AI Augmented Drawing Review in AECO Industry – Ep 101

Do your AEC leaders have the people and project management skills to truly lead? At EMI, our leadership training and AEC project management certification combine real-world tools on demand lessons and live PDH webinars, so your team actually applies what they learn. From confident communication to stronger delegation, our programs are built for AEC professionals. Visit AECPM.com or call 800-920-4007 to learn more. That's AECPM.com or call us at 800-920-4007 because strong leaders build successful organizations. According to recent studies, AI could automate up to 40% of creative tasks in design industries, like drafting, drawing interpretation, and design review within the next decade. That's not just automation, so shift to how we design, build, and collaborate in the AECOs face. And so I'm joined by Luke Reef, principal solutions architect at Twin Knowledge. To speak about how AI is being trained to understand complex construction drawings and how that's transforming design and reviews we'd have it. We'd have in the real-world use cases, the rise in AI-augmented workflows, and what the future holds with AI agents supporting design teams. Curious about how AI is reshaping the tools and processes we rely on every day? This is a conversation you will not want to miss. Before we jump in, I want to tell you about our AECPM Connect event. Built for project managers who want to improve project delivery, communication, and leadership, depractful sessions, and tools. These events provide an avenue to connect with leading PMs in the industry. Our next event will be on June 10th, 2026 at the Forestgate Country Club in New Jersey. Visit AECPMconnect.com and register today. With that, let's jump into today's episode. Okay, it's now time for our conversation of the week with Luke. Luke, welcome to the show. Thank you so much for taking the time and joining us today. Absolutely. Thanks for having me. Excited to be here. Yep, and I think the scope of what you guys do is extremely interesting to me, and I know it's going to be the same for men in the audience, so I'm gonna cut as I just just jump right in. So, Luke, can you start by sharing a little bit more about yourself and then what led you to your current role, which is Principal Solutions Architects at Twin Knowledge, and what generally sparked your interest in AI with an AECO space? Yeah, happy to. So, let's begin, Luke Reef, Principal Solutions Architects at Twin Knowledge. So, I used to be a structural engineer, actually. So, growing up, actually in high school, I used to go to this camp every summer. We roof houses in Anacity Memphis for those who just needed housing. Roof, to be honest, we just couldn't afford it on their own. So, I started early in the construction space. Like, literally doing the work. I'm not gonna say we were great. We frank with you, we were high schoolers, but we do what we could. And I'm going to college, and I'm thinking, you know, what I want to do with my life. And I'm thinking, well, you know, I'd always experience in construction in high school. I like to what I did. I love math. I love physics engineering. So, I went and studied structural engineering. Undergrad at Dr. Tech. So, ended up really falling in love with the physics. I thought about majoring in physics at one point. I'm just like a nerd. I like that. I love the math, love the physics. Turned out I was great at school. I really enjoyed it. So, I wouldn't get a master's as well, because I got it paid for by a researcher, which was really, really nice. And then I got to industry. And what I thought was true, which is, you know, the physics and the math super cool. But you get to industry, and you're doing a lot of plug and numbers into softwares. And I just can't let the software do the heavy lifting, which it should, right? There's some really complex buildings we were designing. But I ended up doing a lot less engineering work and a lot more like workflow optimization. So repetitive things we kept doing calculations or tasks. I just didn't like having to do a manually so often. So I just took it on my own to start, you know, whether it was in Excel with macros, or I did it with a dynamo coding for the first time in dynamo and refits. Which I thought was crazy. I was like, you know, this is crazy. Then I got to grad school. And I said, I'd do a Python and all kind of stuff. And it's funny looking back at thinking how, how Cody I thought I was and I didn't know anything. But spent a lot more time doing that when I was an industry and enjoying that. And a lot less enjoyment on some of the work I was doing. And actually the work I was doing at the time is a lot of the stuff we're automating. Some of the stuff we're automating now with twin knowledge. So it's kind of full circle me being an engineer spending two weeks comparing our steel drawings to fabricators drawings. Two weeks of time I was doing this. And I thought, wow, like, I don't really what I'd be doing this. I think something else, someone else something else could be doing this. So we're back to grad school for three years. I really studied, essentially a applied AI and product development. So it was kind of like