Procore Meters AI, PE Buys the Concrete, and SoftBank Bets $200M on Robots - August 2026 Reviewed as Noted の全文書き起こし

today so i'm kind of hoping 00:02:04.276 --> 00:02:07.346 my kid grows up a little bit to where we can hang out and go do things and stuff 00:02:07.346 --> 00:02:10.216 like that so you know i'll give him 18 months and then we'll be best friends 00:02:10.656 --> 00:02:14.836 can do things i was told like first nine months like it's just kind of a you 00:02:14.836 --> 00:02:19.556 know it's just there but then once nine months hits like he's they start responding 00:02:19.556 --> 00:02:22.856 to you and and that i think i think it's a little bit earlier than that, 00:02:23.316 --> 00:02:27.544 it's six months five months well large was apparently a really good dad. 00:02:29.196 --> 00:02:33.856 Nothing to do with being good or bad it's just that first time where they actually 00:02:33.856 --> 00:02:38.892 like look at you and like recognize you and smile like yeah that's why i'm waiting for that yeah, 00:02:39.656 --> 00:02:44.296 there's also a study done that i found interesting which is in new dads 00:02:44.736 --> 00:02:50.176 your testosterone just drops and and the whole idea is like they want to like 00:02:50.176 --> 00:02:55.206 the idea is like for you to focus on your family like just go home and nest 00:02:55.206 --> 00:02:56.676 a little bit yeah i don't know, 00:02:57.156 --> 00:03:01.336 so so lean into that yeah now well i'll tell you this i've surprisingly been 00:03:01.336 --> 00:03:04.116 working out more than i did before yes so we'll say that. 00:03:05.056 --> 00:03:06.836 That's amazing You're bucking the trend. 00:03:07.436 --> 00:03:10.106 I mean, I just can't sleep. And so like, if I'm up, I might as well do something 00:03:10.106 --> 00:03:12.335 semi-productive. And I, I, I. That's fair. 00:03:12.956 --> 00:03:17.232 Well, now that we've lost all of the listeners who are bored by our stories, 00:03:17.232 --> 00:03:20.672 we've got some interesting stories from this last month. We've got a number 00:03:20.672 --> 00:03:21.852 of things. Let me go through them here. 00:03:22.247 --> 00:03:27.342 We've got the kind of AGC and an entity called the NCCER. 00:03:27.662 --> 00:03:31.402 I don't even know who that is. But they published a workforce survey. 00:03:31.402 --> 00:03:34.790 So the market is soft, but we are still kind of short on people. 00:03:35.202 --> 00:03:38.142 We know data centers are kind of eating the labor supply. There's more data 00:03:38.142 --> 00:03:39.658 that suggests that's still true. 00:03:40.006 --> 00:03:44.622 We've got Private Equity, a company named Command Alcon, bought the software 00:03:44.942 --> 00:03:49.933 that manages the kind of dispatch systems for a lot of America's concrete companies. 00:03:50.803 --> 00:03:58.592 We've got a person named Anirban Basu who did the math that some of the census data wasn't picking up. 00:03:59.192 --> 00:04:02.892 If you strip out the data centers and private non-residential, 00:04:03.292 --> 00:04:07.492 or sorry, strip out data centers and private

non-residential construction is 00:04:07.492 --> 00:04:10.852 down about 8% year over year, which I think we've all kind of felt. 00:04:10.852 --> 00:04:13.626 But this is maybe the first time that's kind of showing up in the data. 00:04:14.090 --> 00:04:20.184 And then we've got- Hey, I'll just say, you have not heard an Anirban presentation, you're missing out. 00:04:21.914 --> 00:04:26.543 Dude so great he makes this really dry topic, 00:04:27.162 --> 00:04:31.252 so interesting because he just he does the best slide decks I haven't seen him 00:04:31.252 --> 00:04:35.242 since AI came around so maybe he's throwing in some weird AI generated imagery 00:04:35.242 --> 00:04:40.212 now but anyway I kind of wish you mentioned him as a dinner guest if you haven't had, 00:04:40.752 --> 00:04:45.059 Anurbozu on for dinner that would he lives in my basement, 00:04:46.532 --> 00:04:48.943 right He's a good buddy. 00:04:49.493 --> 00:04:53.752 A few more stories. We've got Procore kind of rolling out some kind of vague 00:04:53.752 --> 00:04:57.262 reference to how they're going to charge for AI by the credit, 00:04:57.262 --> 00:05:00.340 but not really going into what that entails. 00:05:00.938 --> 00:05:04.972 And then outside the industry, we've got a number of bottles shipping almost 00:05:04.972 --> 00:05:07.184 weekly, it feels like. We've got Fable 5.1. 00:05:07.544 --> 00:05:11.118 OpenAI just shipped GPT-6 named Astra. 00:05:11.594 --> 00:05:17.062 And they've got all kinds of capabilities. And then AGC, they said there was 00:05:17.062 --> 00:05:21.172 an increase in cost in terms of inputs, but then another big acquisition, 00:05:21.543 --> 00:05:24.992 or not acquisition, but a funding check. 00:05:25.312 --> 00:05:29.744 SoftBank wrote a $200 million check to Gratis Robotics, which is pretty interesting. 00:05:31.091 --> 00:05:34.877 So that was a very poorly kind of read through overview of all the stories we're 00:05:34.877 --> 00:05:37.827 going to get into. I think one thing that would be interesting to start with 00:05:37.827 --> 00:05:40.557 would be kind of the workforce survey. 00:05:40.957 --> 00:05:44.277 So softer market, but still short on labor. 00:05:44.577 --> 00:05:48.187 I guess, Victor, do you want to kind of start with this one and just kind of 00:05:48.303 --> 00:05:52.430 give us an overview on kind of where we're at, what's entailed there? 00:05:53.005 --> 00:05:56.437 Yeah, the interesting thing about the softening of labor is, 00:05:56.737 --> 00:05:58.537 one, we all kind of know this is happening, right? 00:05:58.537 --> 00:06:01.927 Like, I shouldn't say that. In a lot of conversations we've had with various 00:06:01.927 --> 00:06:04.890 people, various leaders of different construction companies across the U.S., 00:06:05.250 --> 00:06:11.267 outside of key data center and energy work, I think there's been a softening felt by contractors. 00:06:11.267 --> 00:06:14.047 I think across the board, it's probably been like 15, 20%, which is material. 00:06:14.047 --> 00:06:17.167 But this is coming off some pretty high highs over the past few years. 00:06:17.167 --> 00:06:20.067 The backlog was really large. And so for most of these companies, 00:06:20.067 --> 00:06:22.557 it's honestly a reversion to the mean. It's not that surprising. 00:06:22.557 --> 00:06:26.037 I think on the labor side, the one thing that's unique is that because the data 00:06:26.037 --> 00:06:28.922 center jobs are so, so massive, they're sucking up all of the talent. 00:06:29.190 --> 00:06:32.237 They're sucking up all these people and so even if you do have a job you're 00:06:32.237 --> 00:06:35.387 looking at a contractor or or a team of people that's going to be a little bit 00:06:35.387 --> 00:06:37.893 different than you would have a year ago two years ago three years ago, 00:06:38.427 --> 00:06:41.647 so i think that's also being felt and so you have these like micro pockets across 00:06:41.647 --> 00:06:45.540 the industry and then you have geography based softening as well and so, 00:06:46.109 --> 00:06:49.267 those three things are pretty unique the other thing too is like interest rates 00:06:49.267 --> 00:06:51.787 are still really high and so the types of products that are really sensitive 00:06:51.787 --> 00:06:55.077 to interest rates which are mostly commercial work residential a lot of that 00:06:55.077 --> 00:06:59.280 is going to be slowing down or really honestly at a pause until we see some sort of rate cut there. 00:06:59.890 --> 00:07:02.717 And then that gets into a larger economic picture and so like well what caused 00:07:02.717 --> 00:07:05.457 the rate cut how do we think about that in the future so until then you have 00:07:05.457 --> 00:07:09.277 a lot of things that are not sensitive to that which are data center work because 00:07:09.277 --> 00:07:12.297 the hyperscalers are willing to pay whatever it takes energy because they need 00:07:12.297 --> 00:07:14.494 it and we all know we need more energy in america 00:07:14.877 --> 00:07:17.164 and so you have you go down the list and of course health care because, 00:07:17.802 --> 00:07:21.397 meds and beds never runs out of business and so i think out of those four core 00:07:21.397 --> 00:07:24.107 areas i think you'll probably see some softening until we see a lot of movement 00:07:24.107 --> 00:07:25.680 on the on the interest rate side. 00:07:27.101 --> 00:07:32.028 So does this just mean that the kind of diversification issue for contractors 00:07:32.028 --> 00:07:37.033 is going to get even worse as people kind of lean into where there is work, which is data centers? 00:07:37.785 --> 00:07:41.958 Yes and no. I think if you think of concentration as what type of work are you 00:07:41.958 --> 00:07:44.618 taking on? I think, yeah, so that's naturally going to happen because it's just 00:07:44.618 --> 00:07:46.405 going to be fewer commercial jobs out there. 00:07:46.777 --> 00:07:49.988 However, I think

that as we've seen in other ways and other contractors of how 00:07:49.988 --> 00:07:52.368 they actually make money in their business lines, there's ways that you can 00:07:52.368 --> 00:07:55.206 make more profit, even if you're taking on lower top line revenue. 00:07:55.542 --> 00:07:57.928 And so I think you got to get smart about that. So you think about your costs, 00:07:57.928 --> 00:08:00.878 you think about your different levers within the business. So yes, 00:08:00.878 --> 00:08:04.038 you'll see less diversification on the front end of the top line of revenue, 00:08:04.038 --> 00:08:08.344 but you should, in theory, see more diversification of how contractors actually make more money. 00:08:08.681 --> 00:08:11.258 What I mean by that is like, instead of just charging a single fee for running 00:08:11.258 --> 00:08:14.955 the job, you'll probably see more insurance work. You'll probably see more self-perform work. 00:08:15.251 --> 00:08:17.927 You'll probably see a lot more of this other stuff that contractors do do. 00:08:18.855 --> 00:08:23.868 Yeah. So forcing to be a little bit more creative in terms of revenue generation. 00:08:23.868 --> 00:08:27.678 So as far as maybe you're building in the federal sector, are you seeing kind 00:08:27.678 --> 00:08:31.608 of federal spending? Obviously, federal deficit spending is a big part of like 00:08:31.608 --> 00:08:33.438 big input into the economy. 00:08:34.158 --> 00:08:36.762 Are you seeing any upward downward trend or is it pretty even? 00:08:37.704 --> 00:08:43.430 Continuing to see a lot of money spent over the Department of Defense, 00:08:43.430 --> 00:08:45.383 Department of War, however it wants to identify. 00:08:45.888 --> 00:08:49.800 Someone yesterday told me that right now what's actually sucking up a lot of 00:08:49.800 --> 00:08:55.410 the defense spending is the wall, which I think, believe it or not, 00:08:55.410 --> 00:08:56.649 is sort of flying under the radar. 00:08:56.950 --> 00:08:59.620 They're not really promoting it. They're not really talking about it. 00:08:59.620 --> 00:09:02.530 I haven't validated or verified that, but I did see yesterday on like, 00:09:02.890 --> 00:09:07.740 I think it was GovWin or GovTribe, where it was one of those spending metrics 00:09:07.740 --> 00:09:13.588 or diagrams. And it was a large single color under a single customer. 00:09:14.088 --> 00:09:18.280 And again, that's probably, don't remember that contract being let, 00:09:18.280 --> 00:09:22.602 don't know when it was warded or how it was turned back on, but probably in the cover of darkness. 00:09:23.159 --> 00:09:26.850 But all in all, I'm continuing to see a fair bit of spend. 00:09:26.850 --> 00:09:31.230 And but it's not your I'd say your typical. 00:09:31.230 --> 00:09:35.550 It's continuing to be health care. It's continuing to be power and infrastructure 00:09:35.550 --> 00:09:38.970 as well as security, skiffs, things like that. 00:09:38.970 --> 00:09:44.047 A lot of

