KP Unpacked

Would You Like Fries With Your Engineering

• KP Reddy

Share episode

If you can ship a site plan with one click, do you still need the degree or just better prompts?

Director of Innovation Brandon Blackberg (RTM Engineering) joins KP for a fast, practitioner-level riff on how AI is reshaping civil and structural work. We cover the “McDonald’s vs gourmet” future of delivery, why generalists win in a systems world, and how to avoid getting locked into someone else’s feature stack. We also talk vibe coding, token bills, and what KP would do if he started a firm tomorrow.

Highlights 

  • Why entry-level engineers will start closer to “junior-plus” with AI at their fingertips
  • Generalist thinking vs narrow specialization when failures are system level
  • Rethinking drawings and phases when iteration is cheap
  • Time and materials, lump sum, and the real impact of token costs
  • Multi-LLM strategies to avoid platform lock-in
  • Guardrails that keep creativity alive and production safe
  • Build vs buy: when an internal tool beats a vendor, and when it doesn’t
  • Career paths that don’t look like ladders: engineering, coding, sales

Learn more about KPR Co Mastermind Groups here!

Upcoming KPR Co events

Sounds like you? Join the waitlist at https://kpreddy.co/

Check out one of our Catalyst conversation starters, AEC Needs More High-Agency Thinkers

Hope to see you there!

SPEAKER_00:

All right. On today's podcast, we do a little bit of a different type of episode. Just kind of spending some time with uh some of the folks in our mastermind community to let's say riff a little bit on what's going on and what we're seeing. And um I think they always have unique insights, uh generally somewhat differing from mine since they're day-to-day practitioners. So I thought it's always fun. So today I've got Brandon Blackberg. Uh hey Brandon, how's it going? Pretty good. How about yourself? Not too bad. Work on my radio voice projecting and just gonna start doing drive time radio at some point. Um, why don't you kind of tell us a little bit about like what you do, a little bit of your background generally, right? You know, of course what you're doing at your company, but just general background, so I think it's pretty unique.

SPEAKER_02:

Yeah, so I'm uh director of innovation for a company called RTM Engineering. Um, I want to say I got to where I'm at by kind of weird circumstances. Um, I kind of started as a uh CAD drafter, designer, operator, whatever they called it at the time, um, and kind of just worked my way through going through college and um was a civil engineer for a very long time. At least it felt like it was a long time. And uh what kind of took it off the rails is one of my friends showed me how to build a computer. Um, and then from there I was kind of hooked on the technology side and what I could how I could enable technology to do what I needed it to do in a better way, essentially, than what the company was already doing. And uh ever since then I've always looked at things like, all right, I know this is the way I should be doing it, but how could I do this better?

SPEAKER_00:

Yeah, very cool. Um so I you know, we we have to talk about AI within the first 10 seconds, or AI gets mad at us and it makes a note that we're we're not pro-AI, right? So we always have to stay pro AI because it's watching. Um but you know, I always view like kind of some of the dynamics that's happening, you know. Some people are uh afraid of it all and what's coming, uh, and others are viewing it as an opportunity. Uh, I think I can assume where you're at on that, but how do you how do you think about like younger folks, kind of that younger folks versus maybe the more experienced senior people and kind of that dynamic, specifically as it relates to maybe like hiring and adding talent?

SPEAKER_02:

So I think the younger folks had a have a uh a big advantage here. Um, I think as some of some of us older folks, I would include myself in that group sometimes, um, we're kind of get stuck in our ways a little bit. And I think the younger folks are willing to challenge the way things are done. And as far as like, you know, getting into the AI, and you know, they're already there. Um, a lot of them are learning this in high school and now they're going through college. So they're already used to prompting, they're already used to kind of what the power of AI can be. So I think when they come in with that fresh perspective, that fresh look on things, I think it's it's I think it's a great thing. Um, I I always always tell people to look at things from a different angle, and I think that's perfect.

SPEAKER_00:

Do you think um, you know, we do this thing now where we prompt for writing, right? I've I've I've written a few books and things. I don't consider write a lot. I don't consider myself like a professional writer. Um I'm kind of a hack that enough people like like to read my stuff, so I keep writing. But um so we've seen AI take like non-writers and make them better writers. Do you think there's a time where we're gonna say you don't have to be a civil engineer? Because AI can help you, right? Like you don't actually have to have because in the world of like creative and writing and having lots of space, so to speak, lots of white space for what the indeliverable is. You know, one can argue that everything that's ever been done in civil engineering is highly documented, right? We have a blue book and a green book and specs, and there's uh a lot of orders of constraints, right? Both at a system level and at a granular level, that kind of the computer can do. So, like, do I need to have a civil engineering degree?

SPEAKER_02:

Oh man, that's a great question, especially somebody who used to be a college teacher as well.

SPEAKER_00:

Um so you know, uh I always like to look at it by the way, the editing team will pull out the best sound bites on you and make you look worse. So it's okay.

