput that you get. So I think there's some some cases like that where it's uh even in software development, you're seeing that where we're like I said earlier, people are spinning things off to agents to solve multiple problems at once. Uh that that kind of parallelization is harder to do if you're you if you're not AI, otherwise you're doing it all. Or you're delegating to other people to do that. Yeah, I think there's lots and lots of opportunities. But like you said, I I I totally agree with you. I don't think AI is anywhere near um as helpful as it should be right now for for engineers and CAD.

Roopinder 35:11

Yeah. Yeah, especially in MCAD and CAD. Uh I think the EDA companies are doing a lot more, like design with designing circuits, there's a lot more uh AI implementation there.

Mark Burhop 35:22

Yeah, and I think that's a it more easily. I don't not that that's that's an easy topic, it's an incredibly hard topic, right? But uh as Jensen would tell you, um, but it's it's in terms of testing and and validating that you got a good solution, that's pretty good.

Roopinder 35:37

Uh and that's there's a lot more optimization and and uh yeah that can be done and e maybe EDA than can be done. Although I I'm I'm thinking of something else, Brad Rothenberg, of course, right? Ben and kept up with what he's trying to do. And uh oh yeah, yeah, yeah.

Mark Burhop 35:56

He yeah, yeah. He's he's a good guy. I've I've liked him from the very beginning when it was started. So uh yeah.

Roopinder 36:03

He's a he's one of the it's at the forefront, I think, of uh what it what could be, it still could be a revolution, because he's talking about computational design. And and he's he to he's taken it upon himself lately to design airplanes, right? We're talking about designing airplanes like hey AI, make me an airplane. Okay, well, can't not knowing any better, Brad says, okay, I can do that, right? So he's got all these airplane designs for these drones, and and and I've got to say, okay, he's not making that whole airplane, of course, right? But the some of the shapes he comes up with are fascinating, interesting. And that's and that's a great example, I think, of telling AI to just go explore all the possibilities. Come back with me with what works, or I'll select what works or what looks good or what's manufacturable, but it comes up with hundreds and thousands of designs. Yeah, I may not have thought of, right? In that case, it's dealing with designs. I can't intuitively, I don't feel I can intuitively get what a perfect streamlined shape is of a plane that needs to go like hypersonic. I can't get that because like none of those things exist to me and that I can see or feel, right? But let AI go crazy, right? Let generative design go crazy or topology optimization. Do throw a million things up there and see see what works. That to me is like the dream, right? Now, AI could help me filter out those things because I don't have time to look at all those or study them, right? But AI could say could like what's the word I'm looking for? Aggregate or or evaluate or you know, be very critical of things, right? Uh if I say good, say pick me the best airplane, maybe it should know what a best means and look at all those power designs, right?

Mark Burhop 38:01

Yeah, and and that's one of the you know, one of the challenges with AI is is is basically the prompt. So uh how do you give it a good prompt that it does the right thing? Um and and so that's one of them too. It's like you got to make sure you get you have to anticipate what the AI needs to know to make a good airplane to make sure that information is provided in some way or another. So that's kind of the real value of humans right now, I think. Yeah. That experience of knowing what you got the super bright intern, but you know, he doesn't know how to make airplanes, but he keys, oh, he's a make maybe he's a mechanical whiz or computational design whiz. Um you need to tell him all the constraints, right? Because who knows what you get otherwise. But that's a you know, so but you know, what you're talking about with Brad, though, that's we've been want everybody's been wanting to do design exploration for a decade, right? I want all the designs out there, and I want to Autodesk was real good about talking about this a while ago. Yeah, with AI, it suddenly becomes a lot more pop possible just because the AI can just kick it out. It's just tokens, it's just money. You want 10 designs, you want 100 designs, you know, just pay 10 times more and you get 100 designs and pick up.

Roopinder 39:07

Yeah, yeah, but that but that that's creating a if I let it do that, I don't want it to create a lot of work for me. I kind of have time to look at them all or evaluate them all. I want something and evaluating them for me.