the hardening around buildings specifically to. 00:09:45.174 --> 00:09:51.900 You know, communications, ICD-705, things like that, which all relate to high 00:09:51.900 --> 00:09:56.657 secure environments, you know, so sensitized, compartmentalized something, infrastructure. 00:09:56.988 --> 00:10:00.720 All right. Sponsor break. Lars, real quick. When was the last time you got a 00:10:00.720 --> 00:10:05.136 new set of drawings from an architect and they didn't cloud anything? 00:10:05.507 --> 00:10:10.453 Well, the one that cost me a couple hundred grand was back, I think, in 2017. So... 00:10:11.621 --> 00:10:15.210 Lesson learned there. Yeah. So, you know, what I like about what Trunk Tools 00:10:15.210 --> 00:10:21.910 is building is they're looking at tools to help kind of remove the mundane from the job site. 00:10:21.910 --> 00:10:26.560 And, you know, same as you, every time there was a bulletin or an addendum, 00:10:26.880 --> 00:10:30.420 I would go back to designers and ask them to at least highlight or give me a 00:10:30.420 --> 00:10:31.480 cover letter on the changes. 00:10:32.120 --> 00:10:35.190 And that was always like pulling like teeth, but Trunk Tools is now automating 00:10:35.190 --> 00:10:39.197 that. So you get a bulletin come in, it automatically highlights the changes. 00:10:39.586 --> 00:10:40.398 You don't have to do a thing. 00:10:40.915 --> 00:10:44.160 Well, it's also hitting RFIs, right? It's also hitting submittals. 00:10:44.640 --> 00:10:48.745 And when we think about standardizing ops, right, and providing that, 00:10:49.267 --> 00:10:55.340 you know, sort of resource to those field operators, what I see it as, you know, 00:10:56.003 --> 00:11:01.460 it's sort of like this, like, mentor within the company that structure standardizes, 00:11:01.622 --> 00:11:04.560 keeps it moving in the right direction, looks out for risk. 00:11:05.600 --> 00:11:10.394 And just this last week, I had a QC manager just pass along a submittal to an owner. 00:11:11.037 --> 00:11:13.760 Without doing a review because he was too busy and was wrong. 00:11:14.915 --> 00:11:19.182 And, you know, the classic owner statement is, did you even review this? Right. 00:11:19.792 --> 00:11:23.937 So this to me is carte blanche. We absolutely did. 00:11:24.309 --> 00:11:27.925 And think about getting that first pass before your, your, your, 00:11:27.925 --> 00:11:31.885 your, your, your field manager ever even actually has to do the first pass. 00:11:31.885 --> 00:11:35.721 Someone's doing it for him. So to me, it's about speed. It increases accuracy 00:11:36.041 --> 00:11:38.328 and helps standardize ops out in the field. Right. 00:11:38.815 --> 00:11:42.135 Yeah. And this is the kind of ai that we've we've talked about on the podcast 00:11:42.135 --> 00:11:46.055 really appreciating right which is looking at ways to elevate the craftsman 00:11:46.415 --> 00:11:52.295 at the heart of project delivery give me time back so i can be out in the field managing risk right, 00:11:52.855 --> 00:11:58.833 yeah 100 so all right if if you're a little curious check it out trunk tools.com forward slash podcast, 00:11:59.413 --> 00:12:03.615 and if you mention construction conversations you get a discount discount demos 00:12:03.615 --> 00:12:06.695 are free and we get and you guys get a discount so let's go 00:12:07.115 --> 00:12:12.655 so So Tyler, you're interacting with contractors from a data strategy and an 00:12:12.655 --> 00:12:15.925 AI implementation standpoint, as well as I think a couple owners. 00:12:15.925 --> 00:12:23.364 Are you seeing any of this force people's hand in terms of how they try to maybe 00:12:23.707 --> 00:12:28.565 kind of codify some operational efficiencies to respond to these different external challenges? 00:12:28.565 --> 00:12:32.380 Like are people focusing more on the internal because that's a variable they can control? 00:12:33.010 --> 00:12:36.915 Yeah, they're focusing a lot on internal right now and a lot of internal development. 00:12:37.740 --> 00:12:39.023 What's interesting is that, 00:12:39.743 --> 00:12:44.080 people are trying so bloody hard, 00:12:44.689 --> 00:12:48.714 to create their know their processes in-house they're probably using things 00:12:48.714 --> 00:12:52.892 like claw creating skills they're doing a bunch of stuff with ai there's a massive 00:12:53.310 --> 00:12:57.728 mandate for people to start leveraging ai inside of these companies, 00:12:58.180 --> 00:13:02.324 but what's funny is that with all of these point solutions that are being created 00:13:02.324 --> 00:13:06.274 just within the claud ecosystem itself right so it's like i create a skill and 00:13:06.274 --> 00:13:09.314 then victor creates a skill and then lars creates a skill it's steven you create 00:13:09.314 --> 00:13:11.648 a skill everybody has a skill file, 00:13:12.309 --> 00:13:18.324 that does the same workflow so there's no like unification happening internally 00:13:18.324 --> 00:13:22.844 too so at the same time that there being a ton of innovation and trying to find 00:13:22.844 --> 00:13:24.441 this operational efficiency. 00:13:25.428 --> 00:13:30.804 We're also getting way less efficient as and way more confused because we're 00:13:30.804 --> 00:13:32.963 able to create so quickly 00:13:33.328 --> 00:13:39.064 right so we're able to create a new product so fast that we have five people 00:13:39.064 --> 00:13:42.952 internally across an organization of, you know, let's say a hundred 00:13:43.375 --> 00:13:46.614 that are each creating their own version of the same sort of thing. 00:13:46.614 --> 00:13:50.527 And then it just gets muddy. Then it becomes a political issue internally. 00:13:50.916 --> 00:13:55.204 So it's kind of a weird time because everybody's able to be that kind of creator, 00:13:55.204 --> 00:13:58.751 that craftman, a crafter of an automation. 00:13:59.233 --> 00:14:00.702 And there's this, 00:14:01.550 --> 00:14:05.495 There's a

lack of understanding and oversight in a lot of organizations to say, 00:14:05.495 --> 00:14:11.030 okay, great idea, but how do we make this scalable for the entire organization, not just for you? 00:14:11.682 --> 00:14:15.505 And so, yeah, there's a push for a lot of operational efficiency around a lot 00:14:15.505 --> 00:14:17.895 of this stuff. And I don't know, we'll see how it shakes out. 00:14:17.895 --> 00:14:20.803 But like Tyler talking, I can't just have it. 00:14:21.314 --> 00:14:25.285 Wow. To your last point, I'm curious, like whose responsibility is that to make that decision? 00:14:25.830 --> 00:14:28.345 It's the visionary leader in the organization. It's the CEO. 00:14:28.515 --> 00:14:31.060 It has to be. It's got to be the owners. It has to be. 00:14:31.681 --> 00:14:37.811 It has to be, but so often I'm seeing that stuff get delegated and then it atrophies. 00:14:38.049 --> 00:14:39.835 I was going to say, and then where does it go? 00:14:42.195 --> 00:14:47.941 And I see so often it dies. It dies. People make stupid decisions around that too. 00:14:49.695 --> 00:14:52.120 For every two people that like it, there's eight people that don't. 00:14:52.515 --> 00:14:56.909 But the two people and the owner is in this sort of like wishy-washy category. 00:14:57.368 --> 00:15:00.695 And so it's like, oh yeah, it sounds interesting. Go for it, right? 00:15:00.695 --> 00:15:03.945 And they're like they create this thing but all the other eight people are like 00:15:03.945 --> 00:15:08.115 yeah i'm not with that you know chuck says i don't do it i'm not doing you know it's always chuck, 00:15:09.035 --> 00:15:15.805 so laurage you're that it does how are you managing that they're okay story. 00:15:17.455 --> 00:15:20.297 No it's it's it's fired three people last week, 00:15:21.555 --> 00:15:30.826 no no no uh no hiring in fact in fact i'm i'm hiring an implementer type where as, 00:15:31.955 --> 00:15:38.259 so we're juxtaposed with the old way the new way and now i don't even know what 00:15:38.386 --> 00:15:40.284 you call this but it's a fuel, 00:15:40.913 --> 00:15:45.745 and the the rate of burn is insane it's just like we can go to zero to 100 in 00:15:45.745 --> 00:15:51.742 like that right so this change rate is just nuts so. 00:15:52.695 --> 00:15:55.289 In in in a specific instance it's like, 00:15:56.120 --> 00:16:02.455 I am purposely trying to be very mindful of like, how can I do this with as 00:16:02.455 --> 00:16:04.382 few people as possible in the office? 00:16:05.020 --> 00:16:07.690 Totally tip of the spear in the field. 00:16:08.441 --> 00:16:12.225 And that inherently is creating some challenges for how fast we're moving and 00:16:12.225 --> 00:16:16.235 winning work and executing work. Case in point is just like invoices. 00:16:16.798 --> 00:16:21.125 Literally, we're slow on paying stuff. Not because of cash. 00:16:21.896 --> 00:16:25.755 It's how it's coming in. And there's solutions out there. and I've dealt with 00:16:25.755 --> 00:16:30.214 Paperless and Sage and then Nexonia and all these other things. 00:16:30.533 --> 00:16:35.385 But at the end of the day, it's like, wait a minute, how can we do this in this 00:16:35.725 --> 00:16:38.125 future state where we have our own portal? 00:16:38.299 --> 00:16:44.205 How many bots can I have processed, come in, put in collaborative inbox, we're watching it move. 00:16:44.205 --> 00:16:48.865 It's already filtering out, putting into Paperless so that some structured approval 00:16:48.865 --> 00:16:52.065 workflow. So now I'm not creating software for nothing, right? 00:16:52.065 --> 00:16:54.692 To integrate with Sage and a payment plan. 00:16:54.977 --> 00:17:01.764 But yet, how can I move thousands of invoices a month through a very, 00:17:02.240 --> 00:17:07.155 call it small internal team with no human typing each invoice, right? 00:17:07.155 --> 00:17:13.485 So I'm right now, the old way is, but that's my job. That's what I do. 00:17:13.785 --> 00:17:16.050 It's like, time out. I know it's what you do. 00:17:17.240 --> 00:17:21.316 And you can handle it right now. But the moment I go win two more jobs and you 00:17:21.316 --> 00:17:23.753 handle it, no, we're going to have to hire someone. Exactly my point. 00:17:24.901 --> 00:17:30.956 That's my point. And it was like, oh, now granted, with all that we're dealing 00:17:30.956 --> 00:17:36.324 with from a growth standpoint, there's this like, well, is that the right priority? 00:17:36.905 --> 00:17:40.846 Problem is, even, and Tyler and Victor, I think to this collaborative thing, 00:17:40.846 --> 00:17:45.736 which is It's like, how many priorities can you possibly have right now when 00:17:45.736 --> 00:17:49.356 you can sort of tackle them all simultaneously? 00:17:49.582 --> 00:17:53.026 But are you actually getting anything done in this world of like, 00:17:53.026 --> 00:17:54.505 well, I can improve this, I can improve this. 00:17:54.963 --> 00:17:57.489 But are you actually solving it? Are you getting it better? 00:17:58.556 --> 00:18:03.066 What's happening right now, dude, is that I find myself almost acting as an 00:18:03.066 --> 00:18:07.857 intentional bottleneck to innovation in some of these. To slow things down. 00:18:08.710 --> 00:18:12.906 Because you have to slow down in order to speed up. In some organizations, 00:18:12.906 --> 00:18:14.736 it depends on it, on the organization. 00:18:14.736 --> 00:18:17.656 Sometimes they'll innovate themselves to absolute freaking confusion, 00:18:18.570 --> 00:18:22.865 right? And so that, a lot of times, you have to stand in there and say, 00:18:23.528 --> 00:18:26.906 well, slow down, understand the problems that we're trying to solve first, 00:18:26.906 --> 00:18:29.276 and then we can tackle this thing super fast. 00:18:29.729 --> 00:18:32.106 But we got to make sure that there's return on that investment, 00:18:32.106 --> 00:18:34.860 return on that time that we spend putting that thing in