SPEAKER_02:

No, no, all good. Um man, so you know, it's a hard question there. So, you know, there's two different ways I look at the the the engineering space in general, because I wouldn't say it's just a civil engineering prom. It could be, I think it's across almost all the engineering fields. Um, if you look at it, we have places like McDonald's that's you know, hamburgers quickly fast, right? And they're serving them up, but we still have a need for that gourmet burger place that charges$30 for a burger, right? Because that one time you want to indulge in something. And the way they you can't really compare the two because the two are still completely different. Yeah. So in composition. So I think where AI is going to thrive is on some of those repetitive tasks and some of the, hey, we can crank these out, McDonald's style, you know, drive through. Would you like fries with that with your engineering? And then for some of these more gourmet uh engineering projects that require some of that nuance, I think you're gonna see that some of those more complicated projects are still going to need that civil engineering degreed person. Now, what I will say is I think what's gonna happen is some of those entry-level engineers, when you first start, you're gonna start with a level of knowledge at your fingertips that wasn't there before. So that engineer, when he starts out of engineering school, is gonna have is basically starting at what we would consider a junior level engineer now, almost right out of the gate, because they're gonna be able to prompt figure out what their answers are faster than having to go bug your senior civil engineer. Hey, I'm in this situation and I'm trying to grade myself out of this hole.

SPEAKER_00:

Yeah. Do you think? I mean, I have this hypothesis generally that you know the industry's gotten very decentralized and siloed, and you know, especially I look at structural engineering specifically, right? We're not in we're not inventing new steel members, right? Like it's all kind of there. The catalog exists kind of put it together, right? But you know, there was a day when a civil engineer did a lot of things, right? They were like doing hydrology, then structural, then I mean they were very multidisciplined within structural and within civil engineering as a discipline. And in fact, a lot of architects back in the day did their own structural engineering, right? They they didn't actually have, you know, they didn't go to another firm. So one of my theories is that we have to all also become like master generalists, right? We've gotten so specialized that maybe it's time to go back to actually understanding everything about a building uh and how it goes together. I mean, what do you how do you think is that is that a fair hypothesis? What do you think about that?

SPEAKER_02:

I would say that's fair. I think honestly, being a generalist in in general, not to keep going on that word, I can't stand it, but um I think is honestly kind of where it's going to be at. I think that's one thing that's kind of helped even in my own career, is just knowing a little bit about everything coming from just every spectrum of the IT world, the civil engineering world, the drafters, even in my teaching when I was doing teaching, all that stuff comes together to create a more well-rounded uh person where you're thinking about all this AI stuff because you're able to look at it from different ways. Um, and I think it also helps with prompting. So we're not trying to prompt it as an engineer, we're trying to prompt it as a person.

SPEAKER_00:

Yeah, because I I think one of the things that I see is um, especially when I see like engineering failures, right? Um, and I I think I did an episode last year when the big storm hit, hurricane hit Asheville and just demolished. And I was like, well, it was kind of a systems engineering failure, right? It wasn't anyone's anyone's component, whether it was drainage or whatever, right? The watershed, it wasn't any one thing, it was a confluence of events because nobody took a systems approach, right? Nobody looked at, and I was talking to a friend of mine up in Canada that does like wind tunnels and stuff. I'm like, so you do wind tunnel analysis on a single building. The reality is if you're building a skyscraper in New York City, you can't just do wind tunnels on that isolated building. It's a it's a system, right? There's other buildings and other variables, right? Which is perfect for computation versus building dumb little models, right? I mean, how many dumb little models are you gonna build, right? And then run it through wind, right? And so I think um one of the things I kind of look at is that it pushes us to get out of these like component engineering, you know. Oh, I'm gonna work on the pipe sizing, right? Like there's this to really people need to understand more, right? Like, because you have to understand the systems approach of uh of all of it. Um so I don't know. I mean, I feel like that's been lost a little bit.

SPEAKER_02:

Yeah, I mean, we talk about it a little bit, you know, when you people bring up digital twins all the time, but the digital twin is of the building itself, not the entire system around it. So, and that's kind of gonna be that next piece of you know, when is it when is it gonna be the entire thing?

unknown:

Yeah.

SPEAKER_00:

So a little bit of a shift. Um, so I I gave this talk a few weeks ago, and I don't know, people just get scared. I don't know why, but um, you know, it's kind of bringing up the point of how we iterate drawings. It was just kind of a simple idea, right? We iterate drawings because back in the day, if we were drawing with pen and paper or vellum or whatever, the the cost of change was so high, right? We had to say, okay, let me sketch out something, sit down with the customer and say, okay, does this does this seem about right? Okay, yeah, it seems right, and then you you kind of start filling it in. But like, we don't need that. Like, why do we need all these phases of drawings? Um, which means like, why do we need to do all this extra work? I mean, how do you think about like just the the way I mean we haven't changed how we think about schematics and conceptual and like like why do these things exist?

SPEAKER_02:

Well, I think a little bit of it exists. Well, obviously, from the owner's perspective and the developer's perspective, they want to see you know all the multiple iterations of what could be on the site and what works. But I think with AI, a lot of that stuff is all going to be automated now. And that's what that's what I've been trying to do on the uh conceptual side is automate that process is so that uh you know owners and developers can see this without even involving the engineer. Um, they can kind of basically be able to say, Hey, I want to develop this site, I own the property. You know, what's possible with this site and let AI go to town on it and be able to come up with at least you know 10 or 15 good concepts for it.