Mark Burhop 39:18

Yeah, you keep adding, you keep adding layers. So now you got an A. So this is where agents are cool. Um, and and so what you're describing here, so you got one agent kind of going through coming up with all the different designs. That's all they do is think about designs. You got another one out there that's I all I do is evaluate designs, is you know, how good is this gonna be with fuel? How good is this one gonna be noise-wise? Because a lot of places don't like loud noise. How good is this gonna be for all these other things? And it goes through and evaluates those and does some ranking for you. So now you instead of get a list, you get wow, here's here's 10 of them, or or if you're a CEO, here's three of them. Uh, pick your favorite one of these three. Uh and you've gotten some good uh answers there. So I think that's where this these levels of AI is. I think, well, that's where we are at with software development now, especially especially the early adopters. Uh, I think we're gonna move towards that too in in product design and and manufacturing in general

Trust Adoption And Non Deterministic AI

Mark Burhop 40:15

as well.

Roopinder 40:16

You point out in your in your report about the interfaces, we should get the full name down because I'll refer to it. Interfaces is a primary bottleneck for AI and product design. Uh, one of the things you say is, here we are, okay, a drop in confidence in AI. Uh okay, so here's your quote. The implementation wall globally, only 69% of business leaders now say AI will enhance their industry, a 12-point drop from 2024, and their trusted AI has fallen 11 points to 65%, right? So we're actually expecting less. Do you think this is a sort of a correction from the wild eye?

Mark Burhop 41:00

I think this is a a uh a correction from irrational exuberance that was there before. Uh, so I I think people are understanding more uh what AI can do and can't do. I think they're understanding better how quickly it's advancing. Uh because it was, you know, there's a lot of talk of it being on some type of exponential curve. It's like a year for a really good model now, but six months later, and you know, three months and less until changes happen. So uh I I where I've adjusted here is again in manufacturing, I think there's a longer timeline just because it interfaces so much with the real world, and I think that's more of a challenge. Um, the indeterministic nature of models, that that is a huge thing. So AI models, especially LLMs, I should say, are non-deterministic, right? So I I you know give you an example of what that is. I I had it, I had my agent going through and I said, okay, do do five five coding rounds, and then on the on the fifth round, build test cases and test it all and fix all the test cases. So my my my model says, okay, well, this is round five, I'm gonna do it. This is round ten, I'm gonna do it. This is round uh you know 15. Oh, round 65. Well, that's not divisible by five, so we're not gonna do the testing. But 65 is divisible by five. So you you get this unexpected, non-deterministic uh thing that happens. Uh manufacturing has to be able to deal with that. Now they they do already. They they you know, people are non-deterministic, right? So we're we're used to working with people. You I think you just need to have the mindset that AI is going to be more like people than it's gonna be like a software program.

Roopinder 42:44

I think we expect our software to be deterministic, to have exact, yeah, repeatable, reliable answers, right? I don't want it to do two plus two equals four most of the time and maybe another time, right? Because I'm I'm infuriated by the fact that it gives me a different answer every time I ask it the same question.

Mark Burhop 43:03

Yeah, now there's some things you can do to get better, more consistent answers, but it's still What's your trick? Uh well it has to do with the temperature of the model, so it's a lot something you can't necessarily change. So it it kind of controls how how created the model gets. So if you tell it don't be creative at all, uh it comes up with similar answers. But if you say, okay, start it, start thinking a little bigger, a little broad more broadly, then it will go down different paths more often.

Roopinder 43:28

It'll go down different paths, different paths each time. I can perfectly see how that got it to where it was first famous. Like ChatGPT was used to generate a different question to essays that English teachers might have asked them. And you can't have it have the same question because a teacher will be on to you. Hey, this is yeah, this is chat.

Mark Burhop 43:46

So I can still you can still detect it pretty so now that now there's this race going between students. Uh, first they're using AI, and then they figure out to detect AI, and now we figure out how to uh obuscute the AI. AI written code, so it looks like human written code. And now the AI, you know, detecting that as well. And it's it's terrible. I feel bad for the students right now. And I used to teach C at the University of Cincinnati. And uh I would hate to do it now. I I would not know it's not clear to me how much to bring AI into the curriculum versus not to. Because I I think the value of software developers today is they know how the code works and they know with good architecture.

Roopinder 44:28

Yeah.

Mark Burhop 44:29

That's from years of experience doing it. And what do you do for you know new programmers? Do you make them write code without AI? That seems terribly inefficient in nowadays. So, but how do they how do they build that experience too? So yeah, well we'll figure it out though, because I think it's I don't think software development, a lot of people think software development is is gone, it's cooked, but I don't think so. I think there's gonna be uh a lot of really new jobs. AI is just too valuable right now.

Roopinder 45:00

Uh-huh. Uh yeah, I don't see the coders that I've met that are keeping their jobs just have to make more code. It's like, okay, now you can use AI. Now make me lots more code than you couldn't do before. And uh I'm not gonna give you any more money. That's my AI benefit.