place. 00:18:35.527 --> 00:18:40.506 But what you said, too, capacity being probably the biggest word that I'm hearing 00:18:40.506 --> 00:18:49.026 thrown around amongst founders of construction companies, I want more capacity for my people. 00:18:49.026 --> 00:18:51.320 I don't necessarily want to hire more people. 00:18:52.312 --> 00:18:56.128 I want to give capacity to the people that I already have. And then if I need 00:18:56.128 --> 00:18:57.758 to hire, then I can. Right. 00:18:58.575 --> 00:19:02.790 It sets me up. That's later on. Slowing down before you are right now. 00:19:03.893 --> 00:19:07.458 Sponsor break. If you spend any time running a project, you know the answer 00:19:07.458 --> 00:19:10.196 is almost always sitting in the contract document somewhere. 00:19:10.682 --> 00:19:16.348 Catch is that it's almost never in one document. It's split across the contract, 00:19:16.348 --> 00:19:19.948 the specs, and some flow down clause three layers deep that nobody connects 00:19:19.948 --> 00:19:21.747 until it's already cost money. 00:19:21.939 --> 00:19:25.808 That's the whole premise behind Document Crunch. That risk doesn't live inside 00:19:25.808 --> 00:19:28.539 any single document. It lives in the gaps between them. 00:19:28.760 --> 00:19:33.247 And most teams never look at all of them together until they're standing in a dispute. 00:19:33.944 --> 00:19:38.518 The average construction dispute in the U.S. now runs about $60 million, 00:19:39.209 --> 00:19:41.572 according to Arcadis. It's a lot of money. 00:19:41.978 --> 00:19:45.948 And the root cause they point to isn't fraud or bad intent. It's people not 00:19:45.948 --> 00:19:47.580 fully understanding what they agreed to. 00:19:47.789 --> 00:19:50.645 So Document Crunch builds a platform around closing that gap. 00:19:50.859 --> 00:19:56.473 Crunch.ai reads the documents, surfaces the risks, and it cites the exact page and clause that's big. 00:19:56.816 --> 00:20:00.658 So you're not just taking its word for it. Project Assist lets your team act 00:20:00.658 --> 00:20:03.898 on it, ask questions across the entire project document set, 00:20:03.898 --> 00:20:08.278 hand-draft your submitals, notices, and RFIs right from the chat window, 00:20:08.499 --> 00:20:11.518 and the whole thing runs on your own company standards. 00:20:11.518 --> 00:20:16.272 The office and the field are working off the same playbook for bid through closeout. 00:20:16.556 --> 00:20:21.558 No disconnects. It's live now across the U.S. and Canada. Go to documentcrunch.com 00:20:21.558 --> 00:20:24.921 forward slash construction conversations and book a demo. 00:20:25.918 --> 00:20:29.668 I think it's a good segue to talk about the models because I think what's also 00:20:29.668 --> 00:20:34.558 driving some of this is every time I sit down to do something with one of these 00:20:34.558 --> 00:20:37.714 models, the capabilities are increased. 00:20:37.882 --> 00:20:41.678 And not just a little bit, in some cases a lot. And so, you know, 00:20:41.678 --> 00:20:46.248 this week we got Fable 5.1. I've been playing with Fable 5 for quite a while 00:20:46.248 --> 00:20:50.665 and it's very good in some ways. and it's a little expensive than others. 00:20:51.429 --> 00:20:56.247 I think OpenAI's kind of Astra 6, that's not out yet, right, Victor? 00:20:57.937 --> 00:21:05.557 Not to the general public yeah just the some of the elite Victor do you have a copy of it I do not. 00:21:07.357 --> 00:21:13.377 I do not maybe I do not but I think if like even you've got this model from 00:21:13.377 --> 00:21:19.917 the world lab Atlas the world model have you seen this Tyler sounds familiar it is so good, 00:21:20.645 --> 00:21:26.047 what is it it's a world model and it was built by I can't say her name but she 00:21:26.047 --> 00:21:32.170 was from open ai's founding team i believe and it gives you the ability to, 00:21:32.710 --> 00:21:38.267 provide snapshots like simple you know static images of an environment and it 00:21:38.267 --> 00:21:42.014 creates a 3d environment around that image, 00:21:42.577 --> 00:21:47.187 and so like i said this internally to my team like what does this do for our 00:21:47.187 --> 00:21:51.685 scan to bim deliverable like it should inform it in some degree obviously there's some some constraints. 00:21:52.501 --> 00:21:58.856 But people are creating like entire 3D worlds from a simple image and it's just 00:21:59.210 --> 00:22:01.127 simply incredible. Yeah. 00:22:01.527 --> 00:22:05.197 I mean, I think that this is one of those like there's so many there's so much 00:22:05.197 --> 00:22:06.467 to impact in the last month. 00:22:06.467 --> 00:22:09.987 This is one of those big things that, quite frankly, is should be larger, 00:22:09.987 --> 00:22:13.497 particularly in our industry. Steven, you mentioned like the BDC world, 00:22:13.497 --> 00:22:16.187 I think, is going to be felt by this tremendously. I think the other side is 00:22:16.187 --> 00:22:17.667 going to be pursuits and pre-con. 00:22:18.027 --> 00:22:20.257 Some of the things that you try to show an owner of like what is the building 00:22:20.257 --> 00:22:22.130 going to look like over time. This is going to be massive. 00:22:22.682 --> 00:22:26.157 The other thing that I think is underappreciated, under talked about, 00:22:26.157 --> 00:22:27.767 though, is like the negative side of this. 00:22:28.127 --> 00:22:32.017 So in the new models, in the new world, we have new capabilities coming out. 00:22:32.017 --> 00:22:35.817 But historically, what I've seen is a lot of people trying to apply the new 00:22:35.817 --> 00:22:39.011 way of working or the new model to the old way of working, rather, 00:22:39.499 --> 00:22:40.417 but not much is changing. 00:22:40.417 --> 00:22:44.117 And so like email stay the same, you know, document, all that stuff stays the same. 00:22:44.117 --> 00:22:47.777 But in this, like Atlas is a great example of why can't I be

able to take a 00:22:47.777 --> 00:22:51.907 couple of pictures on a job site and then stitch together a model and then send 00:22:51.907 --> 00:22:56.788 that over to a team that I want to or help that inform some other sort of like 3D rendering. 00:22:57.270 --> 00:23:02.717 The other thing that I think is personally scary about this is this could be 00:23:02.717 --> 00:23:05.557 fake. Like there's so much that could be faked with this. And I think about 00:23:05.557 --> 00:23:08.337 like progress reporting. I think about documents. As long as you talk about 00:23:08.337 --> 00:23:09.317 invoices, like you think about 00:23:09.317 --> 00:23:11.485 all these different things that like prove what happened in the field. 00:23:11.862 --> 00:23:15.507 I could fake an image way easier today than I could yesterday because of this 00:23:15.507 --> 00:23:20.957 stuff so I think this is one of those like super exciting super scary I shouldn't 00:23:20.957 --> 00:23:23.761 even say scary it's not scary it's just it's an unknown it's an unknown. 00:23:24.937 --> 00:23:28.177 Can I want to ask a question and I got to be careful of this because I almost 00:23:28.177 --> 00:23:33.944 feel like this is an entirely different you know session but open source models, 00:23:34.610 --> 00:23:36.682 versus closed I'm not going to be able to do that. 00:23:37.385 --> 00:23:41.971 Specifically relating to controlling what you just described. 00:23:42.617 --> 00:23:49.592 And when we think about truth, security, do you think it's going to head in 00:23:49.592 --> 00:23:52.287 this open world, source models? 00:23:52.881 --> 00:24:02.492 And if so, how do we, as a collective species or call it industry, 00:24:02.492 --> 00:24:05.406 control the truth around what you just talked about? 00:24:06.117 --> 00:24:09.372 Is that the way to do it? I don't know. Yeah, I think truth is something I've 00:24:09.372 --> 00:24:10.652 honestly wrestled with so much. 00:24:11.372 --> 00:24:14.032 I don't even know that universal truth really exists in this industry outside 00:24:14.032 --> 00:24:17.132 of what actually happened on a project. And so it's so hard to even... 00:24:17.312 --> 00:24:19.382 I'll give you an example. I will answer your question, Lars. 00:24:19.382 --> 00:24:25.540 But you know if a piece of concrete failed a slump test. You know if the steel beam was installed. 00:24:25.767 --> 00:24:28.992 However, when someone goes and submits a pay application, they may submit what 00:24:28.992 --> 00:24:30.660 happened plus the materials I've already ordered. 00:24:30.985 --> 00:24:33.982 You look at a schedule, it may say what's shown up. You look at all these different 00:24:33.982 --> 00:24:38.212 things versus what is truth And it's not always been clear to me. 00:24:38.212 --> 00:24:40.872 You ask three people how far along a project you are. It's not always clear. 00:24:41.156 --> 00:24:42.683 So I think that's part one. 00:24:43.164 --> 00:24:45.932 And I think if

