Latest / Construction Technology Podcast / Construction Technology Podcast - Episode 005
Transcript
- Luigi La Corte: What's up everyone. Welcome to another episode of construction technology podcast. I'm here with Andrew from join and Mike from build vision. What's up boys. What's on your mind this week, Mike. Andrew Zukoski: from Claude is on my mind this week. Mike Powers: Yeah, contracts from Claude is an interesting one, for sure. Luigi La Corte: I this is in the, this is in the, notes for today, but I actually feel like Claude, Claude has had its moment in construction over the last four weeks. People have just woken up to Claude being a thing within construction. Obviously we've been using it for quite a long time. this is what we see in like SMB kind of mid-market contractors. Do you guys see the same thing? Andrew Zukoski: don't know anyone who isn't. pick chat GPT clot like individually, regularly, like at the corporate level, I feel like there's a huge bifurcation may even be right, but like, there's like a wide range of companies picking things up or having some formal initiative around it. like everyone seems to have chat GPT and Gemini and potentially Claude. I feel like I have heard this off. Luigi La Corte: Copilot is surprisingly embedded. mean, it makes sense given that most people are on SharePoint. But I'm actually really surprised that people are as happy with Copilot as I've been hearing. Andrew Zukoski: To what degree is that lockdown, like around security and the established vendor of Microsoft? I don't have a good feel for that and I thought you guys might. Luigi La Corte: I think it's 100 % lock-in. think it's whoever is on the Injection is a big problem. Getting everything in the right area to be able to ask the prompt and it get the right data that has the right metadata is like 80 % of the problem. obviously when Microsoft released Teams, there was famous case study of kind of lock-in bundling versus kind of like new product entrance, given that Teams ended up just... completely dominating Slack. And I think it's a very similar story here. People are just bundling or teams are accepting that copilot is bundled in and it's like, well, I have all my documents in SharePoint. And so I already have this LLM that's baked in and has all this data access. But the cloud is interesting, right? Because it's like, you can upload documents from your desktop. You don't have the same lock-in. Like I think it's... Mike Powers: luck. Andrew Zukoski: That's it. Mike Powers: Claude Cowork is now available on Microsoft computers. Claude Cowork is now available on Microsoft computers. So process of updating changing an Excel sheet or a PowerPoint is actually better than Copilot. I don't think that's been dis-remediated across the market. But I had an interview with, ⁓ not an I did a podcast with a general contractor this week. ⁓ And I her, was like, before you sign up for any software, download Claude Cowork. Luigi La Corte: See you again. ⁓ great. Mike Powers: and try it out for yourself and try to do what you're trying to do. And my bet would be you can get 80 % of the way there. And then it's on the vendors, on the tech solution to prove that they can do more value than what you can do because of the expertise they have, the X, Z they have. But again, like having someone download Claude Cowork or Claude on their computer, getting it approved by IT is I think a bigger hurdle at enterprises ⁓ than at to small market companies where you kind of allowed to run. Then we're talking a little bit later, but the DOD Department of War also just ⁓ released supply chain risk on Claude for any the, any of it's, again, it's, I know it's getting litigated, but the position right now is that any vendors that with the U.S. cannot use Claude. So we've had customers ask, is there Claude? usage in your application. So that's an interesting conversation. If you're the IT person, you're obviously on the more conservative side. You're not going to implement Claude in order to possibly lose a project with a government agency. And then this changes every week. I couldn't imagine being in the seat of a CIO right now or a head of IT at a large firm. Andrew Zukoski: It's one of the, I feel like that's like, we were, I had a conversation with these two CIOs at big eight hour firms and we were talking about like AI vendors and this was years ago, but they looked at me and they were like, you know, have very good relationships with Microsoft. Like Microsoft is probably their largest like IT line item. Like we think about all these domain specific firms like CMIC and Procore and Autodesk and all that. But like first there's Microsoft and that's a company that's done right by them as a CIO and as an organization for 50 years. Like, yeah, I would, I would reach for cowork too. Right. ⁓ my God. I've reached for Copilot too. Yeah. ⁓ Mike Powers: Mm-hmm. Microsoft's business is just like a random city. think Microsoft's business arm, so selling into businesses is a $200 billion a year That a crazy stat. Luigi La Corte: They're going to do really well. They like, they've done a good job of just not even fast following, but pretty slow following, but using their locked in distribution to completely win. mean, teams came out like what eight years after Slack or something like that and completely won. Um, and they're moving pretty quick now. They recently launched, believe co-pilot health and now cowork, which is relatively fast. Um, and if they can just emulate even like 75 % of what. Mike Powers: Yeah. Luigi La Corte: ⁓ can do, Cloud Cowork can do, I actually believe that they'll win. Because again, it's all getting their documents in. There was a LinkedIn post that went somewhat viral