Turning AI Adoption Into Real World Result in AEC

Let's talk quick BIM and plan aid. So obviously you know tell us more but then I think [music] the other thing you know our audience would really love to know is you guys are an architect you know have an architecture arm right and companies try to do this internal development all the time. It's like how have you guys been successful [music] in getting something to market whereas others you know tend to kind of flame out. >> I think one it's our shareholder backing. We raised 17.5 million dollars in the IPO which [music] the vast majority of that went to our flagship product called quick BIM. It's basically a product AI tool that autogenerates BIM [music] modeling. So in terms of being able to be successfully able to do it is we raise adoption at AC has reached an interesting inflection point. Firms are no longer asking should we use AI? Instead they're asking why hasn't this turned into tangible results. So the gap between interest and execution is where most initiatives stall and closing it requires more than enthusiastming. It takes a deliberate process of the right data foundation and a very clear view of which problems AI can solve. Today we're joined by Thomas Gaffne, chief operating officer at OFA Group where he leads operations and digital strategy across a publicly traded diversified portfolio spanning architecture and engineering. In this episode we'll speak about how OFA Group built and deployed two AI products quick BIM and planade. what it actually takes to turn them and project data into actionable [music] intelligence and how leaders can think about preparing their teams for the next wave of technological change in the built environment. So, if you're a firm leader who's moved past the question of AI adoption is now wrestling with how to make it tangible, this episode will give you a ground look level into what that [music] process involves and the decisions that determine whether it sticks. But before we jump in, I want to tell you about an AEC PM Connect, a series of in-person events we've created at EMI for AEC project managers and the leaders who develop them. If you want to stay on top of PM trends, elevate your team, and deliver stronger project results, these fullday events are built for you. We're coming to Kansas City this October. Learn more, and register at apmconnect.com. With that, let's jump into today's episode. Okay, it's now time for our conversation of the week with Thomas Gaffne. Thomas, thank you so much for taking the time to join us on the show and welcome. >> Hey, Nick, thanks for having me, man. I really appreciate it. Looking forward to it. >> Absolutely. And I know we had a we had a nice kind of pre-recorded chat here, so I'm excited to learn more about you and and OFA and uh the unique career path that you've gone down. So speaking of which, can you can you talk to me more about your career path? So started in legal and it's eventually led you to leading operations here at OFA. So what drew you to your current opportunity? Um and how's that been going so far? >> Sure. I mean, it's kind of a interesting long story. I mean, when I was in college, I was supposed to go in the military. Uh I was contracted with the RTC program. Long story short, broke my jaw in three places playing sports. tore tore my ligaments in my left shoulder. So, I was medically disqualified my senior year and uh I mean I had been planning on just going to the military and I political science degree and I was like, well, what am I going to do? So, I was like always kind of wanted to potentially go to law school. So, I was just like I go to law school. I got to law school and about two weeks in I was like oh this is not for me. It is a lot of just reading and writing. That is just all it is. And uh it was kind of the best and worst thing that ever happened to me cuz it really taught me how to think and analyze stuff. Uh but I always realized when I was in law school I was like I want to be on the other side of the table. I want to be doing business rather than the legal components, right? And so uh I worked my way up uh to end up working with a bunch of startups uh in the venture capital world. I did uh all of their financings, very high high-profile deals. um learned kind of like the ins and outs of kind of like team building and all that. And uh I finally actually just had the opportunity to kind of meet Larry, our CEO. Uh and I was telling him about some of the things I've done like I had started my own company. Uh we made like basically NFT video game. Uh sold all that out was pretty successful and always had very much an entrepreneur spirit. like my wife owns her own business that I help uh operate on a daily basis and kind of just met Larry at OFA. Got started talking. I was like, "Hey, well, he wanted to go IPO with uh some of these architecture AI design uh products that he's building and I was like, "Hey, well, I'll help you do that. I'll come on COO." And he's like, "Yeah, let's do it." And so, just got into the whole operations of the whole company and uh the growth expansion plan. And it was kind of like a perfect fit cuz even though we have a legacy architecture design firm, right? Uh our tech side is kind of like a startup, right? Because all the architecture people, they're not dealing with any of the tech related things that we are. Uh and so and vice versa kind of. And so basically just got in, got on the ground, started developing, we got the money, we went IPO, and and now I'm here. >> Beautiful. So it's uh really interesting, right? because you don't see you don't see somebody with that sort of let's say like background business acumen and expertise like such as yourself coming into this industry but it's cool because um >> it sounds like Larry's a >> pretty visionary guy and saw the opportunity that many others just haven't yet. >> No, totally. Uh I totally agree. And so it's cool because it's funny learning more about the AEC industry. It's kind of like the law industry. It's like it's slower to adopt some like tech because I mean it's very technical, highly complicated stuff like in engineering and architecture and so uh there's like a grace to it. So it takes a little bit of time to uh kind of expand and adopt the new technologies for it. >> Speaking of which, so you had you had alluded to a couple that you guys have. So let's talk quick bin and plan aid. So >> sure. Okay. >> Yeah. So obviously you know tell us more but then I think the other thing you know our audience would really love to know is you guys are an architect you know have an architecture arm right companies companies try to do this internal development all the time so like how have you guys been successful in getting something to market whereas others you know tend to kind of flame out >> sure I mean I think one it's uh our shareholder backing uh >> we were able to raise a ton of capital and so we were able to go public right so Like I said, when I met Larry, who was talking about wanting to uh build out these tools, and we also met this uh investor who kind of believed in this idea and wanted to help take us public. Uh and so we raised $17.5 million in the IPO, which the vast majority of that went to our flagship product called QuickBim. All right. So QuickBim, uh BIM meaning building information modeling. It's basically a product AI tool that autogenerates BIM modeling, right? And so uh and it costs a lot of money, right? So in terms of being able to be successfully able to do it is we raised 17.5 million to put into this product and then we raised another $50 million in a pipe that we continued for operations, expanding, being able to do marketing stuff, right? um being able to keep um employees paid while we're developing these products and trying to grow, right? And so I think part of it is just the backing that we had cuz even though we're an architecture design firm, like that is like our legacy business, we really see ourselves now OFA Group, right? Because Offa Offa stands for office for fine architecture, right? And that is the architecture arm that pretty much is the one vertical that they just operate in their lane. And yeah, they use our tools, but we've really branded OFA Group to be a tech finance company, >> which is which is pretty incredible because you you kind of alluded it to how law and then the AC space are somewhere, right? Both very mature industries. There's a certain way business has been done for a long time, but you guys are really branching out and essentially creating a separate division of what sounds like a very innovative company to take advantage of what the past couple of years have offered to us. >> Yeah, I think you kind of hit it on the head right there. like uh like Larry has been Larry had started a separate business with uh some tech stuff. Uh but he's always been very techoriented and very futuristic looking and that's why when we met I was talking about crypto cuz my background is all like in the web 3 digital asset space right and so and over the last 10 years that it's just come roaring on the scene and so he was very interested in all that and stable coins and stuff and so we're looking into a bunch of different things and that's when we came up with this RWA platform right so an RWA what is that it's it's a real world asset right? Um all it really is is just something that exists in the real world, say a property, a building, construction, whatever, right? Uh and you put the legal rights to that into a smart contract and you put it on chain and make it into like a tokenized form. Uh and so we've had two very big projects um service procurement projects that we just landed for that over the last uh three months. one is um so this arm of the company as well is called Herth Labs and they're a tokenization service provider and so we landed one contract in Long Island Beach which uh the projected value of the property is billion dollars uh and we're going to be tokenizing that and fractionalizing it and helping the investor uh kind of put them on chain. All the the the rights to that property is going to be on the blockchain, right? And then we have another one in Vero Beach, Florida, which the value of that is going to be roughly 500 million. Uh, and so these are are two kind of large commercial uh contracts that we have for the RWA platform that we build out >> in the AEC industry. Skilled project managers are the backbone of success. EMI's AEC PM certification offers tailored training, ongoing support, and exclusive access to premium content that keeps your [music] team sharp and effective. From foundational training to advanced leadership tools, set your team apart with certified growth. Learn more at aecpm.com. That's aecpm.com. >> Beautiful. So, um I think what our what our audience here would really appreciate is dig into more of the specifics. So, why don't we start quick Ben Thomas? So you know you talked a little bit about how it came together but what does it exactly do for the architecture profession and >> you know what are you guys seeing as some of the the tangible benefits for practitioners. So the what it does is it automates BIM modeling right. So, so BIM right building information modeling traditionally drafting to like an architect will go have the blueprint, right? And then you need to get all of these other professions to input and create the building the BIM model, right? And so where [sighs] that takes a lot of time, energy, and effort. And anytime that something takes time, energy, and