you apply that now with all these new models and all this AI 00:24:45.932 --> 00:24:47.882 that everyone's talking about, it's only going to get more blurry. 00:24:47.882 --> 00:24:52.683 And so you think about like OpenAI, Anthropik, Grok, whatever against like open source models. 00:24:53.008 --> 00:24:56.452 To me, it's honestly, it's a nothing burger. I think the open source models 00:24:56.452 --> 00:24:59.692 are, at least for construction, these are great tools for people that have the 00:24:59.692 --> 00:25:01.896 requisite teams that should go build the solutions they need. 00:25:02.157 --> 00:25:05.733 And I'm not convinced a lot of contractors can or quite frankly should be doing this. 00:25:06.081 --> 00:25:09.682 And so to me it's like it exists it's out there but it'd be like everyone wants 00:25:09.682 --> 00:25:13.002 to run a race like just go to the car dealership and buy a car like no one wants 00:25:13.002 --> 00:25:15.862 to build a car like that's not the skill set of what the general contractors 00:25:15.862 --> 00:25:17.232 are doing like just go buy a car. 00:25:18.273 --> 00:25:25.216 Yeah, I mean, I think a lot of it's, there's a view of the world that I think 00:25:25.576 --> 00:25:28.345 open source is good for many reasons. 00:25:28.472 --> 00:25:34.860 Oh, for sure. In that it, you know, can drive a more open variety of kind of innovation. 00:25:34.860 --> 00:25:40.090 That said, I do think the prevalence of open source models has more to do with 00:25:40.090 --> 00:25:42.860 kind of the trade war with China than anything else, right? 00:25:42.860 --> 00:25:48.370 Like they're kind of throwing open source models to kind of defund a key pillar 00:25:48.370 --> 00:25:50.110 of our current economic growth. 00:25:50.110 --> 00:25:55.530 So, but I think to your point, like if it's competition and we, 00:25:55.530 --> 00:26:00.527 in theory, believe in that in the West. And so, yeah, respond accordingly, 00:26:00.632 --> 00:26:02.176 Anthropic and OpenAI, right? 00:26:03.230 --> 00:26:04.629 I also think those guys are stand-by again. 00:26:06.110 --> 00:26:10.440 Do you think I hope you would be able to keep up with an Anthropic or an OpenAI 00:26:10.440 --> 00:26:11.775 though? Six months behind. 00:26:12.383 --> 00:26:14.350 And that's kind of what I was thinking. Yeah, it's like, okay, 00:26:14.670 --> 00:26:21.690 fine. Didn't Kimmy K3 like kind of do that? I mean, maybe it like met Fable 00:26:21.690 --> 00:26:25.160 5's benchmarks like three weeks after it was released. 00:26:25.982 --> 00:26:28.270 But we normally do this on the first story or the second. 00:26:29.530 --> 00:26:31.885 We're off topic, but this is interesting. 00:26:34.644 --> 00:26:38.100 Go ahead, Lars. I was going to say where I think it's going to get real interesting 00:26:38.100 --> 00:26:42.142 is government regulation and government oversight around open source to control, 00:26:42.774 --> 00:26:51.450 whatever concept of truth is left and

because when so much can be manipulated then how, 00:26:52.155 --> 00:26:58.075 is is there an element of control that they're going to try to manage you know or inject, 00:26:59.097 --> 00:27:01.810 i think that's true i think part of that though is also like the competition 00:27:01.810 --> 00:27:04.750 aspect that steven mentioned right like i think the fact that there's going 00:27:04.750 --> 00:27:08.990 to be multiple parties and across multiple countries trying to go for this thing 00:27:09.370 --> 00:27:11.137 should keep people a little bit more honest. 00:27:12.050 --> 00:27:13.430 We'll see if that plays out. I agree with you. 00:27:14.370 --> 00:27:16.749 The other thing I wanted to mention was. 00:27:18.587 --> 00:27:23.117 On the open source side, I think that, Stephen, you mentioned KimiKate 3. 00:27:23.518 --> 00:27:26.506 The cost of, like, frontier intelligence is increasing. You started the pod 00:27:26.506 --> 00:27:29.793 saying, like, it's getting more expensive if you use Fable for certain tasks versus others. 00:27:30.071 --> 00:27:32.496 But, like, the general intelligence of, like, using Sonnet or something that 00:27:32.496 --> 00:27:36.496 was, like, GPT-4 or 5 is way cheaper, way cheaper today than it was, 00:27:36.496 --> 00:27:40.076 you know, a year ago. So the cost of, like, baseline intelligence is getting 00:27:40.076 --> 00:27:42.853 cheaper. The cost of frontier intelligence is getting more expensive. 00:27:43.224 --> 00:27:45.476 And so to the open source question, it's like, well, if you just have to wait 00:27:45.476 --> 00:27:48.706 six months to get what was Frontier today, like maybe that's worth it for your 00:27:48.706 --> 00:27:51.386 use case. I don't know. It depends on what kind of application you have underneath it. 00:27:52.390 --> 00:27:56.496 Yeah. Yeah. I mean, I think what it is doing, to your point, is, 00:27:57.504 --> 00:28:02.986 it's making it a bit more complicated in terms of thinking through the architecture 00:28:02.986 --> 00:28:08.662 of what you're building and aligning kind of the model, the default model to the use case. Yeah. 00:28:10.018 --> 00:28:15.745 But let's move on to this next story. So I think the details of this are maybe 00:28:15.745 --> 00:28:16.715 a little less important. 00:28:16.715 --> 00:28:21.175 We've got Francisco Partners kind of buying a majority stake in Command Alcon. 00:28:21.175 --> 00:28:25.445 Command Alcon is essentially the technology used for dispatch, 00:28:25.445 --> 00:28:30.638 ticketing, logistics, and QC for concrete asphalts and aggregates. 00:28:32.000 --> 00:28:38.695 I think the question to the group is, is this just like an allocation kind of play, 00:28:39.575 --> 00:28:46.505 for a PE firm, or is this hinting at a trend which increasingly we're seeing, 00:28:46.505 --> 00:28:51.899 which is other sectors attempting to get upside exposure to construction? 00:28:52.520 --> 00:28:56.585 You know, we talk a lot about M&A activities in our industry, 00:28:56.585 --> 00:29:02.587 mainly in the tech category, but now we're seeing it at kind of all layers of our industry. 00:29:02.889 --> 00:29:09.803 And so, I don't know, Victor, do you see this as a larger play and an indicator of more to come? 00:29:10.751 --> 00:29:14.765 Yes but one of the things about command.con for for those in those types of 00:29:14.765 --> 00:29:17.265 markets at least from what i've heard is like it's not a great software it's 00:29:17.265 --> 00:29:20.245 one of those that you know the customers are their hostages kind of deal and 00:29:20.245 --> 00:29:24.268 if you look at that it's easier for a pe firm to come in and buy that kind of company out 00:29:24.686 --> 00:29:27.915 you can cut a bunch of people out of the mix you know streamline the business 00:29:27.915 --> 00:29:30.895 increase price and now for a pe firm you could flip that around for three to 00:29:30.895 --> 00:29:33.780 five x so i think that playbook is relatively straightforward for them 00:29:34.199 --> 00:29:37.535 so they won't lose so they won't churn a bunch because if you're if you're a 00:29:37.535 --> 00:29:40.415 dispatcher for concrete there's not that many options for ready mix company 00:29:40.415 --> 00:29:43.565 to go buy software that does what command alcon does same for aggregate same 00:29:43.565 --> 00:29:44.592 for any of these companies so 00:29:44.883 --> 00:29:48.735 i think yes but but only within the confines of what specifically command outcome 00:29:48.735 --> 00:29:50.201 provides to those types of companies. 00:29:51.874 --> 00:29:57.250 I mean, what would it take to build? I mean, I guess the question I have, too, is also, 00:29:57.994 --> 00:30:01.604 as we just stopped talking about all the different capabilities of these new 00:30:01.604 --> 00:30:10.194 models, is it going to take more time for a bad incumbent to just be outright replaced? 00:30:10.194 --> 00:30:14.481 Like, why do these things need to change hands? Why doesn't someone in that 00:30:14.603 --> 00:30:17.974 kind of subsegment of the industry just build net new? Obviously, 00:30:17.974 --> 00:30:19.374 there's customer relationships. 00:30:20.031 --> 00:30:23.420 There's the network effect. There's all those factors, but it does feel like, 00:30:24.017 --> 00:30:27.344 if it's going to happen, Tyler, isn't it like, can't someone just go build it? 00:30:28.502 --> 00:30:32.333 It's really interesting because I won't say who, but I mean, 00:30:32.333 --> 00:30:35.223 I was talking to some people the other day and there's some acquisitions that 00:30:35.223 --> 00:30:40.039 just happened and we're talking about, you know, is the company that got acquired, 00:30:40.289 --> 00:30:41.473 are they even profitable? 00:30:41.853 --> 00:30:46.878 You know, and they're like, oh, the company wasn't profitable, 00:30:47.011 --> 00:30:50.370 but it's sold for a lot of flipping money. 00:30:51.459 --> 00:30:54.403 And I'm looking at how much, you