within construction tech circles this week, where an estimator was ⁓ people how to use Cloud Cowork. And the video, I think he released a couple of videos, but in the video, he has his file structure and says, like, this is where I have all my projects, and this is where I have my AI personalities, and this is where I my templates. And you like, you already know that like, okay, this was a mid-sized firm with multiple offices. It's just wouldn't work because everything's in SharePoint or everything's in God forbid, a custom drive that's on-prem, is like a different story. And could just imagine, okay, well any big company who's not on SharePoint, this isn't going to work. Well, at least not now. Correct me I'm wrong, but Claude Cowork doesn't ingest documents the cloud. Andrew Zukoski: I gotta imagine you can connect it. Like I've been able to point it at our hub. I haven't done anything document centric, but I pointed it at our MCP server at HubSpot pretty effectively. I gotta imagine. I mean, at the same time though, I am finding that my downloads folder like has a lot more things in it than it did a year ago or two years ago. Like I'm just, around a lot more. Like I bought an Excel license. Mike Powers: You got it. Andrew Zukoski: because the process of working with like these cloud generated Excel files is just smoother locally than like dragging and dropping it up to like Google drive every time I want to edit it or like change something. So like on the one hand, theoretically it's just data. It's just a get call to an AWS bucket somewhere. But on the other, like, I guess in practice, like I'm spending a lot more time with my local file system. yes. Mike Powers: Hehehehe Luigi La Corte: Yeah, on a weekly basis, I have to clear up my memory because I just don't have enough space on my computer anymore. But that's because... Andrew Zukoski: ⁓ whoa, what is it? ⁓ Luigi La Corte: Yeah, honestly. Well, that's because I have a lot of folders locally because I'm either working with Claude and cowork and I need to dump in a bunch of files because I'm some tests because it just doesn't grab from the cloud very well from what I see. So that's why I think Microsoft is going to win. It's like they'll figure it out. They have all the data and if they could just route their agent to the appropriate place, then they'll win. Claude, Anthropics going to zero. You heard it here first. Mike Powers: you Mm-hmm. Andrew Zukoski: That's really good. Mike Powers: I think. Andrew Zukoski: You heard it here first. I have a question. Speaking of going to zero or not, you guys both installed Claude bot or whatever the fuck it's called these days. ⁓ Mike Powers: 20 zero. ⁓ claw, Over claw. Luigi La Corte: Yeah, open clock. Andrew Zukoski: OpenClaw, yeah. it still running? Okay, are you using it more or less than you were two weeks ago? Luigi La Corte: Yes, was chatting with her this morning. I'm using it about the same. I'm using about the same. I have ⁓ a growing bookmark stack on X that I need to go back to to projects for my OpenClaw. But this is where, this is where. Mike Powers: Yeah. Andrew Zukoski: Can't you make, I thought the whole point was you can make OpenClaw go back and. Luigi La Corte: That's right. Yeah, I should, but this is where it's better than let's say like a Claude Claude code with Claude code. you need to have your computer open so that it's running in the background with open claw. It's completely on the go and it's a single, it's kind of like a single pane of glass, a single pane of glass into any of my API services and to coding. So it has my GitHub access. has access to Claude code. now it has something called ACP. ⁓ Agent communication protocol. You can bring in additional agents like a Claude code. and so let's it's ⁓ 10 at night. have an idea. I want test something early in the morning. My laptop's closed. have my, I have my phone. I'll test. I'll text my open claw whose name is Bucatini. and I'll tell let's, spin up a GitHub repo and let's, ⁓ clone this project, but do this so that I can test it in the morning and it can run a sync for 45 minutes. an hour without having to open up cloud code because it's already interacting with the cloud code. If I ask it to spin it up. So it's, it's still good, but like clods catching up pretty quickly. And I think a lot of other, model vendors are just trying to copy what is, which is basically like a react loop with lot of API access and its own, its own terminal, its own shell access. Mike, are you still using it as much? Mike Powers: Exactly. Yeah, I think it's really good for like shower thoughts. If that's like the simplest metaphor for it. Like for example, our, our COO was like to workshop some like TikTok ads as an example. So he was like just talking to the claw bot, ⁓ open claw, you call it now. ⁓ And just being like, Hey, like let's spin up like four or five of these, like go do that. And then it sends him an email to do that. We also. We did an interesting sales motion where we're going to apply for job at contractors reps to fill positions using agents. So we did like a ⁓ kind of a communication stream that. So it's more of these like ideas of like, Hey, we're just like chatting through ideas. We're brainstorming. can go do work and continue the conversation. Andrew Zukoski: in the shower. Mike Powers: I think it come back, it kind of, yeah, I like that. I like this, like that thesis of like the sour thoughts perspective where you don't have access to your computer, or your wife's not yelling at you to like get off of your computer at 11 o'clock at night. You can be like, you know, texting the bot instead. Andrew Zukoski: Yeah, don't you feel like maybe maybe maybe the problem isn't it being hard to