effort, it costs a lot of money, right? And so where we really see it being helpful is it can take months to get a BIM model out. Whereas quick BIM, you can get a model within with if you put in the right parameters, you can get a model of a BIM for a building based on a blueprint uh within uh an hour and a half, two hours, right? You just have to plug in the certain parameters. Like basically it's like prompt engineering. If anyone's used Chat GBT, Claude, whatever. Uh that's kind of how BIM uh quick BIM is set up. You upload the the blueprint. You put in the parameters of like certain things about the property like size, dimension, the scale, the the quality materials that you want, and it will generate after you spend probably like 20 30 minutes, and then it takes about 30 minutes to completely render the model all of the information. And so how much concrete that goes into the building, how much rebar you're going to need, the type of rebar, the type of concrete, the amount of wood, paneling, everything, the electrical, plumbing, HVAC, it generates it immediately, right? And so where this is a huge benefit is say you have an investor who uh is soliciting bids from architects, right? And they're like, "Hey, here's the budget. We want to submit a budget, right? Normally, you would have to have some kind of BIM modeling to figure out how much it's even going to cost. Uh, and so instead of just guesstimating, you can get in there, render a BIM modeling within a couple hours instead of paying tons of money and having it uh take forever. You can actually get like a a a good estimate of the actual cost of the building. So then one people, so architects can then use that for soliciting bids. engineers. Same thing. Investors can use it themselves to see if it's like, hey, would this even be valid or or logical to make this investment on this property and build this type of building or cuz are we going to see any types of returns? And instead of having to spend hundreds of thousands, if not millions of dollars modeling this whole thing out, uh you just sign up for QuickBim with our paid program, spend $300 a month, and you can get a very very good estimate of what the how much the property is going to cost, what it's going to take, the materials is going to go to, etc., etc. So, what it really does is it helps expedite and the construction planning process. instead of taking months, it takes matter of hours. >> And you know, to the audience, if you're thinking like, hey, like, well, you know, I've got a BIM model, right? I'm a designer or I'm already creating it. It sounds like Thomas, there's a there's another level to this because even if um let's say the the construction documents, obviously, they come out as plans. >> The model from which they were generated from may not be as detailed or informative as what QuickBim is is outputting, if I'm understanding correctly. Yeah. Yeah. Yeah. Complete. >> And it can and like you said, it can be a manually it can be an extremely time consuming process to get these models to the level of detail required to do the things you just discussed. So that's it's pretty incredible. >> Yes, exactly. And so uh we would say like currently it's not at the point where it's like, hey, this is 100% shipped. We would say like it's like 85% of the way there because every building's slightly different. And so you would want someone to come in and tweak it a little bit like a professional and just make sure it's all good, but it gets you about like 85% of the way done instead of starting from scratch. And >> we find that to be it's it's fairly common. Um I think where people get into trouble is if they right if they try to play expert as a novice instead of letting the expert kind of wield the tools and use their judgment because >> Exactly. We don't expect any of these tools to be 100% perfect, but they're certainly >> much better in the hands of an >> cool. So, same sort of question for plan a Thomas, right? So, you know, the audience Thomas and the group have two products that they're we're talking about here. So, >> QuickBam is what we just discussed. Now, tell us more about plan a Thomas. >> Sure. So, Plan A is uh less further along in terms of development because we were focusing on QuickBam, but it's still still a good way there and we plan on releasing it uh probably in Q3 or Q4. And so, but Plan A essentially they kind of work in tandem together, right? So, Plan A is for like the building and permitting portion of like a blueprint, right? So ideally, so what you do is an architect, say they have a 23 story building or something. They upload the blueprint to the planade software and plan a then so you would type in the address, right? So uh you get the jurisdiction and plan a pulls all of the local jurisdiction codes, right? So for example, say like the fire code, right? So when it goes in, it'll look for all of the different distances and the requirements of based on the blueprint um for the fire code, right? So um say it'll there's a room that is uh say too far away from an exit, right? Like it'll come up and say, "Hey, here there's the flag here." It will like flag the the issue and uh it'll say not in compliance. and it'll go through and read all of the code and just evaluate it whether or not it would pass code compliance inspection from a building office validator, right? So like um traditionally it takes like two or three or four times to get a building permitted, right? Like based on like a blueprint just because they're like, "Hey, well you got to change this. This door is too small. Uh it needs to be wider. this uh uh this room has to be closer to the exit so it's not within compliance, right? And so because mostly architects previously was doing this by hand, you