SPEAKER_00:

Yeah. Somebody told me it was because we wouldn't know how to build. Yeah, we wouldn't know how to build complete build completion, which was the argument with them, right? The argument with them was all the work is being done up front, but I'm still getting paid on the deliverable. And so like I end up upside down on my cash flow. Does that come up at all like the the billing part of things?

SPEAKER_02:

No, that doesn't really come up in uh in our talks here, not too much. I mean I mean the only part of billing that comes up with AI is you know time and material versus you know lump sum.

SPEAKER_00:

And do you guys do much time and materials?

SPEAKER_02:

Um so we do have a bit of time and material, especially on the civil side. Um, but the uh the other parts of the company would be mostly lump sum.

SPEAKER_00:

Yeah. And how do you guys think about that? Like is I think everybody has this fear that somehow AI is going to reduce fees and this and that.

SPEAKER_02:

And so I still remember the time when they used to bill for my usage of a CAD workstation. Um because that was considered you know technologically innovative at the time. And uh they would bill per hour on the CAD station. So um I I do see something similar maybe coming down to recoup some of the fees that we would, you know, pay time and material is gonna be a lot less because we're using this AI with one button click, we're doing a lot. But I think that one button click might be tied to some sort of uh credit or fee or token, as you would say.

SPEAKER_00:

Yeah, do you think that? I mean, I think you know, I I was talking to someone the other day on an enterprise level, and they said the the biggest challenge with AI deployment in the enterprise is some of it doesn't work out, that's okay. But some of this stuff does work out, and then you get a token bill. Yeah, and you're like, oh, okay, like is this, you know, let's start turning it off, right? Like, because people aren't optimizing maybe the AI work against the type of token or the type of LLM you should be using. If you start, I mean, I saw this with cloud, like a lot of people 15 years ago were building stuff on Amazon Stack, and then um they kept using a lot of the higher order features within the Amazon stack, and then it got expensive, and then the switching costs were too high because they built so much of a dependency on the Amazon feature stack that they couldn't switch to Google Cloud or Azure or whatever, and so I stopped funding companies that really at day one didn't have a multi-cloud strategy. So, should we be having like a multi-LLM strategy to optimize against costs? Because it feels like right now everything's free or cheap or whatever, but it's not. I mean, we know at some point, like they're gonna turn off the spout.

SPEAKER_02:

Yeah, charge us. Yeah. Um, so uh, you know, I don't tie myself into one in one into one LLM or one you know AI tool, really. Um, I try to pretty much explore between all of them. Um, you know, sometimes I'm Gemini, sometimes I'm in anthropic, sometimes I'm in open AI's platform. Um, I'll be honest, if I had to choose my favorite one, if I needed a favorite LLM, I guess it might be controversial as there's some people like the other ones, but I actually like OpenAI. Um I technically use a lot of it. Um, but for the most part, you know, I don't tie myself to one single cloud or one single, I almost said cloud provider now, but one single uh LLM.

SPEAKER_00:

Yeah, what do you think about? I mean, because I mean there, I mean, like someone sent me Agent Kit. Oh, okay, this is cool. But uh, I you know, I look back in the day, right? Like I started off my my first programming was like Fortran and Basic back in the day. And then really when I got into it was the web, right? Once you know, 1992, I was at Georgia Tech and started like building for the web, which meant there were no editor tools, right? It was text file. HTML is a text file, and you built websites from scratch, and then you saw in some of you, some of the kids listening might have to Google some of these terms, but you know, then stuff like Microsoft Front Page came out, and all these uh HTML WYSIWYG editors came out, right? And um, for us purists, it was like, oh, it's throwing all this bloat, it's throwing like it's throwing all this garbage on top of it, right? To something that was very simple, which is just markup text files, right? It's pretty straightforward stuff. Um so as purists kind of got mad at all these text editors, all these uh HTML editors that create a bloat. Um, and I was looking at Agent Kit and I was like, yeah, like I can do this anyway. Like I don't need open AI to give me Agent Kit. I can do it anyway from a purist perspective, right? But now I'm getting locked into their platform, right? Back to the cloud analogy. Now I'm getting locked into the open AI way of doing things, um, which means that I'm stuck, right? Like there's no undoing that. How do you think about using like the feature stack that these platforms are developing? I don't think a lot of people realize what's going on. That you know, you're locking yourself into a feature stack. Um, you know, whether you're using Amazon's you know, bedrock or what whatever, right? I mean, they all have some lock-in. How do you make decisions around, hey, like I know I'm gonna use this feature stack on OpenAI and I'm gonna be locked in. And who knows, well, you know, Sam Alton decides, like, hey, I gotta make more money, and he tweaks up your tokens a little bit, right? How do you make those decisions for thinking short term and long term?

