0:00Tristan Wilson, CEO and founder of Edgevanta, says it makes AI agents for civil estimating. He worked as a project manager and estimator at his family's Louisiana road construction business before starting Edgevanta in 2022.
2:27Jordan Metzner shows Edgevanta Tycoon, where you bid on projects and unlock agents like a bid brief summary agent, plus a highway game where you crush and heat rocks, then pave and roll the road.
5:18Tristan Wilson says estimators turn specs, drawings and bid packages into a scope and cost. A highway bid package can run 1,200 to 1,500 pages, maybe 80 to 100 of which matter, taking him four to six hours to review.
8:05Tristan Wilson says he once submitted a $6.5 million airport bid 19 seconds late that would have won. He says contractors leave about 8% on the table on average, and AI can now parse documents with human-level accuracy.
10:00Tristan Wilson says agents trained on civil construction data pull out key information for human review. One estimator's two hours of typing bid items from three bid forms into Excel became two minutes and a CSV export.
12:08Tristan Wilson says Edgevanta gathers historical pricing from Department of Transportation websites for agents to query. His family's company cut money left on the table from 11% to about 7%, putting almost $8,000,000 back.
13:30Tristan Wilson says civil contractors get tens of bid packages a week. Agents give a structured summary of whether a job fits what they normally bid, saving 30 to 45 minutes on jobs they will not pursue.
15:45Tristan Wilson says no: AI makes estimators more efficient, but they still review every page and remain responsible for the job. He says AI caught an owner's conflicting completion terms that a careful reviewer had missed.
19:37Jordan Metzner says Nvidia is investing about 2 billion in xAI's round. Sam Nadler worries this circular financing could signal a speculative bubble, while Tristan Wilson says many people who bet against Elon were wrong.
22:11Jordan Metzner says OpenAI's estimated spending is close to a trillion dollars against revenue of about 10 to 15 billion. Tristan Wilson calls AI underhyped and says 15 to 20% of construction firms they talk to use enterprise AI.
AI Construction Estimating: Agents That Read Bid Documents: the key questions from this episode, answered with quotes and timestamps.
Jordan Metzner [0:00]: I built games. Edgevanta Tycoon. It's, kinda like roller coaster tycoon. This one is, another construction game.
[0:13] Theme song: Built this week, breaking it down. Built this week, we show you how. A fresh idea, a clever tweak, you locked in shoe. Built this week.
Sam Nadler [0:16]: Hey, everyone, and welcome to Built this week, the podcast where we share what we're building, how we're building it, and what it means for the world of AI and startups. I'm Sam Nadler, cofounder here at Ryz Labs. And each and every week, I'm joined by my friend and business partner, Jordan Metzner. And this week, we have a guest, Tristan Wilson, who is the CEO of Edgevanta. Tristan, give us a thirty second intro, and then I'll I'll get into the docket.
Tristan Wilson [0:52]: Sounds good. Thanks for having me, guys. Appreciate it. Sam and Jordan. So my name is Tristan Wilson. I'm the CEO and founder at Edgevanta. We make AI agents for civil estimating. A little bit about my background. So I actually grew up in the construction industry. So I worked as a project manager and an estimator, and my family was actually in the road construction business in Louisiana. So we sold our company. I moved up to Nashville and started Edgevanta in 2022. So I kinda felt a lot of the pain that estimators feel, which is like going through really a bunch of bid documents and having to, like, corral all this information and try to make, decisions, and there's just a lot of data to go through. So what we're focused on is helping estimators do their work faster, and more accurately, and we're doing that with the help, of AI in civil construction. So that's a little bit about us.
Sam Nadler [1:40]: Perfect. And we'll dive into the Edgevanta product in a little bit. But right before I cover the docket, please like and subscribe. We launch new episodes every Friday and, available on all your favorite platforms to listen to podcasts, whether it's Spotify, YouTube, or Apple Podcasts. And so today, we're gonna cover, something Jordan built that, hopefully, is is a fun kind of tool with Edgevanta in mind. Then we're gonna go into a little bit more in detail, of the Edgevanta product and then cover some of the latest and greatest news, over the past couple days in a in AI and technology in general. So with that, Jordan, why don't you kinda walk us through what you built?