know, and they might not be 100% familiar with the jurisdiction that they're working in because they live in LA and they're working with someone in San Francisco and designing a building, right? And so uh what Plan A does is essentially acts as like a uh building permitting officer, right? just reviewing these documents and in real time it's checking being able to tell you where you're out of compliance. And so, uh, hopefully the the plan is to be able to use this tool. And instead of having two, three, four rounds of going back and forth with the examiner, uh, you can reduce that hopefully to one because ideally you get the examiner to use it, the architect to use it, and it's kind of updating the code and and reading the code in real time. >> Yeah. And to me, it sounds like right, you're essentially translating the what what drawings are, which are just pictorial descriptions >> and converting that into something that can be compared to the code and then running through your checks. But it also I mean to the the clients of these architects I mean that has to be a huge savings from Yes. time and the permitting cycles and then catching any errors or omissions prior to this actually getting sent to bid and then construction. >> Yes. Exactly. And that's what the whole plan is. Right. So it's like uh in the construction world uh especially if you're developing a property that's going to be rented out uh time is money big time right? So if you can cut down the 6, seven month period it takes to get all the permits down to one two and you can build that building five six months faster, that's five, six months of rent that's coming in for the investor, right? It's less time spent on drawing architects, engineers, whoever's working on it like working. So it's cheaper and more efficient, right? And so, uh, ideally, it's it's there to help speed up the the process, get the per building permanent faster, which makes the building built faster, and then, uh, you can get people and start renting out faster and start collecting income >> and and in theory lower construction costs as well because if there's, uh, let's say less >> less bickering between the design and construction teams, well, that means a lower construction cost to the owner. >> That's exactly correct. Yeah. >> Beautiful. So, so Thomas, I I appreciate the the deeper dive on these two products, but we can take it up a level and talk about data strategy and governance. So, >> we know that's foundational to any successful AI implementation, but >> for you guys, what does that found that right foundation look like? And then where are other organizations maybe not meeting that same standard? >> Sure. So, one of the hardest things that we had to do for QuickBim and for Plan A was getting the computer to be able to read the handdrawn lines, right? Like be able to uh like actually interpret the blueprint correctly, right? So, and what that does, it takes time. We need thousands and thousands of hundreds of thousands of blueprints, right? And then adjusting the model, uh having the data implemented, right? And so eventually what it it it's going to work is so like the machine is learning more as it's working more and as as it's being fed uh more inputs, right? And so uh ultimately it just comes down to really using the product and and training it based on getting as much data as you possibly can. Right? So it took a long time for the the machine just to be able to read the lines and then uh so it took a lot of handmade tinkering process but as you are tinkering with it every time you tinker with it the machine is getting better right because it's saving that data it's saving that interpretation it's learning right so like the whole machine learning process uh I'm not a data scientist but that's my understanding you know on a high level uh but um yeah ultimately it's just the more people that uh up upload and input data into like their schematics into the system, the better the system's going to get. >> And if you talk to the experts in this, which I've had the good fortune of talking to a few >> nice >> so these horizontal or foundational models chat claude, Gemini, >> the corpus of training data that's out there generally excludes a lot of drawings from our industry. That's why they struggle so much. >> So there's a it's a to me it's a blessing because we have the opportunity to develop these custom tools and then for call it job security, right? There's always going to be a need for a human because these um you know offtheshelf horizontal AIs can't do everything. So to me it sounds like you guys have captured a nice little niche in the market because the big providers just haven't yet or can't >> Yeah. They just haven't necessarily focused on it. Right. And it's because it all comes down to like it's so I'll use an example uh with like legal stuff, right? So like as a lawyer, I started using Chat GBD the day it came out. I was like, "Let me check this thing out. It's cool." And originally like was not great, right? Uh and like it could write stuff. It could do some research for you. It was basically Google on steroids, right? But now four or five years later, right? Like it can write full-blown contracts, you know what I mean? that are actually half decent, you know, but it takes a lot of uh like still a lot of like input. You got to adjust things. You got to make sure it's good. You have to know how to prompt it correctly. You have to know you have to then go and read the contract, right? And then take out sections, adjust it, uh add sections in, make sure it's good, you know, like there's still a lot of like legal lawyering that needs to be done for these things, right? But it went from like you can't even use this to oh wow this is now saving me hundreds of