the thesis of what I did. I actually did a thesis on how LLIMPS can extend our minds and to helping us remember and to perform our cognitive analysis. Let's say on what we know and what our extended minds, our computers, our databases know. And that was my thesis. And then I come out and I met Ivan our founder. I'm actually while in grad school. He was an adjunct. He came and I guess spoke to my cohort in grad school. We met. I came out. He had just come out and stuff and just him. So we talked and I was like, hey, I had this great background of the engineering and construction. And then my studies in applied AI and a bit of product management. And I love this, this nexus. Like, see what you do is I'm really cool with this nexus. Like, what if we, we joined forces. So I came on right after grad school. And it's been almost two years now. We're still young. I mean, it's crazy in life of a startup that's still pretty young. But yeah, that's that's a bit of my background and how I get to AI in construction through the pathway of structural engineering. And you, I think you mentioned someone that, right? So my background also in structural engineering. And I too would have spent two weeks just sitting and paging through steel shops and looking at our design set and just mind, not mingly figuring it out. I mean, it's great for your development as an engineer, right? You actually get to figure out essentially how a steel building goes together in theory, of course, when you get to get out on the site and see how it's done in practice. That was a huge, for my mind, when I went out and saw on the field of things, you don't consider in an ideal environment in the office. I was like, wow, constructibility. Huge thing you have to consider. But yeah, so anyways, great having that. Those lessons in the field before I came in and joined the tech side. And then you see. It's at that point, right? Like, because so much of the workflow optimization and structural engineering office revolves around numbers, right? Whether you do an Excel or Python or whatever your tool of choice is. But what really, you know, stood out to me when we first met is that drawing piece, which that leads me into our next question. So how do you train AI to understand complex drawing sets? And then what role does data quality and consistency play in ensuring accurate and effective AI augmented drawing for you, right? Because that's not like, like, you don't just pop open an Excel and start like automated and drawing related processes like you do structural engineering calculations. Yeah, as like as like a practicing engineer. Yeah, absolutely. A great question, Nick. So how are we start up kind of talking about this is drawing sets are the language of the industry. And that's a new language that's artificial intelligence has to learn, right? And artificial intelligence is a really broad bucket. And so two, two or three, I would say sub-buckets in that is how we're using or what we're using to train AI to understand drawing sets. The big one is computer vision, right? So we were mostly with 2D PDS. We can work with CAD drawings, BIM, specs as well. But most of our clients around the world, let's say, it's still the interest standard. And the day design gets onto a 2D drawing set, usually digitally PDF and believe me or something like that could be a pronoun as well, right? Or a scan. So teaching AI to understand that visual language, computer vision is where that comes in, right? So we've trained both a general model, but then also we do specifically for bigger projects or clients. We fine-tune that model. And that's going to be the AI's understanding of the basic elements. So what's a title block, right? Where's what's a detail? What's a detail title, right? It's the circle with the number in it and then the name. And it's a scale as you attached. Or it could be a boxed, right? And there's a square that the numbers in. Like all these things that we take for granted, these patterns that we know are very much the same thing. AI doesn't know that, right? So we've got to teach each of these instances of that pattern. And that kind of gets a bit into the data quality, right? And consistency. So we rely on computer vision first to understand that language and be able to then extract, right? And work of that language. And then we work with retrieval augmented generation. That's a typical very key piece these days. And these question answering, let's say, modes of AI, and how it interacts with construction data. And then of course, there's a bit of fine-tuning that you can employ for individual large language models, right? So that first piece is allowing AI to access the visual information through Channing a computer vision model. The second is, okay, how can those individual then, let's say, details, title blocks, tables be understood? And that's where multi-models come in. But can they understand them well, right? How are they understand that a schedule is actually a table? In our terminology, right? That's a great example. Because in everyone else's terminology, well, it means like days, weeks, months, right? But of course, in ours, it's like, no, like Gacha. So that's where a bit of fine-tuning can come