If Your Building Could Talk, It Would Fire You

What if the biggest crisis in construction isn't AI adoption, it's that we hand over $100M assets with no instruction manual?

In this episode of KP Unpacked, KP Reddy sits down with David Niewiadomski, former Turner Construction executive turned Shadow Ventures operator, to answer a haunting question: if your building could talk, what would it say? The answer isn't pretty. "You don't do scheduled maintenance. You didn't check the caulk joints before the warranty expired. You take me for granted." Dave spent 17 years in the contractor trenches, pre-con, estimating, project management, and walked away to solve the data handoff problem that makes every asset transfer feel like buying a car with no owner's manual.

The conversation weaves between tactical AI workflows (how to automate bid leveling in two weeks, why Claude told KP he was "out of his depth" and should call Barry) and systemic industry failures. Why do cars come with organized manuals regardless of manufacturer, but $100M buildings get handed over with incomplete data scattered across expired Procore servers? Why don't architects visit existing hospitals before designing new ones? Why do facilities teams get involved after walls are already placed? And why, when KP's uncle kept every oil change receipt in a three-ring binder to maximize car resale value, don't we track building maintenance the same way?

Key topics covered:

• Why IT departments are the #1 barrier to AI adoption, not capability, cost, or interest, just permissions
• How Dave would automate bid leveling in two weeks using Claude Cowork if corporate let him tinker
• Why pre-con departments are perfect AI targets: small teams, high expertise, Excel-heavy workflows
• The moment Claude told KP to escalate to Barry because he was out of his depth—and what that means for mentoring juniors
• If your building could talk: "40% of my caulk joints are cracking and my exterior warranty just expired"
• Why cars have consistent owner's manuals but $100M buildings don't, the automotive vs. construction data gap
• How organized building data determines which deals asset managers skip during due diligence
• The CapEx vs. OpEx disconnect: design teams optimize construction cost, ignore 20-year maintenance nightmares
• Why facilities teams review drawings after decisions are locked and walls are already placed
• The hospital prototype problem: architects don't visit 50 existing hospitals to learn what breaks and what costs too much
• Why grocery store GMs kept selling corporate-spec'd deli coolers on eBay, and corporate couldn't update specs fast enough
• How technology creates deflation everywhere (Blockbuster to Netflix, $20 CDs to Spotify), except construction
• Why RFIs and change orders eat 10-20% of contract value, and AI's first impact will be waste reduction, not bid prices
• Whether contractors will pass 30-40% AI cost savings to owners (answer: no, they'll pocket it until competition forces pricing down)
• Why mid-sized GCs will adopt AI faster than Turner, fewer people, less federal red tape, more agility
• The union robotics challenge: layout robots worked in NYC, but full automation requires labor negotiation
• Why institutional knowledge walks out the door with employee turnover, and Procore data disappears when subscriptions end
• The three-ring binder standard: why we track car maintenance for resale value but not $100M building systems

If you're an owner frustrated by incomplete building handoffs, a contractor wondering where AI automation starts, or a facilities manager tired of inheriting broken systems with zero documentation, this episode will make you realize the problem isn't innovation, it's that we never solved basic organization.

Listen now.

BuildingWorks & Brookwood Sponsors

Transcript

Host Swap And AI Curiosity

SPEAKER_01 0:34

Hey David, how's it going? Hey David. Great to join you on your uh the Unpacked podcast. It's been a little while. So uh if you all don't know, this is not Nick Durham. Nick has been uh hit by the flu bus in his outsect. And uh Dave decided he would step in and um play the role of Nick today. How's that feel, Dave, playing the role of Nick? Always happy to play the role of Nick.

Local Models And Data Access Lessons

SPEAKER_00 1:04

Is that true? Oh, you know, I've been asked for to get people in contact with him, so he's an important person.

SPEAKER_01 1:13

So um before we like jump in too much, um I think you're one of those like I've got all my MacBook Studios and minis, and you've been, I would say, equally nerding out on AI and open claw and all the things.