hours. You know what I mean? And so that with the LLMs though is everyone's just it's just words, right? So exactly what you said there not the drawings, right? And so like and AI is really good at words because all legal is is just a set of rules and then words, right? And that's literally what computers are really good at. Uh what they're not great at is like creating images. They're getting better, you know what I mean, for sure, but it's all based on certain parameters and they're pulling from like all these videos, right? So whereas all the data for architects is like that's in AutoCAD, you know what I mean? Like that's like those aren't com those aren't hooked up to these LLM models, you know what I mean? So they just don't have the data >> and it's it's not, you know, right? Typically we we're a pretty you know close to the chest kind of industry. So we're not >> excluding public projects which even still there's um only so many of those. >> Yeah. >> Our work product as an industry is just not in the public corpus of training data. So I think you're >> you're exactly right and you know to the the point of whether it's attorney, engineer, architect, I have this discussion with my attorney a lot where he's you know super capable hands-on practitioner but the difference is right there's a difference between being able to review a contract, put a contract together, redline it, but then there's right the advising the more um >> yes >> let's say qualitative >> what does this actually mean? How do you actually do this? Yes. >> Yeah. And then like, okay, well, and then the seeing around corners piece where like >> if you if you knew what question to ask of the LLM, you'd probably go to get get a good answer, but it may not >> help you consider that just based on the specifics of your situation. Does it have all the context of your situation? >> Yes, I completely agree. But the funny thing is too like you even run into like cuz I I still do a little leader work just cuz uh I have a book of business that when I came to off like my clients were like hey can we stay with you we don't want I was like all right I'll just do a little bit on the side right and so perfect example is like I'll write a contract and then uh get everything ready and then they'll put it into chatbt and chatbt will just be like oh this this this and this you got to consider all these things and it's just like it's like yes I've already considered these and we can talk about them. You know what I mean? But chat also is going to just throw out any question to have questions. Do you know what I mean? They're like cuz it's like legally chat isn't a lawyer, so he can be like, "Oh, there's nothing really wrong." So, he's just going to produce things that aren't necessarily an issue, but he has to produce it. And they just are like, "Uh," and then it'll like spit out something. It's like, "Hey, you should do this." It's like, "Well, if you look at the other contract that's connected to this contract, you only put one of them in." it's already there, you know what I mean? And so it like creates like these they like think they're like, "Oh man, >> uh I can just ask Chad." You know what I mean? They're right on a lot of things because you will get a lot of good um good stuff, but at the same time, they create more problems now than you would have because they are like armchair lawyers with chat next to them. You know what I mean? And uh right and there's a right if you're retaining an expert to do something for you probably trust that they know what they're doing otherwise they're not the right expert for you. >> Exactly. >> So Thomas is uh now now if we talk about like project delivery and how this starts to like kind of boil down to firms dayto day. >> Yeah. So, how do you see the roles of architects, engineers, and builders evolving as AI be becomes more prevalent in project delivery? And then how does that balance with the human part of this? Because really, right, we design for people, we build for people. So, this is not just some, you know, something contained within a computer. >> Yeah. And I I completely agree. And we'll how I see it is I think it's going to AI is going to speed up the design and construction process. You know what I mean? Where like the the whole planning process, right? So like everything that we're doing with AI right now is all about like getting the materials and the information we need before we can actually go out and build the thing, right? And so like buildings in America takes so long to get something done. There's so much regulation. There's so many different uh like entities and people you have to deal with and governing bodies, right? Permits and all that nonsense. And then there's also just the actual work, right? So like I don't think AI is going to replace architects, engineers, or construction professionals. I think they're just going to make them better at their job and there's going to be a boom in economic activity, right? So, traditionally, if a building took five years to plan, permit, and then build, uh, and two of those years were the planning process, right, the investment process, all of that. If you can reduce that down to about six months, right, then all of a sudden that's one and a half year save. that is 40% faster, right? So then investors get their money back faster. They can start collecting incomes faster and then when investors, if you know anything about them, as soon as they get money back and they make money, they go invest their money again. You know what I mean? And so I think that there's going to be more work available because there's going to be a more efficient building process. And that's that's just kind of my view in AI in general, right? Like so like even if we use the