into play with multi-model models and large language models. So we do a bit of fine-tuning on our language in the industry more general. And that's where once again, data, consistency, and actually comes into play. So we see with a lot of our clients, Tobras is a client of ours, for instance. They do design build. They'll cost something one thing. Their consulting will call another thing. Their contract will call another thing, right? So the model's got to understand, oh, when these three things are said, it actually means the same thing. So I'm looking for information related to this. Make sure I also know that this word and this word are basically the same thing, right? And I shall also find. And that's something we, as experienced professionals, just do it pleasantly, right? Exactly. Yeah, great point there, Dicken. This is, I'm a nerd, so I can get into this way too deep, and I hope it won't. But yeah, like all these things that we do it pleasantly, it's just pattern recognition. So at the end of the day, artificial intelligence is a subset of machine learning, which is simply just pattern recognition, right? And the less complex your data is, the more consistent it is, the easier it is to recognize that pattern, right? The more complex, plumplex your data is, the less consistent it is, the more complex that relationship or those relationships can be. So to the point on consistency of data and accuracy, we want to get as close to structured data as we can at a much structured data, right? Unstructured data has a lot of implicit relationships. We don't, you know, we implicitly know exactly how language goes together, like when to use the, and not to use it before a grammar involved, right? But that's implicit. We just learn we were trained over years of growing up and speaking the language and reading the language, right? But it is very, very much implicit the entire rule book, right? We don't actually, I couldn't tell you right now what, you know, what grammaticals exist. I can just tell you that I follow them generally, but I have a lot of Southernisms. And it comes from the South, so that's. And I've heard it's like, if you get like these analogies to artificial intelligence, I mean, well, I think one that's really obvious is like, you know, truly like a junior engineer. Another one which I heard recently, which I actually really liked is like a non-native speaker of whatever language you have, right? Because like they learn it super-formally and they haven't grown up with it. And like all those implicit relationships and equivalentisms that we just take for granted. Absolutely. So I guess gradinality, the more simple you can be with the language you use, right? And the more explicit it is than those relationships are, than each is going to be for that non-native speaker to understand, right? Same with, as we're training AI to understand drawing sets. Strong project outcomes need skilled managers and confident leaders. EMI's corporate training delivers expert led courses in project management and people leadership. Plus, industry recognized certification and exclusive premium content to take your team even further. Elevate your team's leadership capabilities today. Visit engineeringmanagementinstitute.org and click on corporate training. And then, yeah, it's a great segue into the next question, Luke. So what are some of the most promising use cases of AI in the AECO industry? And then what common challenges do firms face when actually implementing these solutions? Yes, so good question there. As we work with our clients, especially getting to pilots with new clients, we're focused on highly repetitive, highly manual things, right? So talking about a first place to enter. And that goes back to what we were discussing about reviewing our design drawings as structural engineers against a fabricator's drawings, right? There's some middle in their shop drawings, highly repetitive, highly manual, right? And I won't say that's unskilled because it's not but a huge part of that's unskilled, which is just connecting, okay, this detail and art is drawings. It should connect to this detail in the fabricator's drawings, right? They should align. It's that connection piece, right? It's connecting all this information across stakeholders in order to do review or to produce some kind of outcome on our project, right? And we focus a lot on review with Twitter, not just really our bread and butter. So we're trying to really help the entire review process. And that goes, that could be the very beginning. That could be obviously the design of pre-construction. And then there's even a bit in construction and then a lot post-construction as built reviews, things like that as operators get ready for O&M, right? So yeah, anything highly repetitive, highly manual is really like a great place for anyone to start in the industry. So some other things we see is going to be related to that connection piece I talked about. There's so much information on a construction project and it was across stakeholders and across repositories, right? And a huge part of what we do and there are studies around this, we have it in our decks, we show. Wasted time finding information, aligning