SPEAKER_02:

Well, man, good question. Um, so I try not to lock myself into any one thing. So, I mean, although I'm messing around with agent kit on the side right now, I wouldn't lock myself into it right now. Um, just because I wouldn't want to be stuck with open AI. I do like the flexibility of being able to just flip a switch and go to one different LLM um and trying out the others. I don't want to be locked into that. So I tend to be, I don't know, I tend to be an N8N user when it comes to a lot of that stuff. Um, never use Zapier, but I know some people like Zapier. Um and I would I don't see them going away. I know a lot of people are worried, like, oh, what happens to N8N? What happens to Zapier? I'm like, well, I think they're still gonna be around for people who don't want to lock themselves into one platform, they want that ease of that usability of just bouncing around and using different things. But the how I would evaluate it from a company standpoint, because there is a little bit of ease of use that comes with locking yourself into one platform. I mean, we've all done it. I'm sure you're I'm sure we don't use a mixture of Google and Microsoft. You know, we're all either on the Microsoft email system or on the Google email system. So we're locked into Google or Microsoft. Switching is a pain to go between the two. Um, so when it comes to the user aspect of it, outside of the innovator aspect, um, I would say for the users, I would probably more more apt to lock them into one platform just so that they didn't have the sprawl of uh going between things. Um, but as a tech evangelist or whatever you want to call people nowadays, um I I just I can't be locked in. I I I I like to explore, I like to move around too much. Um and but I think for you know the user base, I think it's good to lock them into a uh a certain standard, and that way they're not apt to go try a bunch of different little apps every here and there. So you're kind of controlling it a little bit.

SPEAKER_00:

Yeah, my my second startup, we're all Linux open source, all Lamp stacked, um, until we scaled a little bit and we couldn't find enough Apache Linux developers until we had to flip into Microsoft because I could find a gazillion, I could throw a rock and find a Microsoft person, right? So I don't wasn't the better technology by any means, but for the business, um, that was the only people factor, that was the only way I could scale.

SPEAKER_02:

Yeah, the people factor always throws a wrench into everything because we as innovators we wanted to always try the latest technology. Um, but the people factor is always kind of what the that's the end goal, anyways, because adoption of tools is that is that next piece after you figure it out you got to get them to adopt the tool in the first place. And if you're all over the place and sending them to this side of the web and that side of the web, it just doesn't make for a very cohesive solution.

SPEAKER_00:

Very cool. What do you think about vibe coding?

SPEAKER_02:

Oh hard questions, man. Um, I thought you were just gonna riff. So vibe coding. So, you know what? The way I look at vibe coding is it opens up coding to um people who never would have ever thought about coding something themselves, and in that respect, I think it's amazing. I think it lays down a great base, even though some people we like all I do is correct vibe coders all day. Um, but I use a hybrid approach even now when I'm uh starting things. Um and I do push, I push some users to Replit and say, hey, just give me a proof of concept out of Replit. Um, what does Replit you know come up with? Um just I actually I think it empowers users in a sense to basically explore on their own. You don't have to explore through my through me going through encoding and creating concepts. You know what you want out of the piece of software. So I like to I like to throw them towards Replit and say, here, try it in Replit, let me know what you get, and I will help you from there. Um and I I don't look down on vibe coding, I think, like some people do. I actually I embrace it because I think it adds a lot of availability to create solutions to people who normally would have said, Oh man, I gotta go bug Brandon because I need this coded in Python. And I'll be like, and and usually I'm pretty good about it. I just I do it on my weekends because I love doing it. But um, for the most part, I think it empowers everybody else to be able like, hey, I don't have to bug Brandon. I can come up with this and then show Brandon my solution, and then I can help poke holes in it or enhance upon it and bring it forward a bit more.

SPEAKER_00:

So it's interesting, you know, we're doing this by Bahn in a couple weeks. Um giving away a pride, and and it was for me, it was like an experiment, like most things I do. Um, but I think we're gonna start doing them every quarter because I think it's it's interesting, it's fun, it gets more people under the tent, so to speak. But I have a friend that runs an offshore firm, and he was like, Yeah, like my my big business now has been fixing five code, right? And I was like, Oh, cool. So is it a growth or like is that a growth business? As more people start screwing things up and you don't fix, like um, you know, it's like being a kitchen contractor, let people go do it themselves and then come in and charge them a lot to fix it, right? Um, and he was like, no, it's actually going down. He's like, one, the tools are getting better, people are getting better at it. And he's actually been seeing some trends where there's a lot of good tools coming out for DevOps. So he he's actually kind of saying, I don't know if I'm gonna have a business, you know, at the end of all this. Um, which I think is fascinating, from that's a vendor's point of view. Like, hey, I'm worried about staying in business is it is a strong signal versus a novice coder saying, Oh, I don't need developers anymore, right? Um, but I think, you know, one thing that's kind of come out of this is that, you know, from your perspective, you're you're encouraging people and all that. How do you think about guardrails? How do you think about putting things into production? Um, because then all of a sudden now you have this bottle, like you become the bottleneck if you have to like review everything, you know, for cybersecurity, you know, uh, someone live coded something for me. I was like, you know, the login doesn't actually work. I can just like hit enter and it goes right in. They're like, oh wow, how did that happen? I was like, well, you know, like, you know, you gotta worry about that. I mean, how do you think about like how does that increase like your workload of having to review things and put things into production?