Jordan Metzner [2:27]: Hey, guys. How's it going? Another exciting week in Built this week in AI and in the world in general. Alright. Let's get started. So Edgevanta helps with construction projects. So I built a different games. Let's see kind of how fun each one is. And hopefully, you guys like it. The first game I built, it's called the Edgevanta Tycoon. It's kinda like roller coaster tycoon where you build your own roller park except in this case, you need to bid on construction projects, and then you gotta build them out, and then you gotta grow your company, and you gotta use Edgevanta to help you do that. So let's first go through the Edgevanta Tycoon. Let's start building. Okay. So when you're in an Edgevanta Tycoon, you have some available projects. Let's look at the downtown parking lot. This is our first project. Okay. So the budget is 150,000 in thirty days. It says to recommend a bid between $1.20 and $1.50.
Jordan Metzner [3:33]: Okay. Let's do $1.25 and thank God to Edgevanta for giving us the suggestion here. And then oh, wow. Okay. Let's here we go. Well, they're supposed to be like kind of these agents and then you're supposed to be able to buy them. Okay. Cool. I unlocked an agent and now I have, an Edgevanta agent. This is a bid brief summary agent. And there's some additional agents here, but I need to earn more more projects in order to do that as well as buy additional equipment. I just need some more capital. I only have 50,000. But I could try the real Edgevanta if I just go to edgevanta.ai. Okay. Let's go into my next game. This one is another construction game. So let's see. Let's start construction. So I think the first thing you do is you click to crush the rocks. So you got the rocks, you gotta fill your container. You wanna get them filled up, and then once you get it all filled cool. Then we can heat up the rocks because we're gonna build some highway. It's heating up.
Jordan Metzner [4:42]: Alright. Now we gotta go to paving mode. Alright. We're gonna go to paving. Let's see if we can pave the highway. There we go. It's paving.
Tristan Wilson [4:53]: Nice.
Jordan Metzner [4:53]: Alright. Now we gotta roll it out. Alright. Come on. Let's roll the highway out. Alright. Click and hold the roll. Tried my best here, guys. Maybe done my maybe wasn't my all best, but I paved the highway. I used Edgevanta to do it, and it wouldn't all be possible here without Edgevanta. So but, I think all of this just comes down to it. Know, I'm just real really no expert in the highway construction space, but I do like to make some fun games on the Internet. So, let me just leave that into, you know Tristan, why don't you tell us a little bit more about, like, what is Edgevanta? Who are your customers? What do people use it for? And how does it work?
Sam Nadler [5:37]: Question. One of the things that I think is so interesting is, like, this is from my you know, I'm not too familiar with the industry, but it's a kind of a what I'm guessing is historically old school industry. And, like, the part of the process you're tackling, could it will significantly benefit from AI. And I do want you to mention kind of, how that how the presence of AI has really, created some significant efficiencies for the role you're solving for. That's a pretty cool
Tristan Wilson [6:15]: That that's perfect kinda entree into, like, what a little bit about kinda what we do. So I think a few of the challenges that, like, we hear from, the, like, the tens of estimators that we've talked with a little bit about, like, what is an estimator? So an estimator is is a professional that's responsible for on the civil construction side of basically taking a bunch of specs and a bunch of drawings and bid packages and basically figuring out, okay. What is the scope of work? And then how much is it gonna cost us to go do this work? And then how do I make sure that I fill out all of the forms and the requirements and the schedule and everything appropriately to submit a compliant bid? And how you kinda measure success is a couple of things. One, is our bid, been accepted, right, by the owner? And then two, how did we perform on the bid? Were we the low bidder? And then if we win the job, we get to find out, how did what we thought it was gonna cost us compared to what it actually cost us in the field. And the this whole process, like, the way it kinda works today is, like, let's say I'm an estimator in, South Carolina. I get a bid package from a county for redoing a bunch of roadways. So we're widening highway, for instance. I get, like, 15 anywhere from, like, maybe 1,200 to 1,500 pages of documents.
Tristan Wilson [7:29]: So a ton of specifications, bid forms, drawings, etcetera. And I literally have to go through and physically read every single page and then summarize what matters. Because the way it works is there's, like, the like, of the, 1,200 pages, there's probably only, like, maybe 80 to a 100 of stuff that really matters that you wanna make sure that you capture. So you really it's like finding a needle, in a haystack, and it just takes hours. Like, it literally takes I mean, when I did it took me, you anywhere from, like, four to six hours per bid just to review the documents, not even start to take offs or, do my, like, analysis. So that's really important. And then the other part, the other challenge is, like, if you miss a small detail and I actually have a horror story about this. So I was the estimator on a six and a half million dollar airport bid, and I missed a small requirement in the bid form so that we had to submit something from our subcontractors, and they did it in an addendum, which is like a change kinda last minute. And I messed up, and I got all the forms late. And I submitted my bid, which was 6 and a half million dollars nineteen seconds late, and we would have had low bid and won the job.