information, right? And then if you didn't find something, you let something out, or something, was misaligned. Just simply, you know, information being misaligned or not found, errors that happened during construction, costs that happened during the construction. It's pretty high. So connecting information that's needed to do the work we do is also a great start. We're already seeing that with a lot of players in the space. And we're doing that, it's fundamental to anything that we do, right? So once again, understanding a drawing set, understanding that these specifications relate to this detail or this plan view or this schedule, right? It's a huge game I say of like a telephone is essentially what our industry is, right? Hey, this is what it should be. Okay, this is what I heard it should be. Now here's what the design drawings are. Okay, did I get this right? Did I interpret that correctly, right? And so you got this massive web of design drivers and then design artifacts. And those artifacts on the 2D PDF drawings, they have to comply and align with all of those design drivers in this specification or that document or this document, you know, and that can live, they can live all over the place. So I really see a huge productivity gain our industry, silly from from AI, connecting all that, right? And I know it seems like a simple use and it is, but the most practical simple uses are where Indian issue starts with any new technology and AI is no exception to that, right? And then that gets into, once again, as I mentioned, what we do a lot of and that's like non-compliance, misalignment, detection. So if you can connect the information for design drivers with the actual design, then you can begin to do some analysis, right? And that's where large language models, multimodal models come into play. They have reasoning, mimicking abilities, right? So they can say, okay, I have some specs here. I realize that these five specs should be driving this detail. Okay, we'll do we align, right? And that's where that reasoning comes into play. So with that ability to connect, we can then tap in into some of the reasoning abilities of our language models, multimodal models, to actually do that reasoning and say, okay, I think we comply or I don't think we comply, or I don't know how we talk about it with our users. We really have phrased it kind of like as a first pass review, right? To your point, Euler Nick, I'm a junior engineer. That's how we kind of push our AI agents. They're intelligent, they're quick. They have great memory, high processing speeds, but they're learning. And the more you work within the more they learn, and so treat them as a junior engineer, right? Let them do that first pass review, let them make the connections do the analysis, but there always needs to be somebody who's been more in the industry for longer, who knows more the language, right? Knows more the things to look out for. That's maybe they may be implicit, right? As well as explicit information, it's maybe a little more implicit. Well, I just know it. I've been in the industry, I just know these things. So yeah, I think it's really that arena where I see AI coming in and really right now revolutionizing how we do our work. And especially review, right? And that leads to better construction, less RFIs, less risk, less litigation. If we can get all those problems and solve them further up that timeline and pre-construction, we can really avoid a lot of heartache risk and money in that construction phase. And to expand that analogy, it's like your senior engineers or your senior people are so valuable because they have that hard-fought experience that is not so easy to find in these training sets of data that AI then just ingest, right? I mean, I'm sure we could go down the rabbit hole and think of very niche and complex ways that very unique structures are reviewed and that are done in like one or two offices across the country. You're just, and that knowledge just as it captured accepts and like the head of a couple of very senior engineers, right? Exactly. Yeah, and I was going to say like it's just, it's just not something that it may just not be something that just doesn't exist in like a data set that that an AI could be trained on which is why that experience is so valuable. Exactly, exactly. And I could get that example here. So to your point, I mean, a lot of big firms who just want off just to complex and structurally unique conditions, right? You don't see this up really anywhere else. So the considerations of lessons learned, it's going to be one example, right? One set of lessons learned. That's not a lot for an AI model learn from. Conversely, I told brothers once again, a really great partners of ours. They do 10,000 homes a year. They do a lot of townhomes. They have a set of details that they love, that they want to see, you know, followed mostly to the T, no matter where in the country that they build, right? Because they do a lot of repetitive same type buildings, right? So then we can use AI and go, okay, we have all these examples, right, of designs across all these townhomes and history, right? And then we have what we'd like from that. I agree with these like gold