get to your laptop at 11 at night? It's the part like you who's gonna have another kid in two months. But no, I got it. OK, so I'm loving this idea of TikTok ads of you in a shower talking to your Mac Mini like. Mike Powers: You Yeah, yeah, yeah. Yeah, I did. Yeah, I just got a Mac Mini sitting right here. It's gonna go behind me on this desk. It's gonna... Luigi La Corte: Did you get a Mac mini mic or did you just do this on mic? But why? Is the iMessage as a channel really that worth it for you? Or is it more like the storage and the RAM? Mike Powers: It's, don't know, I hate WhatsApp. I ⁓ just like, I like, not a meta fan, I'm sorry. And I just like, can't rationalize, get my head around it. I don't like going into it. I feel like every time I go into it, something else pops up. It's the green boxes. I'm telling you right now, I'm still sold on ID to see a blue box on my text. And I cannot get over that hurdle. When I joined, when I moved from Turner to Building Connected, 2016, one of our first engineers took me to the Apple store and said, you're no longer using Microsoft products and Android products. And I just haven't changed that since. And it's the best marketing ad of all time. Andrew Zukoski: In the same way like software has dark, this is very construction relevant, but the same way like some software has dark mode, WhatsApp should implement blue mode, like all the green in the app becomes blue to like give comfort to us iOS people. Yeah. Mike Powers: Correct. So like tying that back to construction and what I think is very, very interesting, which I haven't seen a lot of companies do ⁓ all right, all these customers going back to our Microsoft conversation are in Microsoft and Microsoft Teams. How do you live in that environment with them? Going back to this point, like ⁓ a, whatever you want to call it, tech person, like a very loose, defined tech person wants to live in the Mac iMessage ecosystem. So whatever we can do to tie to that is super, super valuable. Luigi La Corte: Hahaha Andrew Zukoski: Yeah. Mike Powers: All right, now we're putting on the seat of I'm a contractor, I'm a subcontractor, I'm someone sitting there. What do I actually do on a day-to-day? you looked at my history of like my, what apps I'm in, I guarantee you Outlook Teams is the highest percentage. So why does it even matter if they sign in the application? If everything can live in Teams, can live in email, can live in their Excel. Andrew Zukoski: I mean, we see that Teams is like Teams not like one for one with Slack, which I'm more familiar with. And I imagine most technology people are more familiar with, right? We actually see a fair amount of our application usage from like embedded browsers inside of Teams. Like people are able to like set up links and set up channels or whatever, whatever team, set up teams probably is what teams calls them like around And you're sort of like, even though it's an application, it's sort of like a Mike Powers: Mm-hmm. You Andrew Zukoski: 21st century SharePoint sort of thing. Mike Powers: Yeah, I just, think there's this interesting pendulum swing where, all right, if when was first selling this building connected in 2016, it was getting people off of email and Excel into my UI, into building connected UI. We're going to swing back hard where it's like, I don't want you in my UI. I'm going to want you in Excel and Outlook. And the data points that I'll validate around this are, we use ramp for expense management. Ramp doesn't want anybody signing into ramp unless you're the ⁓ the financial person that's approving everything. Everything else is done via text, via email. And I think that is the motion that all these people want us to do. Luigi La Corte: completely agree. I think I do. I actually really believe in that as well. Well, yeah, I'm not supposed to agree. But I think this is why OpenClaw had such tremendous product market fit is because you had these kind of like walled garden applications like ChatGPT and Claude where you had to go sign in, authenticate, bring your context in. But what OpenClaw did is it flipped it on its head and said, you already have your established channels. You have your WhatsApp, which Mike hates. You have your Telegram, which now I use because of OpenClaw, iMessage, et cetera. Mike Powers: You're not supposed to agree. ⁓ Luigi La Corte: Talk to your LLMs where you work, email or through some other channel. ⁓ I think that's why it's gotten the GitHub stars in all of history is because people are like, well, I'm already here, but now it's completely accessible and it's not through third party application. Andrew Zukoski: the app can add the GitHub star. mean, and OpenClaw no doubt is adding GitHub star, but yeah. ⁓ Luigi La Corte: sure they're all legitimate. They're all legitimate, I'm sure. ⁓ Mike Powers: So Andrew, why not? Give me the pushback, why not? Andrew Zukoski: Yeah, well, like, management makes a ton of sense, right? Like, I hate expenses and I get nothing out of it except money. I like the money. But it's like a completely transactional thing. Like, I don't have to interpret any information that comes out of it. Really, the only person that should have to interpret it is the CFO or like the controller, whoever's like managing the expenses, right? And so makes sense. like, Luigi La Corte: Just one for one. Andrew Zukoski: Like our CRM usage is like split, right? Where there's like, I just have to log stuff. Gong has been amazing because now people who are interacting with customers have to go type notes with their fingers after most customer interactions, right? Like you don't have to log in, ⁓ like, Hey, I want to review this history. Like, yeah, they log in for that. And don't know