SPEAKER_00 1:32

So, what have you been working on? Well, I've been pushing co-work to the Macs. I have tinkered around with OpenClaw, haven't haven't got super far far down that wormhole. Although I did just have a Mac mini show up at my house. So that may get taken a little step further. But it's it's I think it's really good. Like the amount that I'm learning just from using local models and understanding the access requirements to data and you know, everything that's important to get a good output, right? Whether it's co-work, open claw, or or really anything, it's really a really a focus on that data. And just seeing the advancements, you know, we started what using Replit a lot at the beginning of last year. And just from that time till today, it's like it's it's night and day, you know, it's it's literally a 10-year difference in normal time to the capabilities of what it's able to do. But a lot of it comes down to what data does it have access to. Yeah.

SPEAKER_01 2:35

So for those of the don't your background is in construction, you worked at Turner Construction and PCI. So even a on-the-ground uh people who build things type job.

SPEAKER_00 2:48

Yep.

IT Roadblocks And Time To ROI

SPEAKER_01 2:48

Not the uh people that talk about building buildings stuff, right? And so if you think about it, if you were in your old job and let's say they actually let you do the things you're doing, which is the big if, right? I think Nick and I have talked about it. Like the number one obstacle to innovation right now and using all these AI tools is IT, right? It's a hundred percent IT, right? So let's say that we didn't have that barrier, right? How do you think you would how much do you think you would have had to invest in terms of time, not equipment and all that? How many hours until you started seeing a payback on productivity, right? Is it because a lot of people don't know they, you know, one, they get on co-work and like, oh, it didn't do what I needed it to do. And it's like, well, you have to invest, right? You have to invest in teaching it and do, you know, it's it's an investment of your energy, but then you know you invest all this energy and then you start to get the payback, maybe not right away, but it's there's an incremental payback you get. But you have to keep investing, you have to keep reinvesting. It's not like you stop reinvesting your energy. So if you're in your old job, let's say you were doing pre-con, right? How many hours do you think you would have had to put in until you started to see some payback of time using these tools?

Precon Workflows Ripe For AI

SPEAKER_00 4:13

Yeah, no, I mean your point's valid because I used to have to call IT to get programs installed on my computer. So definitely the guardrails are preventing that tinkering in the corporate world. But spending, you know, at least a few hours a day just do testing out workflows, seeing what the outputs are, and then refining those. You know, with cowork, I've done a few projects that I, you know, I got them done, but then I realized I should have done it this way, you know, because there's a lot of different skills, there's a lot of different connectors in cowork that could do can, you know, there's multiple different ways to get to the same output, right? It's just a matter of what's the the right way to do it. And I and I feel like that's been, you know, kind of the AI school of hard knocks, right? I went through school of hard knocks for construction for 17, 18 years, and now I'm I'm kind of doing the same thing with AI, and I'm at least spending, you know, I'm I'm using AI all day, every day, but really trying to push it to its limits. I'm probably spending two to three hours just trying to see like how far can I take this?

SPEAKER_01 5:30

Right. So going back to like if you were in a pre-con role, let's say back at Turner, but IT left you alone, right? You got to do whatever you want to do. Yes. Right. What do you think the the day-to-day job, when you looked at your day-to-day job of pre-con, what would you have been?

SPEAKER_00 5:48

So if I had to say how long until I was able to get an ROI, it would be, you know, I think you'd have to probably spend probably two to four weeks just refining that strategy, figuring out, you know, what makes sense, how how to best approach it, and then how to scale it, right? Because a lot, a big part of AI is how how are you going to be able to scale it?

SPEAKER_01 6:14

Yeah. So what's specific? What what do you think would be your fruit on the ground, right? Not the low-hanging fruit, but just the stuff that's just man, in in a pre-con role, which you've had before, what would I be doing?

Bid Leveling As First Automation Target

SPEAKER_00 6:26

So you know, we're you know, when we're doing bid leveling, there's a big, there's a lot of Excel. So you're you're getting a lot of unstructured data from many different subs. I mean, you're you're at least three subs per scope of work. Right. So you think about all the different scope of works times three, sometimes more. I mean, sometimes you'll get 10 bids on a single scope of work. And just having to level those, what's included in this bid versus that bid? Yeah. You know, where's where's the deltas? You know, where are the the scope gaps? Because a lot of your time is spent figuring out why is this number so much lower or so much higher than this number. You know, they're both legitimate companies. Maybe this one excluded a substantial scope of work, and now I've got a big gap. So that number actually is not lower. And now I got to put plug numbers in for other bids that may not have something included. So there's a lot of comparison and analysis. I so I would, you know, looking at it from a pre-com perspective, starting at the bid leveling part, because it's all you know in Excel today. But then you also have more complex workflows when you're starting to kind of break down that bid leveling sheet. What needs to be included? What different things do I need a breakout for? Where's the gaps in the drawings? So you can go pretty complex with the AI implementation within the pre-com process. But if I was going to start today and I was back at Turner doing bid leveling or doing pre-construction, I probably would start at bid leveling and then start to work your ways towards those more complex areas so that you're you're you know injecting AI into the process and then building it outwardly to more complex areas.