Mark Burhop 45:15

Yeah, yeah. So I most developers, it's funny because you you you everyone clocked about oh, when AI comes, you we're gonna have four hour or four-day uh work weeks, and everyone's gonna be relaxing. And and I know very few early adopter coders that are feeling less stressed, they're feeling more stressed because they go to lunch, and you know, well, I can't go to lunch because my agents might need me, and that's that's five agents maybe working. So if I go to lunch, my whole team goes to lunch. So it's it's crazy things like that. Uh that people are having difficulty with.

Roopinder 45:49

Yeah, it's not I don't know. I don't I've yet to hear encouraging reports coming out of the workforce about AI, all the way from like, oh, AI is going to cost me my job. Now now AI is making me

MCPs Tools And The CLI Alternative

Roopinder 46:00

work harder. One thing I'm gonna draw it back a little bit to the CAD world. I've been very encouraged to see MCPs, model control protocols, right? And uh I think Anthropic put them out.

Mark Burhop 46:13

And it's yeah, Anthropic came out with a standard uh gosh, probably about 18 months ago.

Roopinder 46:19

Yeah, okay.

Mark Burhop 46:20

So it had a lot of problems at first, so there was a lot of iterations onto it as well. Uh, and there's still there's still challenges. So you know, a lot of people saw MCPs and say, okay, what so just explain what an MCP is for everybody else.

Roopinder 46:33

Yeah, please.

Mark Burhop 46:34

Uh an MCP is basically a API in a certain format. So you you give this information on the API to the model, and the model, if the model is able to run tools, which is most developers or anything else, then it can say, oh, you're asking about you're asking about how to get rid of chatter for CNC. And I got an MCP over here that knows all about chatter and C and everything. So I'm I'm gonna say, Oh, the user wants to get rid of chatter. I'm gonna call the MCP with the some information. It's gonna come back with a bunch of really good information on chatter, and it puts that in the context, and now we can have good good answers to things. And or you it's not even just collecting information, you could be doing it for running tools. You could say, okay, go out and copy files for me now, or check my code into GitHub. All those are things that you can do with an MCP. Um, the challenge with it is when you start getting a lot of MCPs. So let's say I I load up 80 different tools in my context. Each one has a bunch of information on how to call the tools. So now that the AI has to remember all the tools it has, how to call all of them. Uh it has to sort through them when it calls the tools. And also all that information is taking up a bunch of space in your context. So you only have so much space in your context. So a lot of people are uh MCPs are still valuable, but they're getting away from using MCPs. They're using uh a lot more you'll see CLIs, command line interfaces, because these models are so good at calling a uh uh a Unix shell or uh Unix Solan.

Roopinder 48:17

Oh our MCPs they introduced about 18 months ago. If you think they're already passing.

Mark Burhop 48:24

So you can actually set these tools out. What's that? Say that again.

Roopinder 48:28

Uh the MCPs are already passing. Are companies are companies going to a command line?

Mark Burhop 48:35

I I think in um I think what their role is is being reduced right now because I think uh CLIs are better in a lot of cases. Because if you call a CLI, the just command line interface, the chances are that the the information that model needs to call that CLA, the command line interface, was already in their training. So they don't have to go out and call other applications to learn how to do that. They already know how to to copy files, they already know how to do an LS in a directory, they'll know how to use curl or anything else to do these things. So all of a sudden, anything that was an old Linux command or an old Unix command, it you can just use as a command line to do those things. And um I think that's a a really nice way to go too. There's another uh agent-to-agent protocol that's being done too, so you can get these agents to start sharing information and context back and forth. So I think that's growing. I I I I expect more, I expect this to be a uh a rapidly changing area uh going forward.

Roopinder 49:40

Yeah, rapidly changing, although maybe not rapidly changing fast enough. Okay, another part of your report.

Mark Burhop 49:46

Yeah, it's it's it was out, you know. I guess like I said, MCPs came out 18 months ago, and uh that's a pretty long run in AI time.

Roopinder 49:54

I guess it is. You say here 70% of respondents believe AI will negatively impact operations, but only 27% actually use them in workflows. You're quoting a 2026 analysis done by Dan Cumberland Labs. Yeah. Yeah, okay, right.