if... Like every piece of a technology stack gets refined through usage, It's like the contract example. Like we were talking about how, like if you use to generate a contract for something, it's a contract that's never been used before. and it's likely to be extremely expensive because both legal teams that are agreeing to this are going to have to pore through it in detail and they might or might not be familiar with the language in it. It's just like by virtue of being new, it represents substantial risk. that's true at the database level. Like, are we going to store the right fields? Are we going to like have the right model for the universe? It's true at like the user level is the user familiar with how this workflow works? they have they done this before? They know what they need to look out for. like, it's also true at the application layer, which encompasses the logic, the thing that matches the to the account or whatever, but also just the UI. Like is this UI does the UI like make it obvious that there's a next page of receipts that you need to go through? Like if it's true and it's not that like open claw would be bad at that. necessarily, it's just that it would be new every time if you're like coming in and doing it via text through one of these communication things. So I think there's still a space for the, I think the UI is still and that it's been beaten on over and over by you the rest of the industry. Mike Powers: So again, like I think the meta point would be that UIs are built for humans and in the future agents are gonna be talking to agents so there doesn't need to be an AI. That's where ramp, that's the path ramps going. So in the construction world, like if we're talking about, let's take bid management, like sending out bids. I'm a general contractor, I send out bids to my subcontractors. Do I care where the button is to add a bidder? Or I want to make sure that I get the right bidders, the right information. if I can sit in my email and say, Hey, like I get a project from an owner for a hospital. the application, whatever the application is, ⁓ is fed my bidders list. so I say, do all I'm getting is, I want to, do I want to bid on this project? Yes or no. If yes, send everything out. There's agents working the background code, pleading these loops. So. ⁓ getting bids, creating bid forms, responding to questions and RFIs. Why can't that just be in an email format or a text format opposed to having to sign into a tool like Building Connected to see what my users did? And if you have a question being like, hey, where's Kevin at on project? Why couldn't it engage an agent to call Kevin and say, Kevin, hey, are we bidding Yes or no? And is letting it respond to your email. or to your phone saying, Kevin's gonna respond or let you call the person, doesn't matter. But completely pulling away the idea that there's any value in that UI, I think is very, very interesting. There's engagements and things that you're like, hey, we need to level out a quote. So you wanna see all the quotes together. I think that's like, there's a UI perspective around that, but why could that just be the Excel sheet that I've been using for a hundred years and just updates and collaborates. So there's all these different things. I'm just like trying to think like if... Andrew Zukoski: ⁓ bid leveling is pretty so heavy. Mike Powers: If you were Andrew, if you were to say, we started Join today and we didn't want to have a UI and just invert that thinking, how would the customers engage differently with the application? What would be the pros and cons of that? It's just an interesting mental exercise that I've been trying to work through. Luigi La Corte: Yeah, there's going to be an expand. Andrew Zukoski: I agree that it's an interesting mental exercise. Luigi, you're gonna jump in. Luigi La Corte: I was just going to say, I don't, I don't think, think there's a limit to this trend, right? So if the is swinging back to let's say, on local memory, lower UI, so existing channels, like let's say through OpenClaw, I think that has a limitation. think that limitation is collaboration software. And I think it's because your existing channels do a bad job at collaboration. It's like barely multimodal. So within WhatsApp, for example, it's effectively just text. It's effectively text and PNGs, but you need a lot more of that if you're doing, let's say, value engineering within construction. If I'm going to advocate, let's say, for Andrew's point, I see a lot of value in the common UI for some of these complex tasks. you're talking about logging expenses, this is like a single player game. You don't need to go line item by line item. You just need to send it to the right place. And ideally, you're doing as little logging as possible. Andrew Zukoski: the other thing is the ⁓ UI. ⁓ us, and I want come back to building connected, but for us, it's the fact that everyone looks at the data the same way is part of the value. That the whole team gets around and looks in at the data and they see a common presentation of it that they can all get comfortable with and then that is consistent through time is lot the value of join. Mike Powers: Yeah. Andrew Zukoski: But Building Connected is this incredibly durable product. It's this amazing, I mean, we're starting integrations with good management things and like, doesn't look like it's a suite of products we need to integrate with to cover our customer base. Like way to go Mike, way to go Building Connected team. And when we talk to people, like Building Connected talked a lot about the network. ⁓ Mike Powers: You Andrew Zukoski: But that doesn't appear to be why people love it these days. People fucking love the UI. Like that's