know, if it was a hundred million dollars, 00:30:54.403 --> 00:30:59.353 let's say that this company was acquired for the company that bought them, 00:30:59.353 --> 00:31:01.692 why couldn't they've just gone out and built the dang thing, 00:31:02.151 --> 00:31:05.646 right? For a hundred million dollars, you can get a lot done for that. I'm sorry. 00:31:06.139 --> 00:31:09.645 And especially considering what this specific thing did. 00:31:10.208 --> 00:31:16.393 And so, yeah, there's stuff in the tea that's getting swirled around right now 00:31:16.393 --> 00:31:17.923 that I'm just not sure about. 00:31:17.923 --> 00:31:20.070 And this is something I was talking to my dad about. I'm like, 00:31:20.453 --> 00:31:24.523 maybe this is the reason why I'm just not sitting in the CEO seat at one of 00:31:24.523 --> 00:31:28.253 these large companies because I can't see around the corners like this and make 00:31:28.253 --> 00:31:31.257 sense of buying a company that, 00:31:31.915 --> 00:31:33.581 isn't profitable. 00:31:33.709 --> 00:31:37.854 I mean, it doesn't have, and, and, oh my gosh, yeah, sorry. 00:31:38.388 --> 00:31:41.441 The multiple that it was bought for was just absolutely insane. 00:31:42.104 --> 00:31:48.096 So I just, as a business owner, I'm like, I cannot wrap my head around spending, 00:31:48.329 --> 00:31:52.154 you know, a hundred million dollars on something that's doing 4 million in revenue. 00:31:52.613 --> 00:31:56.757 Well, I'm going to do a perplexity search on who got sold for a hundred million dollars. 00:32:00.653 --> 00:32:04.140 I mean, I think that also plays into some of these fundraising that we've seen. 00:32:07.691 --> 00:32:10.235 I was going to say, I think that also plays into some of these fundraisers that 00:32:10.235 --> 00:32:15.105 we've seen of like people raising astronomical rounds and like insane valuations. 00:32:15.105 --> 00:32:19.755 Like, well, what if you raise at that number, you're severely limiting your 00:32:19.755 --> 00:32:23.315 exit opportunity. Like you almost had to IPO or exit to mega PE firm because 00:32:23.315 --> 00:32:26.555 like a hundred million dollars, while a lot of money to a lot of large companies, 00:32:26.555 --> 00:32:29.627 that's, that's doable. You could pay that off because you could chart the path forward. 00:32:30.206 --> 00:32:33.275 Billions of dollars. It's a totally different conversation. You know what I 00:32:33.275 --> 00:32:35.815 mean? And So it's, it's, it's changing the game on the field, 00:32:35.815 --> 00:32:38.865 changing the game in the field massively. Yeah. Yeah. 00:32:39.625 --> 00:32:44.645 I think it's a good segue to talk about the Gravis Robotics seed from SoftBank. 00:32:45.025 --> 00:32:49.425 I think, you know, I almost had Ryan on the podcast last year and we're going 00:32:49.425 --> 00:32:52.165 to do kind of a joint episode with something from Granite. 00:32:52.165 --> 00:32:58.955 But I mean, obviously it seems like the leaps in frontier intelligence is informing 00:32:58.955 --> 00:33:02.778 a leap in what's capable in terms of the hardware side of the industry. Yeah. 00:33:04.145 --> 00:33:08.465 And so I'm hearing about all of these incredible things that robots are going 00:33:08.465 --> 00:33:13.862 to be able to do, even at the site level, and whether that's field AI, 00:33:14.059 --> 00:33:15.835 I just, I haven't seen any of it yet. 00:33:15.835 --> 00:33:20.950 And obviously, I'm not on the day-to-day of a job site, but is it happening, Tyler? 00:33:22.045 --> 00:33:26.745 Same thing it's the same thing we just talked about 200 million dollars I'm 00:33:26.745 --> 00:33:31.735 not saying it's them it's the same thing 200 million dollars for a company that 00:33:31.735 --> 00:33:33.438 hadn't even figured out how to make money, 00:33:34.341 --> 00:33:37.580 Well, I mean, I think that's like. Lars, please help me. 00:33:37.871 --> 00:33:42.898 Help me. Lars, are you getting your hair done? What's going on? Where's your camera? 00:33:43.637 --> 00:33:47.927 Is my camera off? Yeah, dude. I can see. I can see him. Oh, really? 00:33:48.205 --> 00:33:49.895 Okay. As long as you guys can see him. 00:33:52.173 --> 00:33:57.613 I'm getting my hair done. It's speculation, guys. It's just speculation. 00:33:57.613 --> 00:34:00.063 So there's a couple things, Tyler, that you said earlier. Like. 00:34:02.364 --> 00:34:05.233 Why would I just go buy it versus, why would I buy versus build, 00:34:05.233 --> 00:34:06.828 right? Build versus buy, buy versus build. 00:34:07.605 --> 00:34:10.253 You know, it's interesting how reluctant 00:34:10.253 --> 00:34:16.645 companies are to change something internally if they're a consumer. 00:34:17.015 --> 00:34:21.503 I'm not going to change the platform I'm on because it took 10 years for us 00:34:21.503 --> 00:34:22.683 to finally figure it out. 00:34:22.683 --> 00:34:28.363 And it's just, I don't want to leave it, right? How many companies aren't satisfied 00:34:28.363 --> 00:34:32.745 with certain, call it, products they use, but they make them work? 00:34:33.168 --> 00:34:37.193 And it's easier to make them work than to ask them to deal with more change 00:34:37.193 --> 00:34:42.083 in the organization. So that is something that, you know, from a human capital 00:34:42.083 --> 00:34:47.076 standpoint, internal, we'll call it like a thousand foot below inside of a company. 00:34:48.043 --> 00:34:50.923 That's some of the sentiment that's going on, right? It takes a very visionary 00:34:50.923 --> 00:34:58.993 leader to create a very elastic organization where the people are innovating at all levels, right? 00:34:58.993 --> 00:35:04.213 At all levels, rather than just doing their sort of methodical plowing of the 00:35:04.213 --> 00:35:06.359 field and trench digging every day, right? 00:35:06.962 --> 00:35:10.132 But relating to Gravis Robotics and stuff like that, I mean... 00:35:11.144 --> 00:35:20.344 Guys, we have self-driving

cars. It's here, right? Like, it's happening. And with what will be... 00:35:21.064 --> 00:35:25.164 So, I actually think it'll be... Like, building a highway, like, 00:35:25.164 --> 00:35:27.444 that'll just be robots out there. 00:35:27.444 --> 00:35:30.814 Like, that can be more easier. Where I think it's going to be more challenging 00:35:30.814 --> 00:35:32.974 is in an existing environment, right? 00:35:32.974 --> 00:35:40.624 Not new, existing, unpredictable, where, you know, these environmental stimuli 00:35:40.624 --> 00:35:44.274 are going to somehow have to be processed on the fly rather than here's the 00:35:44.274 --> 00:35:47.273 model, you know exactly where you are in this model. 00:35:47.860 --> 00:35:52.304 And, you know, I'm going to tell you on a personal side, I'm having my investment 00:35:52.304 --> 00:35:55.145 team, I'm like robotics. 00:35:56.738 --> 00:36:01.314 AI is one thing, robots. And Elon's talking about it. A lot of these other guys 00:36:01.314 --> 00:36:04.147 are talking about it. You can see it in the defense world, Andrel. 00:36:05.084 --> 00:36:09.884 Like, it's all robots. And you layer the AI on top of the robots. 00:36:09.884 --> 00:36:15.534 Now we have it because before we didn't really have that sort of like wait how's 00:36:15.534 --> 00:36:16.744 a robot going to actually do it, 00:36:18.264 --> 00:36:23.824 now it's doing it you can ship a humanoid to yourself like right now so here's 00:36:23.824 --> 00:36:26.964 a question actually it doesn't get shipped to you it. 00:36:28.784 --> 00:36:31.664 Takes a train ride it's like I'm going to go see Tyler. 00:36:36.764 --> 00:36:40.995 Where's my gun so the question is is would you guys feel comfortable, 00:36:41.993 --> 00:36:46.093 with a robot in your house. No. I mean, we already do. 00:36:46.513 --> 00:36:48.763 Let's be honest. I think a humanoid, no, I think that's creepy. 00:36:48.763 --> 00:36:50.793 But, like, I have a Roomba. Like, that's not that crazy. 00:36:51.213 --> 00:36:54.463 I think other people have, like, different types of... They're not really robots 00:36:54.463 --> 00:36:57.213 in the sense that they're doing things for you, but your refrigerator, 00:36:57.213 --> 00:36:58.853 your microwave, your TV... 00:36:59.713 --> 00:37:04.203 Would you feel better about it if it didn't look like a human? 00:37:04.203 --> 00:37:05.243 Like, if it looks like... 00:37:05.313 --> 00:37:11.523 100%. What if I connected your fridge with arms and legs, connected to the internet 00:37:11.523 --> 00:37:15.443 and gave it like Kimmy K3. Like, I think that changes the game. 00:37:15.443 --> 00:37:16.613 Yeah, of course, of course. 00:37:17.733 --> 00:37:23.313 Mobility plus intelligence. That's a different thing than my oven. 00:37:24.089 --> 00:37:29.793 Yeah, I think humanoids are in the home a bridge too far, quite frankly. I would never buy one. 00:37:30.093 --> 00:37:34.443 But I think like general purpose tools are interesting. I think within the home, 00:37:34.443 --> 00:37:38.741 I would only specific task things that I can't. 00:37:38.875 --> 00:37:42.863 Again like a Roomba is not that a lot of people have them that's not that crazy 00:37:42.863 --> 00:37:46.163 people have like lawnmower robots like that's not that crazy but when you start 00:37:46.163 --> 00:37:49.253 thinking about like how fast can they go how tall are they how big are they 00:37:49.253 --> 00:37:52.573 how heavy are they like that's when it starts to get a little scary I'm, 00:37:54.073 --> 00:37:59.139 like if I get time back if I get time back I'm interested I'm interested. 00:38:00.922 --> 00:38:03.546 What about like folding clothes or doing dishes or any of these? 00:38:03.546 --> 00:38:05.635 Do you like folding clothes or doing dishes? 00:38:06.187 --> 00:38:09.356 I don't mind them. As far as like chores go, I don't mind them. 00:38:09.356 --> 00:38:11.096 I mean, would you rather be doing something? 00:38:11.536 --> 00:38:16.986 Of course. Of course I would. So, so this is where like, I want a robot to do 00:38:16.986 --> 00:38:19.476 those things that I don't like to do 100%. 00:38:19.636 --> 00:38:24.687 Well, of course that's the selling point, right? Yeah. But I'm uncomfortable. 00:38:26.882 --> 00:38:34.266 With like being under surveillance by whatever or whatever subscription I'm 00:38:34.266 --> 00:38:36.599 paying to someone to fold my laundry. 00:38:37.376 --> 00:38:39.726 The second you're late for your payment and that thing goes haywire in your 00:38:39.726 --> 00:38:43.816 house like you're cooked. You know what I mean? Like you can't have that situation 00:38:43.816 --> 00:38:45.676 happen. I don't know what to do now. 00:38:48.296 --> 00:38:53.194 I mean I will say It then goes sabotages the laundry it's already folded. 00:38:53.374 --> 00:38:55.986 Exactly. You haven't paid me. Third back in the washing machine. 00:38:55.986 --> 00:38:58.210 Yeah, it's like, where's my clothes, bro? 00:38:58.877 --> 00:39:03.347 Burns all of your clothes because you didn't upgrade it to the newest version. 00:39:04.356 --> 00:39:09.218 So I think there are kind of two different lanes if we get back to Gravis specifically. 00:39:09.834 --> 00:39:14.867 Like, I mean, that's civil. So I think that stuff makes pretty good sense because, 00:39:15.454 --> 00:39:19.626 you just have to be spatially aware of what's going on and coding what needs 00:39:19.626 --> 00:39:21.206 to happen to the earthwork. 00:39:21.607 --> 00:39:25.846 Not that difficult. Now, if we start moving through the actual construction 00:39:25.846 --> 00:39:29.014 process and we start looking at robotics as it moves through, 00:39:29.560 --> 00:39:32.766 humanoids start to get really interesting, yeah, but also, like, 00:39:32.766 --> 00:39:35.620 we've seen people try this stuff with, like, 00:39:36.172 --> 00:39:38.627 I'm gonna call Hilti out just for fun, J-Bot. 00:39:38.854 --> 00:39:41.290 If we look at J-Bot, or we look at