SPEAKER_01 8:18

So, I mean, because I think I think about that, right? Pre-con departments are relatively small compared to the rest of the organization, right? In terms of the number of superintendents and number of project you know, it's just the nature of the business that that pre-con departments tend to be, I would say higher levels of expertise, but also like a lot fewer people. Um, hence that becomes a bottleneck. So, like just bid leveling alone seems like to build an agent to do that using cloud co-work, it's probably pretty straightforward, right?

SPEAKER_00 8:52

Yeah. Yeah. I mean, I haven't tested it, but I think that would be a great use case to be able to test it because you can put all the different files into co-work or give it access to that folder where you're storing all of those and then having it build out that bid leveling sheet with all the bids included, you know, based on what I've done with cowork to date, I think that's a that's that's something cowork could probably handle today.

unknown 9:16

Yeah.

AI As A Mentor With Boundaries

SPEAKER_01 9:16

I mean, I I mean, I look at what we're, you know, when we're doing when I'm looking at investments and acquisitions and all that. I mean, the it does quick work of Excel, right? And especially with the Cloud Excel plugin, it's like next level, right? What you can do. So I think that the question there is like, so do you need software for that? No, you just run Cloud Co-Work and you have some energy, right? The thing that I try to remind people too is you know what it gets better. Like the other day, I was doing something, I was doing something in Cloud Code, and it kind of told me I was out of my depth and I should talk to Barry to get some help. Like it knew who Barry was. I was like, oh, it's really getting to learn me. It's really getting to know me, including my limitations, right? So you can imagine a junior person in a pre-con department going through bid leveling and asking questions, and it says, Hey, maybe you should call your boss. Like, you're now out of your depth of knowledge. You need to go talk, you need to escalate this to your boss to review this, which is kind of fascinating, right? Like for a lot of younger people, they don't know when to go to their boss, right? As a boss, you say, like, go figure it out, kid. Like, go, you know, that's what I pay you for, is to figure it out. And then they sit there and they churn on things, but they kind of don't know when to cry, uncle, and go back to their boss and say, Look, I did what I could, I need help now. But I thought it was interesting that AI was prompting me to go get some help because I was out of my depth.

SPEAKER_00 10:41

Yeah, I haven't got any of those, but maybe I need to get the KP and Barry skill file in my co-work instance. I don't think it's because you don't talk. I think I don't think you talked to it enough. I do need to talk to it more.

SPEAKER_01 10:53

You don't talk to it enough, you don't tell it about everything. What you do is you just give it work. If you treat a if you treat AI like a mule, it'll behave like a mule.

SPEAKER_00 11:02

That's a very good point.

SPEAKER_01 11:04

You know, so so anyway, what do you want to talk about today?

SPEAKER_00 11:08

I mean, my I've been really heads down on the building data, you know, how we're delivering buildings today, you know, when when I came over and left the contractor world and came over to kind of really dig into a lot of the pain points, you know, starting with our Q2 research from last year and really just listening and understanding from an owner's perspective. Yeah, you know, where are the downfalls? You know, we're look talking about AI a lot, we're talking about cowork, and then we start to think about how does the owner use their building or how can they use AI within their building? So, which post did you want to talk about? I think your Substack article that really directly talks about the if you if your building could talk, what would it tell you?

SPEAKER_01 11:58

Yeah, it would tell you that you it would tell you that you're not taking care of me. It would tell you that, like, why don't you do the regular scheduled maintenance? Why do you hate me so much? Why do you take me for granted? Is what it would sound like. I mean, don't you think it would be? I mean, I don't think if you had a 10-year-old building and it could talk to you, I don't I don't think it would have good things to say about you.