money. People how well it works. And I just like, I guess if you think that that whole knowledge the whole bid management thing that like time spent on bid management is going to zero, like, and you don't need a UI. Like if a person, Mike Powers: Mm-hmm. You Luigi La Corte: I think human time spent was going to zero. Andrew Zukoski: Human time spent on bid management, if you think that's going to zero, then you don't need a UI. But if you think it's going to 2 % or 1%, like it's like a ton more is automated, but I still have to get in there. Doesn't it still come back to? like the value, like the way that the person will engage, like that comes into audit it and to QA it and like, instead of 50 people doing bid management, you have one person doing bid management, but like, boy, do they have to get in and really quickly understand like a wide array of projects. They need to do it consistently. Like the UI doesn't live, like the UI doesn't live just in the computer. The UI also lives in people's brains who become familiar with it and know how to interpret it. Like, Like I have dreams sometimes where I'm just like looking at like Visual Studio Code. They're actually great dreams. They're, it's great man. But like it lives like I have inside my head neurons that are configured to understand various UI as well. I'm familiar with them and I can very quickly interpret information that comes from that much faster. I would say then like similar text information because text you got to like read it and you're saying well you can start to like have a standard format for it and use like bullets and so forth. I'll be like yeah we're recreating DOS like if start talking about consistency of presentation like comes to the UI. So I don't know man. like... anyway. Luigi La Corte: No, on that point, you will still need a UI, but it's gonna change. It's just the user is changing. The user no longer has emotions like us. The user has like a single job. And I think it's actually really exemplified by this post from Zachary Batt. This is the viral post that I was talking about. So he says, my coworkers think I've gotten dramatically better at my job. He goes on to basically say that he's discovered Claude Cowork, okay? And he has a bunch of commands and this unique folder structure. And let me just kind of jump to the point here, um, where we can talk about how this is going to change. here he's using Claude cowork Claude cowork. I mean, if you don't know is, um, it's a made by a company called Anthropic. Anthropic is an open AI competitor and a cowork is kind of like chat, except it has access to a specific folder. here you can see he's pointing into his local drive. Now, like what, what's the point of that? Andrew Zukoski: you Luigi La Corte: Take this off. So what he's doing here is he's going through his folder structure. has some instructions for his agent and then his agent and taking in some documentation like an RFP or maybe drawings or specifications spits out a bunch of stuff. And then he ends up generating scopes of work. ⁓ So what's the point of this? If a general contractor and we've, we've had this conversation before where we think, let's say CM fees are going down or trending down over time. If a general contractor is just taking an information. And now conducting some operation on that information to spit out questions and scopes. What is it? Like, what is the real value add here? It's probably not as much as it was in the past where your output and your bid was actually a function of how well you did that process. If that process is, you know, being, let's say unified by agents. Well, number one, now your bid manager is actually an agent. And if you need to use building connected, now that agent has to go use building connected, but it would be. more efficient if you gave it a command line interface. And now you as a general contractor, your value is no longer in how good your process is because it seems like it's gonna somewhat diffuse. The value of that is gonna diffuse over every general contract. Andrew Zukoski: Like if everyone has, it was one of our, I'll give a shout out as well, to get Bolt in a conversation and raise this really interesting question. He's like, look, what if the differential in proprietary goes zero between us and our customers? What if owners, ⁓ market, completely prices the work. Like it's just knowable to everyone at all times what a given scope of work is gonna cost. Like how does our business change? that's, that in that case, like if that has happened, actually I would argue human building connected usage does go to zero and it's like best done through text. what, here's a set of like. What does this cost? Could just be text. Yeah, that's pretty fair. Luigi La Corte: I do think that any operation, like any deterministic operation within a set of documents will be competed away. Meaning, you know, people who are need to, okay, need to set up, have a set of documents, generate my sub middle review. Okay, that's going to be competed away because over time it's just becoming really easy to do that without anything proprietary. And now out of the box with Claude, you can get like 80 % of the way there, maybe 85 % of the way there. So. The value is not gonna be in your process for understanding this information. It's gonna be, it has to be other things. Andrew Zukoski: But won't the, one of the, I'm of one of our customers helped us really understand pre-con over time and she described us, she was like, yeah, look, we do all this analysis and all this back and forth, but the fundamental value of pre-con is actually like mobilizing and engaging, like the people are gonna perform the work, the superintendents and the ops people, like getting