even Dusty. 00:39:42.444 --> 00:39:46.726 There's setup that has to occur with each one of those platforms, right? 00:39:47.039 --> 00:39:52.274 And sometimes the setup doesn't merit deploying it out on site. 00:39:52.274 --> 00:39:57.764 So when we went out to race JBOD back in the day, this was, I mean, 00:39:58.330 --> 00:40:01.844 2019 when Eddie and I did this, maybe a little late, maybe it was 2020. 00:40:01.844 --> 00:40:02.586 I don't know. It doesn't matter. 00:40:03.317 --> 00:40:06.277 We went out and we raced it to see who could win. 00:40:06.853 --> 00:40:12.871 But the thing that is not shown on camera is the hour and a half that it took 00:40:13.134 --> 00:40:18.045 them on the phone to get the freaking thing oriented with the total station, 00:40:18.867 --> 00:40:23.164 and so oh this isn't aligned we got to get this file right like back and forth 00:40:23.164 --> 00:40:27.199 back and forth back and forth meanwhile eddie can just go zip zip zip zip zip 00:40:27.389 --> 00:40:30.547 move now if it's 40 000 holes, 00:40:31.277 --> 00:40:34.487 starts to make a little bit more sense i'll give you that, 00:40:35.178 --> 00:40:39.284 but then there's the there's the modeling aspect of it the model fidelity right 00:40:39.284 --> 00:40:40.914 now the 3d models that are being 00:40:40.914 --> 00:40:46.647 produced do not have builder grade accuracy in them in a lot of cases, 00:40:47.464 --> 00:40:51.544 because the tooling there doesn't allow us to get that specific because revit's 00:40:51.544 --> 00:40:59.844 a pig and it sucks yeah that's the headline you do it yeah i do think i mean it's. 00:41:01.135 --> 00:41:06.874 The money from i mean softbank SoftBank, I mean, infamously throws money around a little bit, right? 00:41:06.874 --> 00:41:10.484 And, you know, they're not a sponsor, so we can say all kinds of things about 00:41:10.484 --> 00:41:14.844 SoftBank. But like, they're notorious for all... I'm sorry. 00:41:16.784 --> 00:41:20.204 For all the people on the sponsor list that are waiting, we'll get to you. 00:41:20.204 --> 00:41:21.954 Don't worry. The list is long. Forgive us. 00:41:22.184 --> 00:41:26.128 SoftBank, if you're listening, we will take your money and we will stop saying these things. 00:41:27.104 --> 00:41:32.754 But, like, I mean, you know, they've made some investments that we'll say were 00:41:32.754 --> 00:41:36.726 pretty ambitious. But I do think, I mean, Gravis has a real product. 00:41:37.684 --> 00:41:42.734 They're obviously banking on continued kind of increase in frontier intelligence, 00:41:42.734 --> 00:41:45.372 which is happening. So I think it's interesting. 00:41:45.964 --> 00:41:47.549 I've got a little bit of a different take on this. 00:41:48.764 --> 00:41:52.514 I think $200 million is actually not that much money in the grand scheme. 00:41:52.514 --> 00:41:56.669 It's, again, large ticket number. But for what we're talking about in robotics, it's not that crazy. 00:41:57.197 --> 00:42:00.444 I do think the valuation of it is pretty expensive though. So there's a little 00:42:00.444 --> 00:42:01.214 bit of a disconnect there. 00:42:01.214 --> 00:42:05.284 So I think when you think about construction robotics, particularly on the civil 00:42:05.284 --> 00:42:06.594 side, you have two flavors. 00:42:06.594 --> 00:42:10.534 You have, could you buy a fully autonomous unit from a manufacturer or from 00:42:10.534 --> 00:42:15.284 a provider? Or can you buy hardware and software combined to retrofit your existing 00:42:15.284 --> 00:42:17.344 fleet? Those are really the two options you have. 00:42:17.762 --> 00:42:20.354 And so you've seen a bifurcation of the kinds of companies out there, 00:42:20.354 --> 00:42:23.794 right? Ready Robotics, Race Robotics, Build Robotics, Bedrock, 00:42:23.794 --> 00:42:24.849 like you go down the list. 00:42:25.453 --> 00:42:27.544 And so for a company like Kiewit that already has, you know, 00:42:27.544 --> 00:42:30.554 three, four or five billion dollars worth of equipment, it makes much more sense 00:42:30.554 --> 00:42:34.394 to go and buy retrofit equipment to do that to your existing fleet than it does 00:42:34.394 --> 00:42:38.298 to go spend billions of dollars to go retrofit and rebuy everything you have. 00:42:38.844 --> 00:42:41.404 And so that in that world, then, yeah, Gravis makes more sense. 00:42:41.404 --> 00:42:44.574 So like the valuation is expensive, but spending two hundred million dollars 00:42:44.574 --> 00:42:48.184 to go build enough of these units to then go ship to all the civil contractors across the country. 00:42:48.631 --> 00:42:51.214 That's not that crazy to think about. I agree with you, Tyler, 00:42:51.214 --> 00:42:54.994 like net new buildings on Greenfield sites. it's pretty easy. 00:42:54.994 --> 00:42:59.364 But when you get into existing utilities and strikes, that is really freaking hard. 00:42:59.924 --> 00:43:03.399 That is unsolved. Rod Radar, God bless them, it's such a hard problem to solve. 00:43:03.649 --> 00:43:05.054 You can't really get there yet. 00:43:05.437 --> 00:43:08.014 But I think if you're doing data centers, if you're doing energy, 00:43:08.014 --> 00:43:10.654 a lot of these are greenfield, they're out in the middle of nowhere type projects. 00:43:10.654 --> 00:43:12.229 And so it's much easier to see that happen. 00:43:12.565 --> 00:43:16.269 And then again, from the capital expenditure side, from a GC or civil contractor, 00:43:16.553 --> 00:43:20.924 it's much easier to make a decision to spend tens of thousands of dollars to 00:43:20.924 --> 00:43:24.285 buy one of these units than it would be to go buy a new Komatsu piece of equipment. 00:43:25.329 --> 00:43:29.529 Yeah, I think that's fair. I think though, with any kind of piece of technology, 00:43:29.529 --> 00:43:35.499 like the concept is, is a component of the valuation, but so is traction and 00:43:35.499 --> 00:43:37.692 like real world kind of implementation.

00:43:38.122 --> 00:43:42.849 And I think that's where there's like a devoid of, of info on, 00:43:42.849 --> 00:43:45.709 on how, how this is actually being done. That's why they're expensive. 00:43:45.709 --> 00:43:47.129 It's unclear. It's unclear. 00:43:47.569 --> 00:43:49.929 I don't think like a lot of these companies have their revenues that they, 00:43:50.529 --> 00:43:52.750 they're aspirational, but they're not there yet. 00:43:53.348 --> 00:43:57.959 Yeah no i mean like lars said i mean it's speculative it's it's speculative 00:43:57.959 --> 00:44:02.769 investing at this point it's r&d it's r&d gravis makes sense like if i'm an 00:44:02.769 --> 00:44:04.407 investor i'm looking at gravis i'm going yeah okay, 00:44:05.069 --> 00:44:08.899 i can see how that would track but some of these other robotics platforms that 00:44:08.899 --> 00:44:11.805 we see coming out in the construction space i just don't. 00:44:12.489 --> 00:44:15.239 They haven't found their product market fit yet they haven't figured out how 00:44:15.239 --> 00:44:19.787 to integrate themselves into the job sites so well so that they become indispensable, 00:44:20.310 --> 00:44:25.129 i mean i shoot like working in steel you look at robotics that are actually 00:44:25.129 --> 00:44:30.729 in the steel fabrication shops i've had to model steel to run through and fix 00:44:30.729 --> 00:44:34.236 the nc files because the robot couldn't read it right, 00:44:34.772 --> 00:44:37.442 and so i've had to drive robotics through 3d i'm, 00:44:38.056 --> 00:44:43.279 like if we can't even get freaking steel right good luck out in a job site that 00:44:43.279 --> 00:44:48.239 is very unpredictable with crap all over the floor and people that don't want 00:44:48.239 --> 00:44:52.610 the robot out there because they're scared of it taking their jobs so it's a harder, 00:44:53.174 --> 00:44:56.779 you know it's a harder task to get this implemented in the field than i think 00:44:56.779 --> 00:45:00.982 we give credit so it's yeah something like gravis i can see it. 00:45:01.780 --> 00:45:06.479 As we go down more into the actual like bricks and sticks of things that's where 00:45:06.479 --> 00:45:09.483 it starts to get really complex on the robotics end. 00:45:10.405 --> 00:45:14.545 Well, from hardware to software, let's talk about Procore's kind of AI strategy. 00:45:14.545 --> 00:45:19.335 So I think in July, we talked about the kind of digital coworker and some of 00:45:19.335 --> 00:45:21.005 the agents that they were starting to roll out. 00:45:21.385 --> 00:45:25.683 They've put together kind of a more kind of defined package. 00:45:26.083 --> 00:45:30.625 And let me look at that. It says there's one kind of tier, which is started 00:45:30.625 --> 00:45:33.655 with flat rate, six months, capped at three projects, five agents. 00:45:33.655 --> 00:45:38.750 And then there's like pro and enterprise, 12-month terms, credit usage-based pricing. 00:45:39.430 --> 00:45:44.075 So i guess how should how should the industry kind of view this obviously everyone's 00:45:44.075 --> 00:45:48.305 kind of skeptical of of procor's pricing methodology historically and so, 00:45:48.785 --> 00:45:53.005 as they roll out a big component of their ai strategy they're leaving out the 00:45:53.005 --> 00:45:58.345 price of how they're going to charge people's use of these agents that are getting 00:45:58.345 --> 00:46:01.339 integrated into every you know level of their platform um. 00:46:03.104 --> 00:46:09.064 How on top on top of a percentage model that was already that was already sort 00:46:09.064 --> 00:46:14.584 of like wait is is there is the real value here if if we're just talking rfis 00:46:14.584 --> 00:46:17.871 and submittals guys is it real value right, 00:46:18.486 --> 00:46:23.914 dude seriously so lars we've talked about you know this idea of like what does 00:46:23.914 --> 00:46:25.314 owning the data look like, 00:46:25.904 --> 00:46:32.724 as as an owner on the construction side and development side would this be prohibited 00:46:32.724 --> 00:46:38.144 for you like say you were kind of neutral i went i know you're not but if you were would you, 00:46:38.724 --> 00:46:41.784 would you be like okay i'm just gonna try this for a few months and go through 00:46:41.784 --> 00:46:45.154 the pains of change management only to find out that it's way more expensive 00:46:45.154 --> 00:46:48.540 than you want it to be or is this prohibitive i think, 00:46:49.213 --> 00:46:56.080 i think having having fallen many times and stubbed toes and knees and lost some teeth, 00:46:57.144 --> 00:47:01.764 you you're like yeah no this is this is just an overreach this is this this 00:47:01.764 --> 00:47:07.864 is another I think, unreasonable grasp at something that I'm just, I don't see the value. 00:47:07.864 --> 00:47:11.394 The value is existing over here. This is a bolt on, you know, 00:47:11.394 --> 00:47:14.391 for some other stuff. Yeah, I guess it's convenient that maybe some of this, 00:47:14.646 --> 00:47:17.839 we'll call it data, has already been uploaded to the platform. 00:47:18.001 --> 00:47:21.693 But then it's like, wait a minute, why am I even giving them that stuff? 00:47:23.024 --> 00:47:26.161 Maybe I should take a different approach, which is how do I not give them this? 00:47:26.649 --> 00:47:32.036 And again, what value are they bringing to my organization at what cost, 00:47:32.622 --> 00:47:35.654 right so i don't know if that i don't know if that answered your question but, 00:47:36.153 --> 00:47:41.749 you know i'm immediately very speculative like like like or concerned about like, 00:47:42.473 --> 00:47:46.654 why aren't you our customers just not being loud enough about around your the 00:47:46.654 --> 00:47:50.374 pricing pain or is there just that much am i the only one that just, 00:47:50.994 --> 00:47:55.494 maybe he's like am i getting the value for which i'm paying for a year am i 00:47:55.494 -->