standard details. Now, using that knowledge in these details, go make sure that they're followed on every summer project going forward. And you know, it's a very summer design. You know, I have a detail for a windowsill at Woodtrim because they're very, once again, very summer designs. And for this, for artificial intelligence, as like a first entry into our, into our industry, is a great use. So, Lord, how would you define augmented drawing review? I mean, how does it differ from traditional review processes that you'll see across the industry? Yeah, good question. I like that the phrase "use" was as augmented. And when we talk about our product, we use augmented and automated, depending on different levels of, I would say, I have functionality that a certain product might deploy. So, augmented to your question. Really, for me, I'm once given the lens of twinage and what we do in my experience in industry. Augmented is almost like an assistance that will go find you what you need. And then, like, you in the end, do what you do best. And for a designer, that's a design. Like, incredibly creative minds, beautiful minds to design. We'd like them to be designing more. Symbite engineers, let's be solving the hard, complex, like engineering task challenges, right, and come up with better ways to do things. Beautiful ways to frame things, right? There are a thousand when I got to industry, but as there were a thousand ways to frame a building, even like a two-store, simple building, right? There's so many different ways, depending on what outcome you want. So, that's the fun stuff, right? And that's why I got into engineering. So, I think, augmented, and how, at least as an engineer, I would like it to be instituted in the industry. It's that, like, what are these repetitive, manual, mundane, maybe more project management things? I don't want to project management to a huge deal. So, when I use that, I maybe mean more, but I was referring to earlier, finding information, collecting, centralizing, to then do your engineering work, to the new year design work, right? So, I think, augmented, I think, more and more along those lines. If we get into automated a bit more, that's that first pass review that I think of, right? This assistant agent has gone and found you what you need, right? The relevant information across documents, across repositories, across stakeholders. And now it's going to do a first pass of what the objective is, right? Whether it's a standard detail that needs to be followed, or it's spec items that need to make sure they reflected in the designs, right? It's that first pass analysis. So, when I think of augmented, first automated, that's kind of what I think about. And then, how would you say, like, like, if just in very simple terms for, you know, anyone who maybe has it used this type of technology before? How does that differ? Then, like, what were you still, like a traditional or a traditional drawing review process? So, yeah, once I think about that, there's two weeks I was sitting there comparing our steel drawings against our fabricators drawings. And I was literally just going through our details one by one, and then going to find them manually. And this is, I guess, on computer screens, so we get two computer screens there. Is a blue beam review. And I was just going manually through, and I was trying to find, okay, detail in our drawing, detail in the fabricators drawings, where, you know, which ones need to align with each other. It's the same way with standards, right? You have a, maybe it's a checklist, one of our clients is a checklist, four pages, they put it on one screen, they have their drawings on the other, and they just go one by one. They go down the list, they go find the relevant information, and they do the comparison. The thing there, there are two issues. One is, that's a lot to check. Humans, we have limited cognitive capacity. We get tired, we get bored. You know, we're going to make mistakes, right? That's just going to happen. And then also the complexity of a drawing set. Finding every piece of information that should align with a certain standard or detail, whatever, it's just not going to happen, right? And it's not feasible. There's so much information in drawing sets. And to be comprehensive, there we go. And this is a big, and we're good to challenge I think in a bit here, Nick, with applying AI, comprehensively, stuff for humans, also tough artificial intelligence, right? And I think that's currently an issue we have with human review that we think we can solve with artificial intelligence, but that comes with training, and it comes with just more work with AI agents. And more advancements in AI in general, right? Like we're working a lot with, or the frontier of artificial intelligence. And so every day, new tools are coming out, new open source efforts, and new breakthroughs are happening, right? And so it's interesting how we're working hand in hand with the tech community, with the AI community, as they advance, we can advance, right? And vice versa, right? So. And you mentioned AI agents there, which I think after just the generator AI, I think is like the next most common buzzword, but as both engineers, try