SPEAKER_00 12:19

Yeah, it's like, hey, my exterior warranty just ran up, but you didn't check all the cult joints, and you know, 40% of them are cracking.

SPEAKER_01 12:28

Right.

unknown 12:28

Right.

SPEAKER_01 12:29

Why don't you it would just basically say that you don't care enough?

If Buildings Could Talk

SPEAKER_00 12:32

Yeah, exactly. I mean, it's you know, and and to the point on on the article, I think the automotive comparison is is a really good one. Yeah. Because you're when you anytime you buy a car, you know you're you're gonna have an organized manual. You know all the data, all the pieces and parts, the information, whatever you need. I mean, it is gonna be there in a consistent manner, regardless of manufacturer, you know, which you could kind of relay the same comparison to construction as you know, Ford is Ford, and you know, maybe Ford is Turner in the real world, yeah, right, in construction. So we don't have any of that consistency in construction to be able to truly leverage all the data that you know that the building could talk to you back to in reference.

SPEAKER_01 13:21

Um it's pretty interesting. A good friend of mine was an executive in New York City, hadn't owned a car like since forever, like literally like since college or something like that, like way back when. Hadn't owned a car, he retired and he moved to Florida and he bought, I think like a Tesla, right? And I was like, wow, and I was like, how does it feel to be driving again, right? Because he hadn't really driven that much. And if he did, it was like a rental car, right? But not owning a car. And said the first thing he did was like spend a day on the owner's manual, yeah. Because he was like, I haven't owned a car in forever, and I haven't owned an electric vehicle ever. Like, I don't know, like, I don't know how this thing works. He's just like, Yeah, I just spent a day like trying to understand how all of it works. So I understood, you know, everything about it, and you know, and that's a car, right? So hand over the keys of a building to somebody, you know, when are they getting trained up on how everything works?

SPEAKER_00 14:25

It it's what I'm finding from all the conversations I'm having with people that are actually operating buildings and and the systems that are operating the buildings, right? Your facility management or CMS platform, whatever you want to call it. They're operating the building before they even have all their building data. And then when they do have their building data, they have to re-break it down and then re-upload it into their system so it's in a way that they can actually use it. And and it's not all complete data, too. It's just if do they do you have fields for these documents? Okay, we got to put them there, but there's still other documents that there aren't spots for within those systems that are still relevant to the building that you're you'll you'll probably need at some point in time. Maybe it's not an everyday thing, but a lot of it could be when there's an emergency. So it's a critical time when that that information is needed.

Why Building Data Lacks Car-Like Consistency

SPEAKER_01 15:16

Well, and then also, I mean, I think that the the real thing, right, is and also the people that were there when the building was commissioned and put online, three, four years from now, they may not be there. All that institutional knowledge, right? Yeah, which institutional knowledge is really honestly just you know, the stuff that people didn't write down.

SPEAKER_00 15:36

Yeah, I remember that's what institutional knowledge really means.

SPEAKER_01 15:40

It's like, oh, the old guy that never wrote anything down, it's all the stuff that he knew and walked out the door with, right? And so, like, so then you think about like employee turnover, and then you know the bigger thing is when you maybe sell the building, and sell the building could be to an absolute buyer, it could be a refinancing, it could be whatever. There's a transaction that happens, it is an asset, they'll probably morph over time, right? It might get refinanced, it might get borrowed against like who knows, right? And I had an uncle when I was in college, I borrowed his car and he said, Look, if you get an oil change, be sure to keep the receipt. And he had a three-ring binder of everything. And when I was like, Why are you so like and this is before you know the dealers had online systems and all that, right? He kept everything in a three-ring binder, and he told me, he's like, This is the difference when I go to sell my car, getting what I want for it versus getting less than that. And it was because I could say, like, here's all my records, right? I've done every oil change, I've done every tire rotation, et cetera, right? I've done all the things. And so I think that's where the rubber meets the road. So you think about that, and my uncle was like worried, he was being kind of a little OCD about like the car. You know, you got to keep track of everything, right? And that was a car. Back then it was probably a$15,000 car, right? So what do we, you know, when we think about that and we think about that this is like value add to the asset value, I don't know quantitatively what it is, but I do know I have friends that buy and sell buildings, and they always say the due diligence part, you know, they the confidence they get when someone has really good records of all the maintenance and all the warranties and everything else definitely gives them more confidence than someone that doesn't.