SPEAKER_02:

Well, I'll be honest, it hasn't affected it yet. I haven't had a lot of people jump on making their own proof of concepts or they've just started. Um, but the way I look at it is you know, guardrails from a perspective, I would say I'm pretty strong on cybersecurity. Um, that's always been my thing. I always make sure any tools we're vetting are SOC2 and all that other fun jazz. But um I don't really want to put too many guardrails on it because it also gets rid of some of the creativity. I'll be my I'll be honest, my only worry is people are gonna be vibe coding instead of doing engineering. Um so um because they're gonna have this cool idea, and instead of working on production, they're gonna be working on you know some cool idea, which you know it's one of those things I do outside of work. Um but it's that's that's honestly my biggest worry about it right now. But I could see how the bottleneck would appear and start to happen where you know all of a sudden Brandon is vetting all this new stuff that's kind of flowing through. Um, so I think that next piece will really be you know coming up with you know some way of just, I mean, we have something called RTM Labs, which is technically a filter for all this stuff. Um, it's not really used too much right now. I'm trying to get it out, but I feel like RTM Labs will be a great filter to see all the proof of concepts come in and being able to prioritize because really what it comes down to when it comes to prioritizing and putting the guardrails on is you know, how is this gonna move the needle at the company? How is this going to help production in a way that's positive and not negative? Um, you know, how are we gonna go net positive with this tool? Um, get more time back in the day, essentially. And that's and that's really my my main judge. And then I'll look at it from a cybersecurity standpoint at the end and say, okay, listen, you're logging, I can log in with the enter button. That I don't that's awesome. But um, you know, and then one cool thing is we had with this is I actually had an intern from a college come here at RTM and I didn't really have anything from the work on. They were only supposed to be here for two weeks, so I kind of gave him what I called a uh a research paper, a research development uh task to create an RFP tool. I gave him nothing. Um, I didn't even tell him what tool. He actually I said cursor, I think. I think I gave him cursor, but and uh what I actually got out of it after the two weeks was an actual fully function RFP tool, and I might add with a working login that tied in. So it's definitely getting better.

SPEAKER_00:

What do you what do you think about? I mean, you know, you know what I do for a living, I see all these startups. I kid you not, in the last four weeks, I've seen 50 RF AI RFP tools, and and I find it fascinating because these startups are trying to raise money, and I'm like, you literally built something in the last month, and there's been some that have been funded that I passed on, um, because I just don't I don't see getting anywhere, right? I I kind of see it like you know, you're not gonna dominate anything, you're not gonna be a billion-dollar company. Um so it's interesting that you bring that up, that your intern-built stuff. I mean, I'm sure you get hit up by startups. I mean, are you seeing anything interesting out there that you couldn't do yourself?

SPEAKER_02:

Oh, that I couldn't do myself. I'll be honest, not really. There's stuff that I I don't want to do myself. I guess that's part of the better question. Um you know, the hard part being, you know, we don't have a ton of uh coding power here, you know, it's just me and we have two other coders. Um is what what projects do we take on to code ourselves and what do we go out and buy? Um, one of the things that couldn't have been more relevant is when the intern, you know, had to leave for the summer and he had coded on his own, this entire RFP tool. And then one of us, and one of luckily one of my coders just jumped out of nowhere and said he wanted to look at it. Um, so he learned it. But you know, learning somebody else's code is it could be kind of a tedious task at times, um, especially when you have to edit it or make a change to it or add something. So um I I always look at tools two ways. Yes, I can build it myself, but is it worth my time to build? Um, you know, we always use this the term here best and highest use of our time. Um, and we and that's kind of how we've been is it the best and highest use of my time to code this myself? Should I give it to somebody else at the company to code, or is it something that we should go out and buy because it already exists? Um, and honestly, I I tell you the one downside of buying something that already exists is you're stuck with it. That's it, that's the tool. So if you wanted to expand upon it, you had to make it yourself. Um, and and that's honestly how most of the tools I've coded have come about. Um civil site. So I created something called Civil Site Designer about four years ago, um, before a lot of the AI buzz was around, and it's kind of been evolved since. I keep adding more and more AI tools into it. And I know I say I'm not gonna plug anything, but this would be the one thing I plug, I guess. Um, but the one thing that I kind of kept coming around to is hey, I don't like the way that this other product works, so I'm gonna make my own. And that's kind of what drove me to create my own kind of conceptual design plan designer AI tool. And so it there it's not without, hey, I might buy it, use it, and go, man, I can make something better, and then scrap it later and create your own.

SPEAKER_00:

Yeah, I'm ahead a little bit of a different direction. Okay. So if you uh decide tomorrow to go start your own firm, who would your first five employees be? Like what would what would functionally what would they be?

SPEAKER_02:

Functionally, what would they be? Man. Oh man. Well, I guess uh am I starting my own engineering firm or am I starting a uh AI yeah, you're on civil engineering. Civil engineering. Okay. Um so honestly, I would start with a uh I would get one CAD guru or CAD manager, um, and I'd probably get a bunch of coders.