00:48:00.324 the only one feeling that and then wait you're now going to charge more on top Yep. 00:48:01.518 --> 00:48:05.628 I thought Sass was dead, right? I thought Sass was dying. And now they're like 00:48:05.628 --> 00:48:07.713 doubling down and charging more. 00:48:09.258 --> 00:48:12.897 Come on, guys. But don't worry. We're going to send you all the goods in the mail. 00:48:13.778 --> 00:48:18.378 And we're going to have a racing team. And we're going to spend all the money, 00:48:18.718 --> 00:48:22.262 right? We're going to spend all your money on us. 00:48:23.998 --> 00:48:27.768 I strongly, strongly, strongly disagree. Wait, Victor, before you provide your 00:48:27.768 --> 00:48:33.230 perspective, let's try and construct like what would Chris Lengiza say? 00:48:33.415 --> 00:48:36.978 What's like Procore's perspective on this and then we can all provide like then 00:48:36.978 --> 00:48:38.919 I want to hear your actual take. 00:48:39.064 --> 00:48:43.598 I have a bow tie do I have a bow tie somewhere? Yes so Tyler you're going to 00:48:43.598 --> 00:48:49.558 be Chris for the next minute and a half give me your explanation don't do something I can't actually. 00:48:52.958 --> 00:49:01.058 So go Victor Well, yeah, Victor, do you want to try and craft Procore perspective? 00:49:01.058 --> 00:49:04.038 I'll try to steal a minute from Procore, and then I'll give my own personal. 00:49:04.038 --> 00:49:06.938 I think this is actually one of the best things Procore, and smartest things 00:49:06.938 --> 00:49:09.128 Procore's done in a bit. I think this is sneaky sharp. 00:49:09.696 --> 00:49:12.541 I agree. Are you acting as Procore? Is this your take? 00:49:13.158 --> 00:49:16.368 I'll let you decide by the end of this. I think, Lars, to what you're saying, 00:49:16.368 --> 00:49:19.568 a lot of the industry definitely has been upset at Procore, their pricing model. 00:49:19.568 --> 00:49:21.633 I think it made sense for Procore as they were growing. 00:49:22.103 --> 00:49:26.683 But I think there's a lot of heartburn right now. I think that's totally... felt and heard. 00:49:27.672 --> 00:49:32.292 But if I was at Procore, let me paint the picture. I think you know the base 00:49:32.292 --> 00:49:35.292 floor and you know the ceiling roughly of what people are willing to spend for 00:49:35.292 --> 00:49:37.292 your product. And it's a percentage that you've already been charging. 00:49:37.292 --> 00:49:40.282 You can slightly increase it every year, but there's no meaningful increase 00:49:40.282 --> 00:49:41.972 to that. People are going to churn. They're going to go to Autodesk. 00:49:42.012 --> 00:49:44.332 They're going to go to these other companies. So you know roughly what the max 00:49:44.332 --> 00:49:45.272 is that someone's going to pay. 00:49:45.921 --> 00:49:49.522 With the data grade acquisition, you're starting to try and get to this world 00:49:49.522 --> 00:49:53.671 where you're doing less reliant. You're less reliant on what a core, 00:49:54.135 --> 00:49:56.362 what a contractor is actually doing in the top line revenue. 00:49:56.362 --> 00:50:00.393 You're trying to get closer and closer to verified actions of what your software could do. 00:50:00.811 --> 00:50:05.722 And the idea being, if you could use an agent to go complete an RFI for you, 00:50:05.722 --> 00:50:07.201 I could charge for that. I could charge. 00:50:07.839 --> 00:50:11.572 Continue down the path. They bought DroneDeploy. And the idea behind the DroneDeploy 00:50:11.572 --> 00:50:13.462 thing is less about the data that they already have. 00:50:13.462 --> 00:50:16.302 But if you're still going with like, okay, if I need an agent to go do something 00:50:16.302 --> 00:50:20.312 on my behalf, and now I have the data and the insights and the visual confirmation 00:50:20.312 --> 00:50:24.482 from a project site, it is much easier for me to verify what happened in the RFI. 00:50:24.482 --> 00:50:27.372 It's much easier to generate the submittal and do all this other stuff the Pay 00:50:27.372 --> 00:50:29.127 application within the Procore application. 00:50:29.824 --> 00:50:33.734 All the way from start to finish. Like if I'm Procore, it's starting to make 00:50:33.734 --> 00:50:36.884 more sense why I would buy those companies together in a world where I know 00:50:36.884 --> 00:50:39.644 the max of what someone's willing to pay. Here are the components of why they 00:50:39.644 --> 00:50:41.608 use the software and now I can use agents to do it. 00:50:42.242 --> 00:50:44.727 That's the, from the product side, that's how I would think about it. 00:50:45.348 --> 00:50:48.512 From the pricing side, this is really, really complicated. 00:50:48.842 --> 00:50:52.664 This is like, if you're Procore, this is almost a burn the boats moment, 00:50:52.664 --> 00:50:56.004 quite frankly, because if you've been entirely relying on your contractors to 00:50:56.004 --> 00:50:59.114 grow their business in order for you to grow your business, you're in a sticky 00:50:59.114 --> 00:51:03.063 position, especially if revenues go down and you have some softening in other markets, 00:51:03.563 --> 00:51:05.824 against a time where people are getting pretty frustrated with how much you're 00:51:05.824 --> 00:51:08.874 charging so you have to figure out a new way to create value and you have to 00:51:08.874 --> 00:51:11.713 create a new way to actually capture that value and they're trying to do that here, 00:51:12.160 --> 00:51:15.734 so against the backdrop of the products of how they're doing using agents and 00:51:15.734 --> 00:51:18.414 the verification of the field now they're trying to create a way that they could 00:51:18.414 --> 00:51:20.484 charge you for those actions it actually allows you 00:51:20.924 --> 00:51:24.524 in theory as a contractor to eventually stop buying the full modules the way 00:51:24.524 --> 00:51:27.614 that they break up all their products suites into modules you don't in

In the 00:51:27.614 --> 00:51:28.834 future, you won't have to buy all of them. 00:51:28.834 --> 00:51:31.165 You just need to buy the ones, the actions that you need. 00:51:31.485 --> 00:51:34.039 You could buy submittals. You could buy RFIs. You don't need to buy finance. 00:51:34.393 --> 00:51:36.364 Like the fact that if you buy finance from Procore- Hold on, has that come out yet? 00:51:36.904 --> 00:51:40.194 Has that come out? They don't. I think that's, again, if I'm at Procore, 00:51:40.194 --> 00:51:43.054 this is a future I'm trying to build. This is what I'm trying to get to so that 00:51:43.054 --> 00:51:45.110 I could charge just for what you actually need. 00:51:45.510 --> 00:51:48.558 Because in that world, it is much cheaper for me. It is a higher margin product 00:51:48.976 --> 00:51:50.711 and it's easier to plan for. 00:51:51.048 --> 00:51:54.474 Now, at the contractor level, it's totally different because as a contractor, 00:51:54.474 --> 00:51:56.864 you have no way to actually forecast that. You have no idea how much you're 00:51:56.864 --> 00:51:58.894 actually going to spend. So I disagree with that. 00:51:59.224 --> 00:52:06.221 Oh, why? Because I believe in a consumption-based model, I can predict exactly... 00:52:06.913 --> 00:52:09.761 What my actions are at bid time. 00:52:10.973 --> 00:52:16.334 I can assume based on historicals with this customer and designer, X number of RFIs. 00:52:16.712 --> 00:52:21.263 I can predict how many subcontractors, how long the project's going to go on 00:52:21.263 --> 00:52:23.533 for, how many days on site people will be there. 00:52:24.493 --> 00:52:28.543 100%, that's predictable to me. And I'm a fan of consumption-based models. 00:52:28.543 --> 00:52:33.389 That's what my platform's based off of because I can predict at time of bid 00:52:33.604 --> 00:52:36.773 how much my software will cost on this project. 00:52:37.553 --> 00:52:41.313 But I don't even think we're arguing about the price methodology, 00:52:41.313 --> 00:52:45.993 because I can understand token variability is a huge variable in this kind of 00:52:45.993 --> 00:52:52.173 product. And so I can respect the fact that there's a need to do some metering. 00:52:53.313 --> 00:52:56.153 Steven, I think you said it eloquently, or maybe it was Tyler. 00:52:56.153 --> 00:52:58.012 It certainly wasn't me. Probably me. 00:52:58.314 --> 00:53:04.561 But it was like this Procore was built on the back of contractors, 00:53:05.273 --> 00:53:10.093 bludgeoning them on like, no, this is what this means. No, this is what I need. 00:53:10.093 --> 00:53:12.367 No, this is how this works. 00:53:13.516 --> 00:53:21.183 And I just think, like, why are they constantly, well, we know what the max 00:53:21.183 --> 00:53:22.205 people are willing to pay. 00:53:22.846 --> 00:53:26.032 Where's the value prop in that? This is already a tight industry. 00:53:26.549 --> 00:53:30.733 And by the way, if I predict my revenue is $80 million and I give you a check 00:53:30.733 --> 00:53:33.973 for my $80 million because you've asked, right? And that's what my pricing is 00:53:33.973 --> 00:53:37.799 based off. But guess what? I only do $60 million. Victor, do they give me a credit back? 00:53:38.498 --> 00:53:41.627 No. Yeah. And that right there is wrong. 00:53:42.306 --> 00:53:47.780 And that is what I don't appreciate. Even the insurance industry gives me a credit back. 00:53:48.549 --> 00:53:49.855 They're worse than insurance. 00:53:51.039 --> 00:53:54.933 That's how you really feel, Lars. It's not me. give us your take and then we're 00:53:54.933 --> 00:53:57.795 going to move on to our last story before we go Tyler where's your go. 00:53:59.262 --> 00:54:04.340 All right, so I would say the biggest hiccup in all of this, 00:54:04.526 --> 00:54:08.541 I'll refer to this plan to charging per use. 00:54:08.981 --> 00:54:15.496 I'll refer to this plan as Operation Bendover Contractors who idea, 00:54:16.355 --> 00:54:21.202 that I can release an AI tool out to a thousand people, 00:54:21.864 --> 00:54:25.260 and they can just go prompt effing 00:54:25.591 --> 00:54:32.916 crazy on this thing and the CFO gets a bill for $150,000 at the end of the month, 00:54:33.224 --> 00:54:37.624 how long do you think that product is going to last in an organization 00:54:37.949 --> 00:54:43.492 if you cannot track a cleaner return on the investment? 00:54:43.492 --> 00:54:48.445 And that's the problem that I have with this, is the return on investment is 00:54:48.584 --> 00:54:51.246 not calculated clearly. 00:54:51.890 --> 00:54:57.666 So I'm fine if you tell me I put $150,000 in here, I get $15 million out. 00:54:58.142 --> 00:55:03.522 Woo, I'm great. But nobody has been able to do that eloquently. 00:55:03.522 --> 00:55:06.182 And that's where I was going to give you some props, Lars, for what you guys 00:55:06.182 --> 00:55:11.177 have built, is because within your own system, I can look... 00:55:12.576 --> 00:55:19.256 Email comes in, it does this, it sends this, boom, ROI, clear. 00:55:19.256 --> 00:55:24.819 But people sitting there and treating this thing like, where did I put blah, blah, blah? 00:55:25.976 --> 00:55:33.388 You can't predict that, Victor. $150,000 a free month and also beyond that, it's variable. 00:55:33.852 --> 00:55:38.712 The API costs are going to go, so as soon as I become reliant on this thing, 00:55:39.281 --> 00:55:46.652 and then open AI or anthropic, starts charging the actual cost of compute plus their problem, 00:55:47.320 --> 00:55:50.060 then what then it's going to be 500 grand a month, 00:55:50.960 --> 00:55:56.840 on what planet is that okay for any other software that's where it's gone kind of haywire for me is, 00:55:57.461 --> 00:56:01.936 ai as people are just turning their turning their brains off and saying like 00:56:01.936 --> 00:56:08.107 well it's ai it's amazing it's all the bull crap it comes back to return on investment every time. 00:56:08.571