legal example right so uh traditionally a lot of the projects that I've worked like AI has made me 10 times more efficient 10 times right and so like since you're more efficient and you're better you can charge a little bit more but you're spending less time so it's actually cheaper for the client overall. You know what I mean? And so and then instead of because like one of the things with lawyers it's like the client's like they see you as an expense, right? And so, um, but if instead of something costing $10,000, it's only $2,000 because you're 10 times more efficient, right? Uh, and then you charge us a little bit more because you are that good, right? Because you already saved them AGs. It's like, okay, now we'll send you another one. You know what I mean? Oh, we'll send you another one. So, because instead of spending $10,000 on one thing, now you're getting 10 things done for a just a little bit more than something for one, right? Right. And so I see that the same way in uh the AEC industry, I think that like so using quick bin for an example, right? Perfect example is say a construction person is soliciting bids and you're a smaller firm and you don't have the skill set or the the money money and resources to do a BIM modeling before the the bid selection. You can use quick BIM for $300 a month and you can generate the BIM modeling and then come in and have an accurate pricing on your bid so that you can know whether or not it's going to be sufficient or not. And so instead of spending uh couple days on one bid, you can do 20 bids in one day. You know what I mean? And you have more efficiency to be able to do that. And you have more eyes on you as a smaller firm. and then you can actually be more competitive compared to some of the bigger firms. Whereas with QuickBam with the bigger firms, you can be submitting thousands of bids, you know, a day for everything that's out there. And it just uh I think in and eventually that generates more business. It's more efficient, right? because now you actually have a legitimate estimate of how much this is going to cost because you already ran all the the modeling for the materials that it's going to take to really build this building, right? And so, uh, I think that there's going to be a lot of room. It's going to change just like AI is changing every industry really, but I think there's I mean architects aren't going away. Engineers aren't going away, you know what I mean? So, uh, I just think that there's going to be more work out there for them, >> which is right the classic case of if you know price drops, >> demand can spike because more people can afford it. And then you can have that that economic discussion, which is great. And part of what, you know, I'm hearing from peers is like, well, now architecture and design is becoming accessible to more of the population instead of that just that like tippy top. >> Yeah. Exactly. So, and it's and it's cool because, you know, we're we're really just at the the start of it, right? Because uh like you said, it's not even been four years since Chat GPT dropped and there's still a lot of uh a lot of room and potential. >> Oh, so much. I mean, it's crazy like the the increase of what it can do is is wild. And so, I do think over time it's just going to continue to get better. And I think over time humans are going to learn how to interact with it more and just it slowly becomes like a tool to work with, right? Like instead of being like, oh, this is going to replace you, which I think it's going to replace some tasks, but like like there's always going to be lawyers because like you said, looking around the corner and being able to know the right answer, right? because like there is a human intuition and an experience that just goes far and beyond what the current capabilities are, you know, and so uh uh I definitely think that humans are going to be needed for the very long future in my >> and you know and I I love the term tacet knowledge because there are things that we know like you said intuitively >> that we may not be able to explain >> and that's you know you call it taste. judgement, tacid knowledge. It's all great, but the key part of it is for now >> that's not embedded in the training data of these models. So there's still a need for right this this very important thing between the ears to uh >> to to do some stuff that our AI friends just can't yet. But who knows it it may uh it may turn out that they can. >> So Thomas, if you're talking let's say adoption of AI innovation any new technologies. So from a leadership and then team development standpoint, how do you get people prepared for all this change? Because it's certainly a different way of working. >> So how I've done it is I just tell them to start using it. You know what I mean? I'm like, "Hey, uh, just use this." Right? So for example, like I'm just one of my employees will just be like, "Yo, just throw this into ChatgBT and see what it comes up with." right? See what the question is and then they're like, "Oh," you know what I mean? It's like, "Oh, hey, put into Claude a financial model and just get some projections, you know, like let's see what it looks like, you know?" And so, because it's like I do believe that prompt engineering is very important for when you're using AI, right? So, like you you get into AI when you get out of it, right? I mean, you put into it, you get out of it, right? And so, um, having them just actually use start using AI, uh, they get better at it. It's like a skill set, you know, uh, being able to talk to these LLMs, you know, like there is a legit like someone who just started using JTBt or something for the first time is not going to get the same output from me who has been using it every day for five years. You