SPEAKER_00:

Okay.

SPEAKER_02:

And the reason I would do that is the CAD manager could help make sure that CAD standards, um, on this I'm gonna do to say a civil engineering company. That the civil engineering company would run with a certain level of CAD standards that look good, and the this way we're producing things that look correct. And then I would hire just a bunch of coders and I would sign and seal everything myself.

SPEAKER_00:

And okay, so if that if that's your ideal start state, right? You're working at an engineering firm right now that has, I would say, uh lots of artifacts and background and history, right? Not a blank sheet of paper. How how do you think you work? Because I know you work closely with the executive, with the you know the CEO and all. And how do you guys like contemplate like that as a potential end state to the business, right? That you know, because you know, I've gotten really sour on this word change management, specifically because of AI. And I'm like, it's really more like change journey, like change management implies you know what the end state is, and I don't think anyone knows what the end state is, right? Um, but there's just so much effort. The consultants all show up to help you with your change management process, right? I'm like, change to what? Like, we don't know what the end state is, but if you took that as like your end state in a way, right? Because that's your starting state. Do you think it's possible to like take a firm that's already been in existence for a long time and kind of digitally transform them to look more like that?

SPEAKER_02:

Uh, I think it'd be difficult, um, to say the least. I mean, you got a lot of people that have been doing things a certain way. And again, you're fighting people and you're fighting feelings. Um, two of the hardest things to go against, um, especially feelings is the number one thing. Um, but I think if we look at it from a perspective of do engineers have the knowledge through vibe coding or the ability to learn Python, which both of them already had a Python class, or depending on where they came out of C sharp or C. Um, so I think they could easily learn the coding aspect of it. Um I I think, you know, outside of five five-person example we used, I think the next the next portion of the company as it would grows is really you gotta you gotta grow that sales muscle as well, which engineers typically were not the most uh you know extroverted people, um, to say the least. Um we're mostly you know, you I would just like to be in a dark office all day and leave me alone. I mean, sometimes that's my attitude. Um, but uh so I honestly I think it could be done. I think you could make the shift. I think there's gonna be a lot of feelings around that. So I think it also you also have to make sure not only are you going to change the way we do things, but you're also going to affect culture a little bit. Um, and culture is kind of one of the big things a lot of us hang our hat on is we have a great culture at the company. Um, and I think that would be the next biggest shift is you're gonna change the culture just a bit. Um, and kind of maintaining that, I think, is key because that's why people come to work every day is you know, you come to work because the culture's great. I love working with I work with. Um yeah, it'd be it'd be hard, but I think it can be done. I I mean I think you just got to get engineers who you're gonna honestly, you're gonna get engineers who love coding and you're gonna they're gonna love the empowerment that comes with being able to code and make their own things and they're getting to affect all these cool tools. And you're gonna get other engineers who are A, you know, they're just they don't want to do anything other than engineering and they're senior-level engineers, but they're gonna be great because they're gonna be able to add that knowledge into the LLM and get that knowledge over into this new tool that we're going to need to be able to scale the business in that way. And then you have other engineers who are just gonna love to go sales, or maybe they don't know they love sales yet, um, but they love project management, they love clients, um, you know, things like that. Are the I think they're just gonna reach out, so it it'd be a huge lift, especially a well-established company.

SPEAKER_00:

Yeah, it was interesting. I was talking to a uh a younger, you know, seven years out, so had their PE and everything about sales, and it turns out he's like, he tells me this thing. He says, Hey, I think I'm a mediocre engineer, but I really enjoy sales, but my company won't let me like because I'm not a principal, I'm not a part. I thought it was fascinating. I was like, Well, are your part are the partners and the principals really good at sales? He's like, No. I'm like, why it was it was an interesting dynamic because I do think we have a generational shift where we have to kind of break some of our linear thinking around career paths. You know, if someone comes in and two years into civil engineering, it turns out they're not great at the details and they don't operate that way, but they do have the uh ability to go out and do sales or something. Like we should like, you know, I don't know, we've got to rethink some of these career pathing ideas that we, you know, I think that's a huge thing if we want to continue to attract and keep talent.

SPEAKER_02:

Kind of like a choose your own adventure career path.

SPEAKER_00:

Yeah, 100%. I mean, some people do have, I mean, uh, I run into civil engineers all the time that are in sales roles. He's like, How'd you end up in sales? Civil engineering was driving me insane. Like, I just I wasn't that great at it, I didn't enjoy it. I wanted to spend more time with people, and my boss told me I had to like sit in a cube farm for the next five years of my life before I could uh you know actually be recognized as a human. Go sit in that cube farm and do civil 3D all day, right?