Mark Burhop 50:12

So So the kind of the the feeling there, so this this gets back to using a I I did another example or another essay earlier about kind of the difference between coding and manufacturing. So in coding, if you make a mistake, you you roll it back, right? You don't uh you can run all the tests you want. You can run a thousand tests, and then if 999 fail and the last one works, you're still fine. You can't do this in manufacturing. There's there's uh, you know, first of all, you gotta get a you gotta get a machine to do testing because you're doing physical AI, which is hard enough. It's always been hard for researchers to get a hold of the machine because the machine's got a job and it's gonna go to do that. So now and now you're you're saying, oh, but we're gonna create a thousand bad parts first before we ever get one. You you can't do that in manufacturing. That's that's too much scrap, that's too much cost, it's too much machines offline. So there's a lot of barriers there that I would say are gonna make AI look slow and and not be the best solution for things. Um, I think there's a lot more up front that is valuable. For example, you can stand behind one of these big CNC machining tools, you know, one of these multi-axis ones. They have a big monitor right there and they got all the documentation. And if you get an error on the machine, it would be certainly nice if the AI piped in and says, Oh, you've got error such and such, it just means you need to turn this dial half a turn. You don't have to go searching through the documentation or or doing anything like that. So those are nice ones, those are ones that help people out and relove some barrier, and the operators are happier because they don't want to go searching through documentation, the parts come out better, uh, and uh the throughput is good too. So those are the times of things I think where AI is is really good and helpful.

Roopinder 52:05

Um users aren't putting them in their workflow. But imagine, like you were saying, to out of the tool makers, the CAD vendors are starting to put them in. So they're getting some use of AI without actively trying to hook up with a AI program themselves or you would use us, even try a startup or something like that.

Mark Burhop 52:27

I think they're just but I think they they have to to a certain extent. The you know they they have a lot invested in these these big um systems, right? And so it's not something you it's not just oh well, we'll change off it and do this instead because all these connections to all the other systems is hard. That's where the startups I think are a little more interesting, and they can make these big statements. One of the things, so I I work with OpenClaw a lot, but I don't know if you know what that is, but uh the creator of OpenClaw uh was doing these agents that just go off and do things, and he said basically UIs are just slow APIs, meaning what we can do now is is I could create right now an agent that would go through and actually click the buttons in your favorite CAD system and do all the the work there. And and I'm sure there's examples of that already on the on the web.

Roopinder 53:22

Um all right, that may seem like an ordinary thing for you, but for somebody uh dubby like me that doesn't program, that's that I would say. That's huge, right?

Mark Burhop 53:31

So so so but you go from that and it's like, oh my gosh, that as a developer, that that's it's bad. You don't want to be looking at a graphics window and making a cursor drive around to click buttons. Um, but you can't, and you can make an API out of it. But the best is to just call the APIs directly. So I think that is going to be one of the bigger changes that that come. Um for developers, but no, not for development. No, what yeah, the use so the user with an AI can though, right? So you can you can sit in your your CAD system or your nice your AI and say, design me a deck, and when it pops up, and here's your your deck, and oh that's too big a deck, I need it smaller, and you can you can change it that way. You don't need a CAD system and all those buttons, it's just calling APIs and pulling out the the CAD model and displaying it for you. Yeah, though I think I'm not saying this is coming anytime soon. I'm not saying this is gonna change things in the next three years, even which is a long time in AI time, but I do think eventually the the UI for this will change that that uh the way you you get to a final product, a way a designer gets to a final product is gonna be a different type of mechanism than what we have now with clicking buttons and dragging mouses and rotating models looking for the face that you want. Uh type of thing. I think there'll be better systems with much different UA UIs, much better UIs uh that drive all that.

Roopinder 55:01

So you remember my minority report where they UI, that's like UI hall of fame kind of stuff, right? Where he's yeah, he's doing all these things with the screen. And uh and uh and to this day, UI designers talk about that, but we still don't have it.

Mark Burhop 55:16

That was a we yeah, and it's it and uh and and those are always good too, because it's the same thing with the um the the VR the VR folks. And I and I love VR. I got you know, I'm I've got them sitting over here at my desk all over the place. Um, but I hate taking that thing on and off and putting it on and getting it's knocking my glasses all over the place.

Roopinder 55:36

It's cumbersome, right?

Mark Burhop 55:38

Yeah, and and so I don't use it a lot because of that. So it's it's we a lot of times you see these things in movies that look really cool, but in the real life, you know, there's a lot of practical things. Same ones with your you're moving your hands and gesturing all over the place. My arms would get tired after that, right? So um, I I I think some of these interfaces are cool. I think there will be some things like that come along. Um, you know, but the mouse has survived a long time, so it's gonna be had to be something pretty good that's better than we're it might be okay for us, Tom Cruise.