know what I mean? And so, uh, I'm going to make the computer work harder for me than it would for a a newer user, right? >> And so, uh, I just tell all my employees, I just like, hey, just start using it, you know what I mean? And just start getting better with it. Just test it out, you know? And cuz, like I said, when JPG first came out, it was like Google on steroids. And now I'm sitting over here vibe coding different dashboards for our marketing metrics. you know what I mean? Uh and so it's like uh because it's like oh I I've clawed bot on my computer but it took time to like really like learn all of that and learn how to do it. And so my suggestion to people wanting to adopt stuff is just give them exposure, you know, and make it okay. Just be like, hey, uh it's okay to use JT. It's okay to use Claude for some of your work product, but don't make it so it's so obvious like you know what I mean? Like make it better, you know? So, like when you type in a prompt in chat, it'll say one, and it'll have an underline, and it'll say two, and then another underline, three, another underline, and then uh an employee will like copy and paste that whole just the prompt into it. It's just like, yo, I know this is purely generated from Chad McD, right? So, like that's fine, but make it look better. You know what I mean? And make like actually go in and clean up the lines. make sure that the the the like the conversation of what you're saying flows correctly. You know what I mean? Like make sure the formatting is right and so and say it's okay that you you did this, but here's how you make it better, right? And then you can constantly uh get improvement and then so then they're like, "Oh, I'm not afraid to just use judgment. I can make it better and we can get better responses. We can get more efficient work and we can get faster work, right?" And so, uh, my suggestion is always just, hey, make it available for your employees and make it okay for them to use it and then help train them in the the use of the computer, right? Because I mean, think about Microsoft Word, right? Like, you used to have to have dictation done for taking notes and constantly and then like or handwritten things and then Word came around, you know? now you're just typing it and it's 20 times faster. But people had to get exposure to it. They had to learn. They had to have some training, you know, and I I see AI is the same way, right? Uh it's a tool for humans to use to make them more efficient. And so, uh but they have to know how to use it correctly or else they're not going to be able to be more. And I I think I think what you just stated was very well said because I had a I had a colleague refer to it like Excel. So everybody like you you go to a jab, right? It's just assume that you know how to use something like Excel and everybody uses it slightly differently. Everybody was trained somewhat differently and there's no one correct way to use it. But you don't just see people saying, "Hey, you can't use Excel." >> Yeah. >> Because it's just it's just another tool. So that's how I thought of it. And I think I think the way you described it was along the same wavelength. And the best way to learn about it just just get into and start using it. >> Just start using it. It's so complex and it's so sophisticated. You just have to use it, you know? And like cuz like I said, and it doesn't know you. it doesn't have any context of who you are and going to like that human intuition of why I think that you're still going to need humans, right? Because I like I said like the output I get from the computer is going to be very very very different than the output someone else gets. You know what I mean? And so uh just because little connotations, little changes in your wording, little changes in the style and tone of how you write that computer claw is like it's gotten so good that it is going to interpret it differently than someone typing very specific or using like the actual language tools where it's like using slashes and be like hey slash this equal sign that and then uh like actually like using code language in their prompts uh is a different way to communicate and it understands that on a different level than someone just saying, "Hey, I would like to make a website." Right? And so they come in, they're like, "Hey, I want uh this type of coloring, right?" And it they use the actual code, the computer code language in the prompts, and then Cat or Claude knows exactly what they're talking about because they're using a different language and they have different tones, right, instead of different things. And so uh it it's crazy how smart these computers are getting but also how much the human ingenuity on using it goes into like this prompt engineering. [snorts] >> Really well said. So Thomas, we really thank you and appreciate you for taking the time to join us on the show today. The audience wants to connect with you, learn more about Offa, your guys lineup of products. What's the best way for them to reach you? >> Uh I mean we're all over social media. We have a website off ofroup.com. We'll list it on the NASDAQ. We are on X. We're on Instagram on Facebook. So, we're just everywhere. Just type in offag group in Google and you'll be able to or chat and you'll be able to find us. >> Beautiful. Well, Thomas, thank you again so much for taking the time. Nick, I appreciate it, man. It was a good time. >> Absolutely. Please remember you can find the show notes for this episode at aectodcast.com. There you'll find a summary of the key points discussed in today's episode as well as links to any of the resources, websites, or books mentioned during this episode. Until next time, I wish you the best in all your engineering and technology endeavors.