SPEAKER_02:

Like, yeah. I mean, and that's what it comes down to a lot. I think as engineers, we all hold ourselves to a certain standard, you know. You're told to hold yourself a certain standard, and that's just you don't want to break out of that. Um, so if I'm not doing engineering and I'm doing sales, you know, I mean, I I'm a very big proponent of having a few different career paths. Um, because I think there's a people who just love being in civil 3D. I'm one of them. I I can just lose myself in civil 3D and relax. I'm at home in it, and I can just kind of have a blast all day. Like I don't even know I'm working. I could be here 12 hours just in civil 3D drafting and designing. And then there's other people who just can't stand it, but they'd rather be on the sales side. So I think as a company, and I we do this really well here, is we embrace those differences and we put people again, you know, where are you passionate about and where's the best and highest use of your time? Obviously, unfortunately for me, the best and highest use of my time is no longer in civil 3D, even though I do still love to do it. Yeah, um, but I think just kind of positioning people into the right places. I I want to say RTM does a great job at that. At least I feel like we do, because I hear nothing but great things about kind of it. There's different people though. I mean, here's the thing you can do a choose your own adventure career path, right? And you're gonna have people who thrive in that model. There's gonna be people who love that model, like, hey, I'm not tied down to one thing, I can try over here, go over here, go over here. But there's other people who actually do want to see that career path, like, hey, I'm gonna be civil engineer one, two, three, four, and I'm gonna work my way through this set piece. So it's also a little bit about uh people in that sense, where you know, some of us really do want to have it all mapped out for us, um, and we're fine with that. But for those people who like to choose your own adventure model, I think you can go really far really quickly and get into where you're passionate about fast.

SPEAKER_00:

Very cool. Um any thoughts about other topics since I've been asking all the questions.

SPEAKER_02:

No, I mean, honestly, I would ask you, you know, what do you think about it's all the buzzword right now, and I hate using buzzwords, but agentic AI, you know, we're all talking about it. We're all talking about how we're gonna implement it into our processes. Uh, where do you see it? You know, where do you see it making the biggest impact right now?

SPEAKER_00:

I I think that there's this, you know, what what I really hope to solve for in this industry is that I think we've gone through eras of trying to use technology as a way to beat in a process. I just don't think it works, right? I think the reality is we're highly decentralized. It's not a workflow, it's all the workflows, right? And even within a certain firm, I'm sure within your firm, depending on which engineer is doing the project, they approach it probably with a different workflow, right? Um, and so I think that the challenge with some of the agentic stuff and the opportunity is if you try to apply agentic methods to a rigid workflow, I think you fail. Right? I think you fail. So I think you really have to think about this stuff as a decentralized manner. And um, I think everyone's trying to just like create workflows and charts and apply it to agentic AI, which I think that rigid approach will lead to failure because I think what you'll see is um near-term gains, but long-term, it's like, well, isn't this just what we were doing? Like I look at like customer service, right? The biggest application of agentic AI is this idea of using it for escalation to a person, right? So, okay, I'm gonna triage you. Funny enough, I was talking to someone that said, Hey, I'm actually putting people on the front end and putting agents on the back end, which I thought was like fast. Okay, tell me more. I was just like, tell me more, right? I gotta I have to understand this, right? And the point was that people get in customer service, people get really frustrated with dealing with a robot that doesn't understand what's going on. So what they figured out was if I put a person up front that defines, like, okay, so here's what you're at, here's what's going on, okay, great. They can send you down a rules path and let the road like let the agent take it from there, which I thought was like, oh my God, somebody's like trying to think uh differently about this stuff. So I think the the problem is if we just take agents and apply it to here's how we do things, like I said, my customer service perspective, you know, escalating um a problem, I think it's I think the outcome's not gonna be as interesting. I mean, I think it'll work, you know, at some point. But you know, this firm I talked to, it's like, no, we're putting people up front. And I was like, really? They're like, yeah, because you know, and I think you know, these guys are like, let's just say a telecom business, right? They were saying that like once people decide what they want, then let the agent run with it. But at the initial, like, should I sign up for this plant or that plan? What are the pros and what are the cons? Sometimes people want to have that human touch like to be guided through it, right? I think which is generally the case, except for everything, except for buying a car. I don't think nobody ever wants to deal with a person ever again buying a car. But um, and that's a function of the people you have to deal with, right? But I do think there's some I I think we we just can't keep applying technology against existing workflows and somehow think it's gonna be a better outcome.

SPEAKER_02:

Yeah, I definitely agree on that. Yeah, because and that's where you know, thinking outside of the whole, that's why I think new new people bring new perspectives into some of these things, because a lot of times we get stuck in that as engineers. Well, this is the workflow, so I'm just gonna automate that workflow. But is it the best workflow? Are we doing it right in the first place? What if we tried this? So I always like to ask what if and see where it takes me, essentially. Basically, I start with like a notepad, and I like what if this, what if that? And I keep asking better questions. Um, and essentially I get to one single question where I can't narrow it down anymore. And I'm like, all right, now let's try this. I've gone through the all the what ifs, and now let's see what happens with it. And that's kind of how a few of the tools have started. It just starts with a bunch of it starts on a legal paper asking better questions.

SPEAKER_00:

Yeah, I agree. Well, uh, it's been great chatting. Just one last question. Any favorite, not development tools, but like call let's call it power user tools that are tickling your fancy these days.