to keep very practical. How do you envision AI agents supporting or even automating workflows within the industry? And then can you just break it down again in very simple terms? Like, what is an AI agent? What does it do in practical use cases? Yeah, you could call there a mix of an agent. I'll start with a little thing about chat TBT, right? Actually, it's tough, because these days, they're also their ages, they've kind of ballooned. But originally, it was, you know, chat TBT first came out, a static large language model. It had been trained in all kind of information, but that was it couldn't access outside information, new information, none of that. You asked a question, hopefully you'd knew it, right? Which at the time, it was just mind blowing it, knew anything, and it could respond, like it ain't mostly coherent, human sounding way. That was how it arisen it was. And then as you've gotten into agents, right, we still have that the large language model as that core. But then we have really addition of tools, as it is a big thing here. And then what I would call like a, maybe a master large language model. So a large language model that can understand what's being asked of it, and they can use different tools to then go perform additional tasks, get different additional information, maybe even access additional large language models, right? And then take all the information he gets back, like a project manager, right? And he can synthesize it all, make sense of it all, and then can, you know, so spit out an answer if we're talking about a chat-like experience. So agents that way can access outside information via tools, databases via tools, access other softwares via APIs, right? And that's a big thing with agencies days. MCP, model context protocol is, was a big buzzword there for a while. I don't know if it's died down recently, honestly. It was like this, we hit with just AI, right? We see these huge hype waves, and then we'll see what it sticks, right? Yeah, because that's agents fundamentally. Still an LLM, reasoning ability, ability to understand human language and respond that it's core. But now I can access outside data and pull that in for an analysis and then give you an answer. That's more than just it was trained initially. I've also heard it's, right, like, virtual computer, right? Like it's essentially during the button clicks that you would do as a user. Based on your instruction, go to websites, maybe log in, don't give it sensitive information, perhaps. Yes. Maybe it gets blocked because a certain provider does have a lot of agents crawling its website, right? But the idea, and I think you nicely explained that it's like, take what we do to computer, right? Have it emulate that. But even, I think, I wouldn't even call it more sophisticated, right? Because if you look at some of these tools, like the deeper research in particular, from my point of view, it's incredible. Yeah, I would never be able to do that as human. Yes, I mean, the compute power is incredible. The memory is incredible, right? And so, yeah, deep research is a great example. And then this is interesting. This is where we get a little nerdy. So I'll mention this and you can cut it late if we need to, right? But the ability for, we don't know if large language models, if AI can actually think in reason. Some people think it can. A lot of people think it can't, right? But deep seek was a really cool example of this. So, let's go to chat TPPT. TPPT, the latest ones. They were trained with reinforcement learning. With human feedback. So they had a target, this large language model. They would perform analyses. And then they would be great against that target. And then a human would say, hey, here's how you feel short of that target, right? Here's what it should have been. The large language model will go, oh, okay. All right, I'll remember that for next time. And I'll do it again. I'll do it again. And just, you know, so many iterations just to learn. But humans were in the loop giving it feedback. What's crazy is deep seek was interesting because purported at least, I don't know if we know exactly and this is how it was, it self-reinforcement learning is what it used. So there wasn't a human loop to help it think. It was based on its objective. And it would self-assess its reaching its objective after each analysis each run. And it would then self-adjust, right? And through doing this, they found that it became much more efficient at answering questions and generally better. And that gave it more compute resources to think, quote-unquote, right? And so I started thinking more on its own because it was incentivized to, right? And they found that as it started to learn to think more and be more efficient with its original thoughts that it did improve its performance. So I don't even know how we got on this tangent here. Nick and I said you can calculate if you want to. But this is like, you know, we're talking about, yeah, applications in AC and practical. And we're doing it. Other companies are doing it. Like it's coming. It's here now, right? But there's whole, this whole world of AI and people ask me, Luba, what's the one thing to know about AI? And I always say like, it's