Operating Before Data And The Cost Of Gaps

SPEAKER_00 17:35

Oh, absolutely. And I've talked to some asset managers and you know, showed them what organized data looks like as opposed to what they're used to. And, you know, some of them have told me like sometimes we'll skip deals. We won't even bring them to investment committee because the building data is so messy that we know we're gonna have to spend a lot more time and due diligence than with a building that is a lot more organized. So, so like where there's maybe not a direct value that you can put on it, I think there's a lot, there's a bigger unknown value where you know I'm trying to sell this massive asset. How many buyers just walked because they knew it was gonna take way too much time? And you know, depending on what the market looks like, there may be plenty of other opportunities for them that they're just gonna move on to. Yeah. So there's a there's a lot of different things like that that are indirectly and directly impacted by how organized your building data actually is.

SPEAKER_01 18:38

So I think like I've written about wrote in my BIM book, I've written about it a lot, like that bifurcation of the of the CapEx side of the ownership and the opex side of it. And even within a within a specific owner, those two parties don't even talk. And and I think like if you think about the AEC industry versus the maintenance, they're not thinking about it, right? And and out, you know, to my architect friends, I think I pick on them all the time. They're not thinking about the operation and maintenance of the building. They're not thinking about what the ongoing costs to maintain. You know, you come up with some fancy curtain wall system, you're not thinking about what it's going to take to maintain that and potential leaks and you know, and all that. And and the contractors aren't thinking about it, right? They're not saying, hey, you know, they think about just the capex, right? It hey, if we substitute this to this, that'll save you money on the overall construction costs. What they don't tell you is like, oh, by going with that cheaper product, you might have a a higher cadence maintenance cycle. So nobody's nobody's really thinking about total cost of ownership. You know, when you were at Turner, like would you did that even ever come up?

Turnover, Transactions, And Lost Knowledge

SPEAKER_00 19:57

It's you know, a lot of the value engineering is is it's all mostly on the CapEx side, right? We have you know$100 million to build this building, and that's that's the only focus. You know, you you you'll you probably get some manufacturers that will sell with that in mind, right? They're they're trying to sell you a product that has a bigger value to the owner. But when it comes to the installers, they're just looking at will this reduce my bid, you know, or or my budget. And in the facilities team, you know, within those value engineering processes, aren't really involved at all. Actually, I can't really remember last time I I mean, I've been talking with a lot of facilities teams now, but back when I was building buildings on the contractor side, we never talked to the facilities teams.

SPEAKER_01 20:47

No, when I was doing some work at Georgia Tech, when I was working there, they'd suck me into like capital projects. And we always brought the facilities management team in way late in the game. And and we'd send them a set of drawings. Well, I would sometimes they get sent drawings earlier and they just would be too busy to look at them. And then when they finally got around to look at them, like all the decisions had already been made.

SPEAKER_00 21:11

Yeah, now you're moving walls and moving pipes.

SPEAKER_01 21:13

And then and then, well, then you're just basically telling them no, sorry, you're gonna have to figure it out, right? You're not you're not moving anything, really. Sorry, right? And and you see that pattern over and over again. I think there's almost this weird dynamic that I think it needs to be solved for, right? How do you connect both sides of that equation? And then also, you know, I was thinking about this like if I was building a new hospital and I wanted to understand like the requirements, it would seem like I would spend time with existing buildings to understand what worked and what didn't work, right? To inform how I design the next one. And when I've talked to architects and engineers about this, they are not spending that much time. They're spending the time with their client in front of them, but they're not going out and saying, hey, we're gonna go spend the next month visiting other hospitals, especially think about hospitals, right? That's not a competitive business. Right. I mean, if you're Walmart, you're like, hey, I'm gonna go look at a bunch of targets, that's different, right? It's competitive businesses. I don't think hospitals view them as themselves as competitors. Right. So if you're the if you're designing a new hospital and you say, like, hey, Mr. Hospital Owner or Mrs. Hospital Owner, I'm gonna go, I'm gonna A visit. I'm going to go spend the next three months visiting other hospitals, talking to patients, talking to doctors, you know, the people that matter, talking to the facilities people, and putting some research together. I'm going to come back to you with my findings in three months, and that's going to inform how we think about designing this building. I don't think that happens.