SPEAKER_02:

Power user tools? Honestly, I've been kind of lacking inspiration in that uh field lately. There hasn't been anything that I've been like, wow, this is a great power user tool. This is an amazing tool. Um, what I will say is what I've been liking is just how fast AI is moving. I can't keep up with the tools anymore. Um, so just keeping up with the tools in general. Um, you know, messing with agent kit on open AI, that's been pretty fun. Um, NAN just honestly, I've been automating my entire house. My wife probably hates it. Um, but uh, I don't even know how adjust the temperature anymore in this place. Like you just say it. Um but uh um so and as far as new tools, I've been actually lacking a little bit of inspiration in there. I haven't really had anything.

SPEAKER_00:

Yeah, so I've been playing, I've been down the rabbit hole with the tool called Eraser.

unknown:

Okay.

SPEAKER_00:

Have you played with that?

SPEAKER_02:

I have not played with that one yet.

SPEAKER_00:

I like it because it's a it's it's a um it's a charting tool, it's a process diagram, process flow, including like first if you want to do software architecture stuff, it's just a lot of fun. It's just a lot of fun, and and I kind of I kind of view it as like the future of where CAD may be going, um, because they're they're they're definitely doing the prompt engineering against a visual outcome. It's been a lot, it's it's a lot, it's a rabbit hole because you start right.

SPEAKER_02:

I will give you one that I can't play with it yet because it's not out yet. Um, but I had the uh pleasure of going to Autodesk University this year, and they were showing off some of the new Forma, their Autodesk MCP and their Autodesk Assistant. And I I am kind of excited to kind of see what I what is possible with what they got going on. Obviously, we can see what they got going on with Forma inside of Fusion and a centralized data model that they've now switched to with hub and ACC data coming together into one place. I think they're calling it the hub. Um, so I am kind of excited to see where that goes and what is possible with it and how far I'm allowed to push it outside of what they want it to do.

SPEAKER_00:

Yeah, it's really interesting. You know, a lot of my startups come to me like, oh, Auto Desk will never do this, Procor will never do this. And I'm like, what you're missing is it's not even so much a business thing, right? These companies have software engineers. And generally they like lack inspiration because it's like, okay, let's make Revit whatever and upgrade it to the next thing. And there's like a lot of boring product managers telling them what features need to add, right? But what's really happened is CEOs showed up and said, Hey, what are we doing about AI? And so I think it's inspired a bunch, you know, everybody thinks like, oh, software engineer at Autodesk, they must be horrible. They're not horrible, they just make an amazing living. Um, and they follow what's the product roadmap, right? But I think what's happened in some of these companies, and I've talked to some of these software engineers, they are so inspired right now. Because management's saying, hey, with AI, go figure something out, just go do whatever, right? Let's figure it out, let's experiment. And they're they've really been re-engaged by leadership versus saying, I mean, I mean, what was the next version of Revit or Civil 3D going to be without AI? Oh, let's move the button from here to there. Everybody's saying the button's in the wrong place, let's make it blue instead of green. You know, I mean, that iteration, um, I think it's it's boring, right? And I think foundationally, like if you can say, hey, because I saw this happen during the web too, right? Like when people were moving from client server to the web, first they did lipstick on a pig, still engaging. But when they said, hey, no, we need to do a whole cloud-based rebuild, people got very excited, right? People and you and you saw some inertia there. And then we got bored again. So I think um I'm usually fairly negative about the incumbents, but I think that it's that they still have people working there, and I can't imagine a software engineer anywhere in the world that's saying, I don't want to do, I don't want to do anything with AI. You know, I just want to move buttons around. I don't I don't think there's a software engineer out there that's thinking that way.

SPEAKER_02:

So do you do you think companies are gonna start like again? Obviously, this is something that we've thought about at RTM, um, starting to sell our own engineering software essentially, or companies that are just gonna revolve around, they're gonna change from actually doing engineering to selling engineering software.

SPEAKER_00:

I I think there's an opportunity, right? I think there's an opportunity because um I think there's so many edge cases in our industry, you know, um, where a product out of the box or whatever just isn't a great like I'm seeing a lot of software being built for the data center business right now. Data center boom. It's worth me investing and building some software to optimize how I design and build a data center, and there's a ton of work. Um, whereas I don't see how Autodesk comes up with you know their data center focused product, it might be hard for them to rationalize. Um, but I'm I am seeing some of those ideas around very narrow, like I mean, you know, data centers are narrow, but it's not a small market. Um, you know, I don't think like in the world of wastewater, is there a great wastewater treatment plant design tool? You know, probably not because the companies like Bentley have had to build product that serves a lot of different markets, right? So I think I could see that happening.

SPEAKER_02:

Yeah, I've been trying to work on a uh surgical suite tool that designs because they're very configurable into how the surgeons want them. So I've been working on a tool to optimize how to create these surgical suites within Revit. But uh still working on it. So nothing debates yet.

SPEAKER_00:

All right, man, it was great chatting. Cool. Hopefully, I'll see you around see you around the hoop soon. So uh sounds good. All right, I'll talk to you.

SPEAKER_02:

All right, talk to you.

SPEAKER_00:

Bye.