going to get weirder and crazier way sooner than you expect. It's just a very powerful technology. And people are doing some incredible cutting at reach rates. Stuff you don't even know and see. And we're just going to be amazed. Both in a positive way and in a negative way, I think much through them than we expect to. So buckle up and do our best to push it in the way as engineers. Once again, we're not replacing engineers. Our philosophy at ethos. We wanted to augment, take away this repetitive, boring, highly manual tasks and engineers, you know, design more, designers, design more. So they come about. I think it changing faster than you would expect is very appropriate because if you, right, if you think about, so how humans think, linearly, we experience life, linearly, construction is a very linear process. The development of these models is not. And it's, it's truly exponential. And if you, I guess, read about the history of them, for a long time, we were at the start of that exponential curve. It just kind of looked flat, right? And if you were looking at it from this side, one day, you're going to be looking up because it just went by that quickly. And it's, it's incredible. And that end, like, what we just, what we see today, right, the consumer facing stuff, right, the products that are available that you can buy, then, like you said, there's this whole other piece of development in the background that just isn't known to the public. Right. I guess it's called stealth motor, wherever you want. So, yeah, it's as, as interesting as crazy as it seems. And now it's like, yeah, just, just wait. It only gets better or worse, I guess. Yeah, absolutely. And we'll see. And that's what's interesting. You know, you go up and you read history and you're like, oh, wow, like they should have known that like, duh, the car was going to come and replace horses. And, you know, duh, that was going to happen. But at the time, you know, I don't know if I knew if, Ford and then we're going to succeed. At the place, the horses, they thought it was crazy. And so I'm sitting there living in the time of AI. And I'm thinking, this is another one of those like, we might look back and history and go, duh, that was just going to be disastrous. Or duh, that was going to change everyone's life. But we're right here at the edge of history. We have no clue. And especially in industry like the AC, right, AC industry, that has so much room for improvement and efficiencies. Where we've been lagging technologically to be 15 years behind, you know, every other industry. With this intelligence that we're bringing, right, this outside intelligence with great compute power, great memory, that can see things we don't see without even prompting it. We're going to see some real cool advancements, some real quick advancements in the industry. So, and I'm excited to be a part of that, right? Yeah, I love always been in AC. I love the industry, love the people, right? Like some of the heartbeat, heartbeat of America construction, right? And I come from the south, right? A lot of people grow, and that's what they do as a career. They, they, the construction workers, good people. So yeah, I'm, I'm proud to be, to be working in this industry. And hopefully I'm going to change it for the good. That's the goal. And it is directly, directly correlated to the success of these technologies, right? Because you can't put a server farm out in the middle of nowhere with no roof. Yeah, it starts to rain and bye-bye. So that, and that's, of course, right. Data center boom, and so on and so forth. So we play a critical role in just, yeah. The built environment, America, worldwide, and then advancing these technologies. But we've, last question for you. Any final pieces of advice you can get to AC professionals who are curious about integrating ads and to their workflows, but maybe not sure where to start. Yeah, it's a quick question. I was thinking about that earlier. Even my start, right, was I was doing something, didn't like it. I thought, okay, how could I not do this in the future? That's just my natural inclination, right? And it was highly manual, high-repetitive. And that's kind of what makes things. It's very important work. It's vital work. It's just tough work. It's not the easiest. So I would start there. And it's easy to say, hey, just think and take the initiative. And that could be tough to ping. I'll be the company's right. It's not always easy to take initiative in a company. But it's such a. Everyone's looking, how can we get AI into our workflows, right? And it's like top-down issues a lot these days. And I guess I'll speak through a bit of like who we've been working with at companies. So we find champions at companies, whether it's through a conference or through random in-bounds, or I don't know what it is, right? And we meet with a company. We say, hey, can we meet with a few teams? We find that one person. You're just kind of like, you know, I'm curious. That's really what it is. I'm just curious. What can AI do? I'm the expert in my problem. You're the expert in the tech. Like, what can we do? And it's that core curiosity, right? Just having it, fostering it. And I guess staying persistent