--> 00:56:13.466 Don't get that mixed up. And when I ran a poll on LinkedIn, I was like, 00:56:13.466 --> 00:56:16.663 hey, you know, what's the biggest holdup for buying new technology? 00:56:17.081 --> 00:56:21.546 Everybody's answering return on investment. It's not clear. And especially for 00:56:21.546 --> 00:56:27.118 AI right now, it's still developing. I'm not saying that there isn't ROI to be hot. There is. 00:56:27.530 --> 00:56:31.977 But for stuff like this, that's bullcrap. It's nonsense. It's ridiculous. 00:56:33.776 --> 00:56:38.353 We will leave an opportunity for Procore to respond and see how they come back. 00:56:39.690 --> 00:56:47.016 Last story, 97% of drones on our job sites are from DJI. 00:56:47.442 --> 00:56:52.149 And then we've got those kind of doubling in cost because of some tariff announcements. 00:56:53.284 --> 00:57:00.226 How quickly do you think this translates into increased ProCore slash drone deploy bills monthly? 00:57:00.795 --> 00:57:02.584 Immediately? God. 00:57:03.884 --> 00:57:08.854 Yeah. I mean, I'm the opinion of, like, get them out of the site. 00:57:08.854 --> 00:57:12.654 Like, we should be using more American-made drones. I think that just makes more sense. 00:57:12.654 --> 00:57:15.794 I think the question of, like, what happens to the data is a whole other separate 00:57:15.794 --> 00:57:21.904 topic. About the pricing, it's another thing. But I just don't trust the ulterior 00:57:21.904 --> 00:57:25.904 motive, the genuine motive of some of these companies when we know that they're linked to the CCP. 00:57:26.704 --> 00:57:29.124 Again, this probably isn't a construction tech podcast at this point, 00:57:29.124 --> 00:57:33.484 but I would just get them out. Yeah. Well said. American. 00:57:34.850 --> 00:57:36.986 Well, that's the opportunity, right? That's the opportunity. 00:57:37.259 --> 00:57:40.040 Yeah. That is the opportunity. Opportunity there. 00:57:41.930 --> 00:57:47.664 Well, August was a doozy. Any hot takes you want to finish out the episode with? 00:57:47.878 --> 00:57:50.485 Lars, go first. What's your thoughts for September? 00:57:52.146 --> 00:57:55.767 Gosh, I wasn't prepared for this. Tyler, go. Go. No. 00:57:58.404 --> 00:58:03.024 You can't make me a plot. I was thinking. Victor's the smartest one out of all of us. 00:58:04.044 --> 00:58:08.622 Victor's going to come with a thesis. He's like, well, actually, I'd like to. 00:58:10.480 --> 00:58:14.444 Yeah, I don't know that I have a hot take. I would double down on something 00:58:14.444 --> 00:58:17.994 I said last time. I think Clear Story is one of the better companies in this 00:58:17.994 --> 00:58:20.662 space and underappreciated of their product surface. 00:58:21.097 --> 00:58:24.814 I'm a little bit surprised of how much the product direction that a lot of construction 00:58:24.814 --> 00:58:28.164 tech companies have gone down and what they haven't built. I think there's still so much to do. 00:58:28.624 --> 00:58:32.074 I think we need better alignment, though, between solutions and customers. 00:58:32.074 --> 00:58:35.344 I think that's probably the biggest hindrance of people adopting more construction 00:58:35.344 --> 00:58:39.274 tech is it's expensive and it's hard to see the value. And I think we need better 00:58:39.274 --> 00:58:40.217 alignment between those two. 00:58:40.855 --> 00:58:45.424 So can I ask this question just real quickly, because I've been thinking about this yesterday. I. 00:58:46.803 --> 00:58:52.703 Wrappers are kind of useless, but harnesses are really valuable as it relates to these models. 00:58:53.063 --> 00:58:57.423 And we know that with Perplexity, we know that with Cursor, a few others. 00:58:57.922 --> 00:59:02.853 And I had this thought of like, well, isn't that what every construction tech 00:59:02.853 --> 00:59:04.256 company should be trying to build? 00:59:04.407 --> 00:59:09.563 Is like some really contextually aware harness for our industry? 00:59:09.563 --> 00:59:12.511 And yet, in theory, we're kind of failing at that. 00:59:13.483 --> 00:59:17.549 And they're trying to be almost everything else other than an effective harness. 00:59:18.141 --> 00:59:21.553 Well, I mean, a wrapper is basically just a harness because of me, 00:59:21.553 --> 00:59:23.369 our heart is seeing other LLMs, right? 00:59:23.678 --> 00:59:28.183 So this would be... Well, I mean, maybe it's nomenclature, but I view a wrapper 00:59:28.583 --> 00:59:33.414 as less than, but yeah, potentially, right? 00:59:34.234 --> 00:59:37.083 Yeah, I would agree with that. But I think that the challenge, 00:59:37.083 --> 00:59:40.073 like if you look at Accursion, the underlying thing that they did is code, right? 00:59:40.073 --> 00:59:43.623 You look at perplexity, it's different. And so I don't know what that thing 00:59:43.623 --> 00:59:45.820 is in construction. If you're going to say it's submittals and RFIs, 00:59:46.134 --> 00:59:50.493 you could do that, but that's not the most value return task for a contractor. 00:59:50.493 --> 00:59:53.533 So I don't know that it's worth it to build a really nice harness for that. 00:59:53.533 --> 00:59:54.863 It would have to be something bigger and better. 00:59:55.203 --> 00:59:59.483 And so then I would start asking, well, what is that? Well, what are those things? 00:59:59.483 --> 01:00:02.829 And again, it comes back to financials, probably. It comes back to pay apps and accounting. 01:00:03.282 --> 01:00:05.403 So I go back to what I just said. I think ClearStore is probably one of the 01:00:05.403 --> 01:00:08.587 sneaky best companies out there. I think Shepard is really good in insurance. 01:00:08.971 --> 01:00:11.123 There's a lot of harnesses that could be built around what they do. 01:00:11.123 --> 01:00:18.093 So I'll pause. If you think about it, like coding is to the digital world that 01:00:18.093 --> 01:00:20.252 designing is to the physical world. 01:00:20.548 --> 01:00:28.884 So

like the analog of cursor for our industry is a harness that designs. Anyways. 01:00:29.852 --> 01:00:34.292 Guys, this has been fun. Victor, you had one more thought. No, no. 01:00:34.972 --> 01:00:37.952 I think that the harnessing on the design piece, it's interesting, 01:00:37.952 --> 01:00:40.722 but like a lot of people don't love some of the Autodesk products too, 01:00:40.722 --> 01:00:42.438 because like they're built in a different generation. 01:00:42.816 --> 01:00:46.912 And as we saw with the latest release from Astra and this Fable 1, 01:00:46.912 --> 01:00:50.542 I think that there's a lot to be said about like what is possible with using 01:00:50.542 --> 01:00:52.642 Blender and using your own 2D design. 01:00:52.642 --> 01:00:55.942 And so if you're thinking about truly designing a building, I think that's one aspect. 01:00:55.942 --> 01:00:58.812 But remember, at the end of the day, owners don't want designs they want buildings 01:00:58.812 --> 01:01:00.722 they need usability of the structure itself, 01:01:01.204 --> 01:01:03.482 and so we get caught up in all this technology but at the end of the day like 01:01:03.482 --> 01:01:07.002 that's what we're here for is what we got to deliver yeah i i will give you 01:01:07.002 --> 01:01:12.454 my hot take i finally figured it out steven i figured what it is it said, 01:01:13.290 --> 01:01:14.693 to death with reddit, 01:01:18.292 --> 01:01:20.968 hey you uh you walked those needles this weekend tyler, 01:01:23.632 --> 01:01:27.202 i guess without context that could seem like a lot of different things Just 01:01:27.202 --> 01:01:31.412 for clarity, Tyler was giving blood and might have fainted and doesn't have 01:01:31.412 --> 01:01:34.123 a drug problem. Lars, anything else for you? 01:01:34.698 --> 01:01:38.064 I appreciate this. I mean, I think hot take, it's, you know, 01:01:39.492 --> 01:01:45.888 so many good thoughts here as founder, as, you know, tech construction, 01:01:46.167 --> 01:01:47.747 another growing construction company. 01:01:49.272 --> 01:01:54.909 The amount of human capital it's taking to just implement basic tech solutions, it's just insane. 01:01:55.217 --> 01:01:57.487 It's just nuts. And I think. 01:01:59.793 --> 01:02:05.882 That's the opportunity because so many of these older solutions are just layering 01:02:05.882 --> 01:02:11.496 in. And I think this is the thing, Victor, not to go back to Procore, but I'm going to do it. 01:02:12.512 --> 01:02:15.212 I watched the videos on it. I'm like, this is good stuff, right? 01:02:16.682 --> 01:02:24.442 But we're bolting on like new stuff onto like old stuff, you know? 01:02:24.782 --> 01:02:29.675 And like, like, how can we make this easier? Right. 01:02:30.053 --> 01:02:35.052 And like, that's the thing. And unfortunately, I think it's going to come down 01:02:35.052 --> 01:02:40.623 to individuals in company ops who lean forward and go and create, 01:02:41.321 --> 01:02:45.925 with a, with a, with a company culture that allows them to create and innovate 01:02:46.041 --> 01:02:48.595 and, and, and, and adapt quickly. 01:02:48.850 --> 01:02:52.182 And I think those are the companies that are, that are going to somehow just 01:02:52.182 --> 01:02:55.321 like, where did they come from? Right. 01:02:55.962 --> 01:02:59.200 Who are, who are those guys? It's those innovators, right? 01:03:00.311 --> 01:03:04.630 In the words of my brother, it's like people are putting lipstick on a pig yeah, 01:03:05.152 --> 01:03:10.402 and charging you more oh yeah yeah hard to disagree i think some of this stuff 01:03:10.402 --> 01:03:12.772 should just work out of the box and it should be nice and it should be easy 01:03:12.772 --> 01:03:14.549 and it's just not sometimes, 01:03:15.782 --> 01:03:20.632 in in defense of you know we'll call it on the software side guys i don't it 01:03:20.632 --> 01:03:24.583 is hard holy smokes it is hard, 01:03:25.029 --> 01:03:28.512 so i i respect it but i but i'll also say it's never been easier than right 01:03:28.512 --> 01:03:33.142 now wow have you tried tripling your cost or price and. 01:03:35.328 --> 01:03:39.082 I don't know just an idea just see you.