them engaged and having them understand the project earlier and have a better like, you know, sense of it and readiness, readiness to go build this thing. Like when the work actually commences and that the, lot of the activities that happen, like, yeah, there's a knowledge product that's produced, but then there's also like a change that happens inside, you know, inside the superintendent's head with like, okay, I have an understanding of this project. I better understand the owner's goals and like, I'm more confident in the constructability of this thing. And I'm not going to yell at the pre-contain for throwing it over, throwing it over the wall to us. even if it's still exactly the same when it gets thrown over the wall. Luigi La Corte: I think, I think that's right. I think there's, there's a lot of components of value within pre-con and I think the process element is just, it just won't be as much of an advantage, but it probably will be within your relationship with a customer if applicable. And I think like the stuff that you're doing at join, think is less susceptible because you're welcome because you're not dealing in a, you're not, you're not dealing in like a price shoot out environment for the most part. Right. And like, Andrew Zukoski: Thank you. Yeah. Yeah. Luigi La Corte: is gonna be, I you know, you may even be able to do that completely with agents at one point, Where you have an agent just win the job because it's completely deterministic and it can mobilize subcontractor labor and like there's no value at all. Andrew Zukoski: Like we have a, we moved into a new house and like apparently the suprilateral is 100 years old. We need to get it replaced. The joyous of homeownership. And just getting prices has been, this is different universe than we talked about, right? Like this is a $10,000 construction job, When it happened, I'm not thrilled about that number, but like just getting a price has been very challenging. Like. Like people are busy right now and it's it's interesting. It's just been very challenging to get that price. And there's some site knowledge that you need to have, like, like did you, like I can answer all these questions. I've answered all these questions for so many people already. Like I can't believe I'm like answering it's one unit or two units again. Mike Powers: But how much of that is you like, if you think about in commercial construction is all of these contractors just overloaded with work. And even if an agent can do an automatic takeoff. Like, are you as a contractor to want to do a thousand projects? The answer is probably no. ⁓ Andrew Zukoski: Thank you. Yeah, like, what if you win them all? Luigi La Corte: Why? Why wouldn't they? Mike Powers: So you can't. So before AI, I'll give you an example. Going back to the building connected position, we obviously had a really solid general contractor product, help general contractors bid projects. We had a million monthly active users in the application. So we're like, hey, we have every eye of every subcontractor. Let's go build tools for them. Let's go help them manage their bid board. Let's go help them win more projects. The feedback we got from the majority of the subcontractors is we don't need to win more projects. We need to win better projects. So by giving tooling to making them more efficient, more effective in doing takeoffs, X, Y, and Z actually doesn't move the needle for them. A mechanical subcontractor in Buffalo, New York can only have X capacity. They want to go work with specific clients. They want to go work with both specific general contractors. They don't want to get a bid from a McDonald's if they do hospital projects. It doesn't matter if they have all the tooling in the world to do that. They cannot support the labor infrastructure around it. I think that's the bigger friction point. Going back to our combo last week about Anthropic, we're only really solving this 20 % piece, which is like what's done on a computer right now. And to like the reason why I keep giving props to join, but join is a collaboration tool where there's multiple parties. that are never gonna be on one Claude code or one Microsoft instance. So there's huge amount of value for having a UI, a collaborative environment, so all these parties can come in. Anything outside of that is getting commoditized down to zero. So you're not collaborating with multiple parties, it's gonna be commoditized down to zero. you saw that with like, if you think about like how the banking world, everyone's like, banks became more efficient with ATMs and their moved into other roles. What happened in that actual market? was banks saw their competitors getting ATMs and they were losing customers to their competitors because they had an ATM and you'd have to what you could get cash out 24 seven as opposed to banking hours. So that bank was forced to have an ATM tool. you think about that in construction, all of these contractors are gonna be forced to have an AI tool. So now it's the focus of like, where do you actually provide value to your customer? How do you collaborate better with your customer? do you offer additional services to your customer? And that's where the value prop actually comes in. So if all of these tools and functionalities and workflows are gonna be commoditized down, it's where do you expose and provide additional value to your customers is where the real value will be driven. Luigi La Corte: It's interesting example, Mike, because when ATMs came out, tellers didn't go away. Like it just increased surface area of your offering. Mike Powers: So I think the on it, like the historical story was like, ⁓ it did replace labor because those, like you can see there's more bankers, people