Roopinder 56:10

He does all his own stunts at you know, like he's practicing he's like our age, and he does his own stunts, and I'm sure he can do that interface all day long, right? But yeah, we we can't. But that's the future never caught up to that. I think it was just the fact that it can you know recognize hand gestures is and and fake the way he was moving his fingers and pointing. There there was, and that's sort of coming around, like there are, but still, even to this day with your VR headset, you control it by doing dumb looking things like pinching, right? Here's a pinch or a double pinch or something, or right? And that's like, oh my god, didn't you see that movie?

Mark Burhop 56:50

Yeah, and it seems it seems like it should be intuitive, but uh you know, I I was looking at 3D monitors not too long ago, which was the cool as anything. They they track your eyes, so each eye gets a different set of pixels or not pixels, but proton or photons if you want. But everything looks three to eight, three D and supposed, and then that then you'll they'll hook it up with something that can detect your hand, so you can go out and grab it and move it around, but it's it's still hard to do, right? It's not as easy as as physical objects.

What Comes Next And How To Reach Us

Mark Burhop 57:20

So I um I so I think more work is needed there. Definitely.

Roopinder 57:24

I've taken an hour of your time. It's so much fun to talk to you. I'd love to talk to you. I hope we can do it again and catch up.

Mark Burhop 57:31

Oh, yeah, anytime. I I you know, I'm this is all fun for me. That's you know, you you you say people are worried about it, but I'm I'm excited about it. I never had so much fun in my career.

Roopinder 57:40

Um it's like this is we got a front row seat. I mean, you're directly involved, but I have a front row seat to all this technology. Yeah, just watching it develop so quickly in front of me. Uh yeah, no, this has been it's very good. So, Mark, how do you what do you do now? You retired from are you retired now? You're not retired from the time.

Mark Burhop 57:59

Technically, technically I'm retired from Siemens because I was old enough when I left. Okay. Um, yeah. So I I was doing a little bit of AI before I left. I was working on what they call synthetic data, the last thing, but I really felt there was a lot I could do outside of Siemens, um, just on my own. So yeah, I got together with some friends. We we have a startup going. We're not ready to talk about it yet. We still got a lot of you know, a lot of secret things going on, if you will. Oh, oh I I don't think they should be secret, but you know, you know the money people there. It's like, hey, don't tell anybody, don't tell anybody. But uh um, so that's what I'm working on right now. So I have a little bit of free time in between, between you know, product launch when I probably will never sleep again, and and right now. So that's why I'm trying to get some of these essays out there and and start talking to people. And some of it's just good to get the information out there to see where other people are at.

Roopinder 58:52

If something's looking looking at what you've done and what and your some of your conversations are linked in, if I could call them that, it seems like you were freed from the gravity of Siemens. I know you can do more of the things you're interested in. Yeah, that's that's about right.

Mark Burhop 59:05

Not that Siemens, Siemens was was very good about letting me talk about things and doing different things, but there's there's a lot of rules, right? Any any of the big companies have rules and uh you know things that you you need to be careful about and and so on. So yeah, it is it is nice being free and just being able to to talk about whatever I want to and yeah, you know what works, what doesn't work.

Roopinder 59:26

This is this might be turning into a semen sort of day. Later on, I'm gonna talk to Brittany Ing from uh Siemens Maritime Division. So she was uh so I just realized it's gonna be all all seamen's or ex-Siemens. Mark, it's been great. Thank you so much for joining me on the show. And uh good look good.

Mark Burhop 59:49

Yeah, thanks for Roopinder. If you get a lot of hits, I'm glad to come back. If they're all negative, that's fine too.

Roopinder 59:56

All right, you go, you go make the next big thing, and I'll be anxious to hear.

Mark Burhop 1:00:00

Yeah, that's it. Well, let's let's let's uh when I when I have something really good to talk about, then we'll be in.

Roopinder 1:00:05

All right, all right, Mark. Talk to you later. Bye bye. All right, be back. Thank you for listening to Faux Des, the future of design and engineering software show, brought to you by ENGtechnica. I hope you have learned of a new application or technology that will help you with your job. If you have an application you think would be of interest to other engineers, please let me know by emailing me at roopinder at engtechnica.com or message me on LinkedIn.