Organized Records And Deal Confidence

SPEAKER_00 22:58

No. And, you know, doing a lot of hospitals over my career, you're getting a lot of changes late. When they start walking the job and they start seeing the space, you know, getting spatial awareness of what it's actually going to look like. That's when a lot of these changes are starting to come down from the owner. You know, design changes. And even being able to look at your old internal projects, you know, the other hospitals you've built. You're not able to really query that information. You don't have a proper system of record for your other buildings. So you don't have a really good way to go in and you know, maybe it's housed in Procore. Well, that's not Procore servers. It's not on yours. They own it. So as soon as your subscription's gone, your access is gone. And they're not really thinking about the longer life cycle of that building and how they can use that building's data to tell the story. And so that you're not constantly prototyping your next building. You can actually, you know, base it off of knowledge that you've occurred over all the assets that you have actually built and operated. So I think, you know, just outside of operating the building, you know, where that building data really can come in too is on the future development side. No, and I mean think about that, right?

CapEx Vs OpEx And Total Cost Blind Spots

SPEAKER_01 24:23

Even getting the data on like the maintenance cycles. Like if you went and talked to, if you went and met with 50 other hospital owners and their maintenance people and said, tell me which one of the systems here is like not working for you. It's gotten expensive, it's been too many maintenance cycles, blah, blah, blah. And then you'd go to your engineers on your project and say, Don't you dare specify anything that looks like the XYZ system. Because we talked to 20 hospital owners, and half of them hate it. And it's cost them a given them a lot of headaches in terms of maintenance and all that. And I think sometimes there's this weird like reliance. I mean, think about that. Like we're speccing and buying products that go into a building for the next 10, 15, 20 years, maybe longer. And we're not even trying to figure out like what the realities of the maintenance like of it was, right? It'd be like going out and just buying a car and not looking at doing any research on what the issues were and what the maintenance should be. I mean, we would we'd never just go out and buy a new car and say, like, oh, I mean, I'm not gonna do any research. I just I like it, I like the color and its shape, so I'm gonna buy it, right? Or or not knowing how many miles per gallon. Like, would you ever buy a car and not know how many miles per gallon it's giving you?

SPEAKER_00 25:42

I mean, unless there was a test floor.

SPEAKER_01 25:44

Yeah. But then you're then you want to know range.

SPEAKER_00 25:47

Well, then you want to charge range.

SPEAKER_01 25:49

Then you want to charge range, right? So we asked, I mean, and these are baby purchases compared to buying a building, right?

SPEAKER_00 25:55

Yeah, you're 30 to 50,000 compared to multiple millions.

SPEAKER_01 25:59

Right. So, what would happen if you bought a car and they said it would give you 30 miles per gallon and then you drive it for six months and it's giving you five? Yeah, well, and that's you would take it back.

SPEAKER_00 26:13

Yeah, it's false advertisement, right? Yeah, you need to make sure your systems are operating in the way that the design required them to, right?

SPEAKER_01 26:21

But one, we don't even do that, we can't return the building, right? Yeah, hey, sorry, lemon law, I'm giving you your building back, right? But also, like the design teams, nor the build teams are maybe even asked that question. No, or were they held accountable to it?

SPEAKER_00 26:39

And you had a good grocery store example, I think it was last summer, how they kept spec in the same deli coolers, and then the general managers would put them on eBay and buy the new ones, right?

Late Facilities Input And Rework

SPEAKER_01 26:53

Yeah, because they were like, corporate doesn't listen to us, and then I went to corporate, and corporate said, Oh, yeah, we did, but by the time we find out about it, there were already 20 projects already underway. Like we couldn't get them in the plans fast and the specs fast enough. And and so I look, I think in a world where, you know, I wrote and wrote another article the other day about like if AI is creating AI and technology tends to have a deflationary effect, right? How much did you spend at Blockbuster renting movies versus how much are you spending in terms of consumption? Like it's made it cheaper for you to watch movies, right? Music, it's made it tremendously. I remember when I used to pay$20 for a CD. Oh, yeah, to listen to one song, right? Now we pay$20, we get to listen to all the songs, right? So in by its very nature, right? If we look at technology and innovation, that it creates abundance where there was one scarcity, right? That's the basic tenants of when you think about what good innovation and technology looks like. We haven't seen it in construction, but if you think about these little things, these are the little things that actually start to create impact, I think, on like construction costs, facilities management, energy management, all the things that matter. And these are not small numbers.