in the bank industry than there were pre-ATM. I think the actual like first principle change was that it normalized something and then it gave those businesses saying, hey, this a nice to have anymore, it's a must have for our business. But now I need to figure out how I make more money. Because before if I was the only bank, in my city and I was open from to five and I had four tellers and the customers would come in deposit. Now it's everything's online. I need to figure out how do I as a bank operate to be more efficient? So you saw this massive, like everyone's like, there's no more regional banks, there's no more regional banks because there's so many efficiencies that are layered on in a large organization that why wouldn't I just go to JP Morgan Chase as opposed to local bank? And you think about that, how does that shift to construction? If you're like, hey, I'm able to do these back office workflows automatically. Andrew Zukoski: Yeah. Mike Powers: I'm building a relationship with the customer. So, hey, I'm an owner that's developing buildings the country. I'm going to want a national general contractor that's doing everything automatic, but I have strong relationships with the end parties that provide me additional value. Like I'm doing offsite manufacturing so I can do modular construction of my McDonald's. My McDonald's get developed and installed in two weeks as opposed to eight months, and I'll pay a premium for that. We're seeing this play out in the lawyer space with the same perspective of like, a due diligence process used to take three months, Harvey comes in and says, hey, we're going to help you do it in one month. If that three month bill was $8 million, a lawyer can now charge 16 to $24 million for a one month process because the deal gets closed faster. That's commoditizing away all of the actual due diligence process, but it's giving them the ability to offer these additional services at a faster time slot. Andrew Zukoski: Yeah. Well, wait. Today, the lawyer can charge 16 or 24 for doing it faster, but they compete and they're not spending any more labor on it to do it in a month. So won't that number get driven down as these tools permeate the market and as iBankers and companies learn, oh, we're not paying 24 million. Mike Powers: I think in that world, you're relying on the relationship with the lawyer, you're offloading it to the lawyer, there's a reason why you're an iBanker, you're not doing this in house. So in that example, you're looking at the bottom line, hey, if I close this deal in one month as opposed to three months, I see this amount of ROI. Andrew Zukoski: Yeah, but like you're gonna like other legal firms will come to you and say, look, we implemented Harvey. We can do it too. And we'll charge you 10 instead of 12. Mike Powers: Mm-hmm. Yeah, I think there's a network effect portion of that where like, it actually combined into a couple of only platforms where Lagora and Harvey were the only winners because they'll be able to offer competitive pieces at scale. They'll have proprietary data. They'll have embedded workflows with these customers. Like a DLA Piper as an example, isn't going to continuously swap back and forth between Harvey, Lagora and whatever new tool comes up. ⁓ think there's a lot of foundational pieces that will like prevent that from happening. ⁓ a go-to-market motion. Andrew Zukoski: Like they will be able to screw their scale, they will be able to better use it, and like they will just have a better, they will have a better cost basis. Yeah, okay. Mike Powers: That's all I go to Markimotion. Correct. Luigi La Corte: That hasn't diffused in the market, but I definitely see a world where as soon as you ingest the documents, it's one click processing and you get everything. Here's the scope, here are the questions, here's the uncertainty, here's the quantity takeoffs, you know, here's the contract review. But if that's true, then as a contractor, you either need to be a lot more choosy. need to, what people really want to your point is to increase their win rate, but to do fewer projects. They want better quality projects. Sorry. Mike Powers: win better projects. They want to win better projects. They don't want to win more projects, they want to win better projects. Luigi La Corte: They want to win better projects. I do agree. So which reality do we think is likely? One, general contractors end up bidding a lot more projects. ⁓ bid a lot more projects because it's now it's really easy. The cost to bid a project precipitatively drops. I'm not going to say towards zero because I think it's, we say that too much, but goes down a lot. So if the cost to bid a project is now ⁓ 10 of the cost, I can widen my aperture if I want to grow. Two, they bid less projects, they add value in a way that can't be commoditized through an LLM. Or three, the general contracting model fundamentally changes where now it doesn't make sense if you're just doing CM. You actually have to provide some labor for you to have some kind of competitive advantage. Which universe do you think is most Andrew Zukoski: That's happening. Mike Powers: So my position would be the at the general contractor are gonna have their salaries drastically increase because of all of these workflows are commoditized. The owner is gonna wanna work with Andrew. I want Andrew running my job. I will pay a premium for Andrew. Nothing else matters. It's ⁓ sorry, Andrew as a general contractor. I ⁓ Andrew as the PX on this project running this hospital. I will pay a premium for that because everything else is Luigi La Corte: Andrew as the CEO of JOIN or Andrew as a hyper... Andrew Zukoski: I just like stand in general. Mike Powers: commoditizing