with it, right? Finding ways, okay. We're still doing this manually. It's still highly repetitive. I'm curious, can AI do something. And I think a lot's being rewarded. I hope at least in the industry these days of professionals who are willing to say, you know what, I don't have an answer necessarily. But I think this might be it's highly repetitive, highly manual. I think this could be a good spot to explore, bring an AI in, or any kind of tech in efficiencies, right? And we actually, I mean, I, I hope everyone in the industry hears this, because this would be great for us and other players. Your experts in your pains, right? And in what really bogs you down, where experts in the tech, and when we come together, we create a product, a solution that really can change the industry. So we need these champions out there, right? Here are experts in what they do and those pains and those struggles. We need them to say, you know, like, we can do better than this. You know, I can do better. We can do better. And now let's push it to the market and say market. What can you do for us, right? And once you get a ping on with love, you're on a company. And how big it is, it's tough to get heard. But I do think project managers hire ups. They have an ear out right now for, okay, where can we have mandates to get it in, to get AI into their workflows. And so stay curious. And just know, and you know, I'm coming from the other side, know that it's, it's possible. Maybe now, maybe later. But whatever problem you're facing, like, there's some way, I mean, I don't know, it sounds crazy. AI, there's so much potential. There's so many, I can probably address it, right? And if I, yeah, there's some tech out there, right? It's being a different problem, repetitive enough, manual enough, don't be satisfied with it, right? Push the status quo and I think we'll see some champions arise out of them. Some people succeed in their companies because they have the curiosity and the initiatives that you know it. What if we address this? What if we do things differently? So that's my aspirational advice to an listening to the industry. Stay curious, believe things can get better. And last thing I say, just as with my, my plug, I guess, you know, the work y'all do is incredible. The work the industry does is it truly is. And I remember seeing as engineer many times thinking, this is tough. This is tough. But you forget how crucial what you do, as soon as you can seem like you're just a piece of a puzzle and you are, right? You're a piece of that effort. But in a lot of ways, AEC drives the world, drives America, keeps it safe, right? As an engineer, man, I learned how the industry is like, wow, our work is so crucial. Buildings don't fall often. There's a reason for that because the engineers do their job, right? Safety standards regulate it. Richway bodies do their jobs. And it's the stuff that people don't see on day-to-day. But then a building falls, you know, bridge crashes, it's news worldwide. And so it's hard to remember how important the role we play is, just in the world, the social fabric. So kudos to y'all. Thanks for doing what you're doing. Keep doing it, and to an odd is doing the best we can to come alongside y'all and, uh, hey, work together to make things better. So. Well said, Luke, and if, and if the audience has more questions for you, watch to learn more about you, your background, or the company, what's the best way for them to connect with you? Absolutely. And I'm going to measure that to Nick. I'm glad you brought it up. At this point, I think you have the community still. Do you just want to, please email me. I'm happy to give it Luke.Reve at twennonge.com. I'm also linked in me, Luke Reve. And there was nothing I was going to say, Nick, like, uh, reach out to me. Reach out to anybody in this space. Because I think a lot of us, especially if we cover the space, we want to talk to others who are interested, right? We want to help grow and foster this community. And we enjoy sharing what we know and what we're doing. So, uh, you have reach out to people and there are opportunities all over the place if you only reach out and ask. So yeah, reach out to me. Email, uh, LinkedIn. I don't have a Myspace anymore. I probably do actually don't know how to get into it. I don't go on Facebook ever. I'm an Instagram, but barely. I have no TikTok. I'm a millennial. So yeah. Email and LinkedIn. That's right. That's it. That's, yeah. Email and LinkedIn, the millennial. I will, Luke. Thanks so much for taking the time to join us. And we'll be sure to include your contact information in the show now. So if you are interested in the audience, want to reach out to Luke, feel free to do so. Luke, thank you again so much for taking time. Pleasure. Thank you for, uh, for having me and that's an honor. So. Yes sir. Till next time. Sounds good. Please remember, we can find the show notes for this episode at actechpodcast.com. There, you'll find a summary of the key points discussed in today's episode, as well as looks to any of the resources, websites, or books mentioned during this episode. Till next time, I wish you the best and all of your engineering and technology endeavors. [Music]