SPEAKER_00 28:14

No, I mean, just think about RFIs and changes. I mean, that's 10 to 20 percent of the contract value. You you know, you say, well, you're not getting cheaper bids, but if you say, well, you're not getting as many change orders, well, then AI is reducing costs. Yeah, right. Maybe not directly, maybe the contractors are still marking things up more and their costs are less until the market starts to pull it down. But I I think the the the first initial impact, I think, is going to be the improved efficiency on the job site, which is gonna reduce the added costs after GMP is finalized or whatever contract method is being used.

SPEAKER_01 28:55

You mean guaranteed minimum price?

SPEAKER_00 28:58

Yes.

SPEAKER_01 29:01

And you're in all your years at Turner, did you ever see a contingency uh not get used?

SPEAKER_00 29:07

Oh, it's it's getting used like before it shovels in the ground. Right. You know, it's you you're guessing at what what your contingency needs to be, you know, or are we carrying in five percent? You know, what do we need for expediting?

SPEAKER_01 29:22

So let's say you're just picking on Turner because he used to work there. You could be working anywhere else. Do you think if who do you think the first contractor, if any, that figures how to use AI and robots really well and can deliver a building for 30 to 40 percent cheaper? Do you believe any of them will go in and sell it for 30 to 40% cheaper? Or do you think they'll just get 30 to 40 percent more profit?

Learning From Existing Hospitals

SPEAKER_00 29:50

So, you know, on the robotics end, you know, we really just kind of touch the surface with like layout robots and taping robots. Obviously, there's rebar and you know, various other ones. And, you know, looking at like the bigger ones, we've got to have they're gonna have to get buy-in by the unions, right? So the bigger contractors in union markets are gonna have that battle. I was successful in in that in New York City with layout. But working through that with on the bigger side, I think it's it's possible. But I think as more and more of these robotics become more and more capable and start to remove more and more work, I think that's gonna be the first big hill to get over. In terms of like overall AI adoption, you know, we kind of started talking about if I could do this while I was back at corporate. Though the bigger companies, and especially the ones that do federal work, are gonna have a lot harder time being able to scale AI within their organization. I think the mid-size general contractors are probably gonna be able to adopt a lot of these things a lot faster than the bigger ones, right? Because, you know, bigger projects, a lot more people. I mean, if you're a company of 500 people as opposed to 10,000 people, there that's a lot different lift. You know, I experienced switching over an ERP system at a you know,$400 million company. That was a big lift. So now we're thinking about completely changing processes, injecting new technologies. There's, you know, it feels like it's gonna be a little bit slower on the on the more established bigger size companies, but a lot quicker on the smaller companies. Now, what whether those smaller companies are able to kind of overtake the bigger ones, you know, we'll we'll definitely be finding out here soon. But but yeah, I I think there there is a lot of opportunity in the mid-sized contractors to to really step up the game and do things that the bigger ones can't. Yeah.

SPEAKER_01 32:03

All right, Dave. Well, very cool. It's fun having you on.

SPEAKER_00 32:08

Yeah, as always.

SPEAKER_01 32:09

You know, Nick hopefully will be back next week after he uh recovers from the flu. There's a pretty nasty flu going around.

SPEAKER_00 32:16

Yeah, I've been uh locked inside, so I haven't really had to deal with that yet because we've been getting all the blizzards here in Boston. Yeah.

SPEAKER_01 32:25

Yeah, I was just in Philly yesterday. It was like I left like 75 degrees at Half Moon Bay and flew to Philly, and it was like 20 something, but it was like feels like five. And I was like, this is like not this is not what I signed up for. No, I kind of joked with the client. I was like, I usually get hazard pay for this, like this is not something if that's put more than one jacket on, you have to like pay me more. That's how that works.

SPEAKER_00 32:53

Well, it came recently a couple weeks ago, it was even worse, so all right, man.

System Of Record And Reusing Project Data

SPEAKER_01 32:59

Well, very cool, it's good chatting.

SPEAKER_00 33:00

You take that pay.

SPEAKER_01 33:02

Thanks.