away. So everything comes back to the labor functionality of that. The 80 % of the work that's not done on a computer is the largest value prop that's going to be positioned. think contractor employees will be able to charge a premium for that service. Andrew Zukoski: The the CME employees, because like one of the trends we see is our customers, when we talk to them about year over year growth, they're all like the big ones, it's crazy numbers, a lot of it driven by data center, but their craft workforces are often growing even faster. Like there are big craft labor pools inside of these general contractors now. And so I would say there's evidence of three happening already. And I think that's, I mean, people have been talking about how like how many bidders you see in part driven by building connected as a general contract. order to the job out to bid, it's not uncommon to see like 10 to 15 general contractors bid it, like each of them struggling to get scopes, like three qualified bidders for each trade. that, pre-LLM, technology had reduced the cost. Like you don't have to drive to the fucking plan room and like buy a physical copy of a stack of plans to like. Luigi La Corte: Hmm. Andrew Zukoski: take a look. So yeah, I think you can see three. I think three have the others might happen as well. But I think three I think you like, see more engagement in services are delivering value into the end structure, like which could be craft labor, could be supply chain, it could be like equipment centralization, it could be financial methods like leveraging scale and balance sheet and financial sophistication to deliver things more Mike Powers: So everything Turner is doing at scale, wish. Now, I was, yeah, go ahead. Andrew Zukoski: I was going say everything at Build Vision lets people do it scale. Luigi La Corte: And you also know, like we, think we've touched upon this before, but I also see subcontractors moving increasingly into the general contracting box because now they have alpha, right? They have an advantage. the majority, let's say I'm doing a hospital project. If I, if the majority of the cost is &E, well, I already have a cost advantage over you as a general contractor. And sure, I need to do a few more things, but if understanding that risk and quantifying that risk is now trivial, doesn't. Mike Powers: Mm-hmm. Yeah. Andrew Zukoski: Yeah. you Luigi La Corte: there's not really much of a hurdle for me to do that. like, I could probably compete against you and win, you know, four times out Mike Powers: It's the people. It's the people behind it though, right? they want the mechanic or the guy that's been running Mechanical Scope for 30 years on that hospital project. They don't want Mike, Luigi, Andrew, mechanical contractor that now has an AI tool to go do that project that doesn't know what an elbow is on a duck on pipe. I think like this rubber band back to like the actual like trade. is going to accelerate and they're going to be some of the highest paid people in the country because of that. Andrew Zukoski: bad outcome. Luigi La Corte: Yeah, I can see it. I can see it. Because that work is not, the for it is only increasing, but we know we're very supply constrained. But everything surrounding it, the kind of like the homework to enable that work is becoming a lot easier. Meaning like an owner is going to be able to spin up a project faster, ⁓ somebody will able to underwrite that job faster. And now you connect the demand and the supply side a little bit more cleanly without all this, all this hurdle. I think Andrew Zukoski: Yeah, much less intermediation. Luigi La Corte: Like all the inefficiency, the money that was given towards people who were in the adding value, I think that does go down. But now there's probably a more direct conduit to the actual labor on site. Mike Powers: And I think the most successful companies will be the ones that have very, very strong programs to upskill their employees so they can bring in someone from college and get them into a position and commit to that person for 10 years so that in 10 years that person's there. Andrew Zukoski: Yeah. What's the time to grizzled PS? Can you shift that in? Mike Powers: to value, to value. ⁓ Cause again, this, I literally the first thing I did on my job, the first job I was at a Turner was read the spec book. It was in 17 binders. I read the entire thing. like, don't like telling a chat bot being like, ⁓ if you think about it, it's like, all right, that, that value problem, like now I was understanding how those nodes connect. Getting this answer through a chat bot doesn't tell me how those nodes connect. Andrew Zukoski: And now you're in tech. Yeah. Mike Powers: cool, can text it and chat with it and give me examples. But at the end of the day, like that foundational knowledge of just like you're in the field being like, ⁓ that doesn't look right. You got to run into errors. Like there's not a, there's not a right answer. We were doing like a retrofit of the Eagle stadium and sold the value prop to the owner because we built the stadium 10 years ago that we knew where bedrock was. I spent the entire winter drilling micro piles in the parking lot on our dime to find bedrock. It was there was three feet of snow on the ground. I remember it clearly like this. But like that, like that's that's not solved by someone in Silicon Valley saying here's your chat box. Andrew Zukoski: Yeah. Mike Powers: It was fun. Andrew Zukoski: That's great. Gents, I feel like that's a good note. I gotta drop. Luigi La Corte: Yeah, yeah, we'll get it there. We have one out of three topics for the week, but we'll push it to next week. OK, let me sign off. on. Mike Powers: Hahaha! Love it. Good stuff, Aids.