Tristan Wilson [8:39]: And it really, really, really burned. And those are the types of things that that that happen and that should never, right, should never happen again. So there's that. And then there's also, like, figuring out how to price the jobs. There's, like, market information that's out there. And if you in an ideal world, you would try to predict what do you think the competitors are gonna do because in low bid construction, you don't get any multiple shots at it. You get one shot. So everybody submits their bid. They open the bids up. It's called ribbon read, and you either won or you didn't. And, also, there's a big amount of money that gets left on the table in between first and second place. So those are kinda like the three things that, that we really help people with, and they're really expensive challenges. So, like, if you got a five person estimating team that spends 20% of their time on spec review, that's a $100,000 that you're investing in that every year. If you miss one small qualification on a bid, that could cost you million a million or millions dollars worth of work, and you leave about, 8% on the table on average on a lot of the public work. So, what's different now, like, what's changed is that AI can now, like, parse and understand documents with human level like accuracy, and it was not able to do this, a couple of years ago, and now it can.
Tristan Wilson [9:59]: And I think a lot of people, they hear that, think that, okay. Well, this is gonna all of a sudden, like, replace estimators, but that's actually not what we're seeing at all. We're actually seeing that these tools can actually just make estimators a lot better, improve their, quality of life, and basically make their lives a lot easier. So part of how we're able to do that is through AI agents. So, basically, these AI agents are the way it works is I have a bunch of bid documents. I upload those bid documents into Edgevanta, and then I have a bunch of AI agents that go through and parse through the documents and are trained on civil construction data, and they basically help me pull out all the important information out of the document. And then I review that. I am the, what's referred to as human in the loop. You guys are familiar with this, but I check that as I'm doing it. And a small example, like, was sitting with somebody in South Carolina last week.
Tristan Wilson [10:58]: She was working on a bid. She had three different bid forms. And the way that she would have to do this in the past is she would have to literally upload the p she would have to get the PDFs side by side and then type in into an Excel document because these are not formatted in a really good way. And that that could take two hours to type every single one of the bid items for the three bid forms. With our platform, she uploads the documents, presses a button, and there's an agent that's designed specifically to do that, go extract all those items, and she can literally press two buttons and exports it in a CSV. That two hours, gone. You can do it in a matter of two minutes. And then you can spend time on the stuff that really matters, which is like, how do I go build this job in the most cost efficient way and make sure that I have, a compliant bid? So that's kinda one small example of kinda the broader picture of how do we get people time back in their day with the help of AI. And then the second part of it is how do we make sure that we, like, win jobs but leave as little money on the table as possible?
Tristan Wilson [12:07]: So how we do that is we go to the Department of Transportation websites. We gather critical data, historical pricing information. We bring that into our platform, and then we have agents that are built on top of that that can help us call and query all past historical bidding information so that we can really understand, like, where's the market at for this job right now? How much work do our competitors have? Then most importantly, like, what do we think is gonna happen in the future on upcoming bids? In the past, you would have to jockey Excel spreadsheets. At my family's company, we actually had somebody who did this full time. All they did was bid tab forecasting. With the help of AI, we don't we can actually that we can have some of our team members now do that, and we don't have to go hire, full time staff, multiple, additional team members. We can basically do it kind of in house with the crew that we have.
Tristan Wilson [12:58]: So and that can really help add to our bottom line. Like, just for instance, at our company, we were leaving 11% on the table on all of our bids. That was tens of millions of dollars for us. We were able to reduce that down to closer to about 7 which meant that we put almost $8,000,000, back in our pocket. So that's worth spending time on. That's worth spending money on and investing time in. So that's another part of what we're really able to help people do is not just the time savings, but also, adding money to the bottom line through having better data.
Jordan Metzner [13:30]: That's awesome. It's a great application of AI, and it seems like a lot of these things, as you mentioned earlier, just would not be possible without AI previously. And I guess one also question here is, I mean, if it took so long previously to do a bid, I guess you'd have to be selective on which bids were worth, like, worth working on. And maybe with Edgevanta, you can bid on more jobs, simply since they take so much faster to bid on as well. Right?
Tristan Wilson [13:57]: Yeah. So part of another thing that we have helped with is the go, no go process. So, like, you get, like, our customers. So who are our customers? They're civil contractors, so they do site. They move dirt. They put in pipe. They lay pavement. So it's a lot of kinda heavy infrastructure, site work, road building, that those types of folks. And they get tens of bid packages in the door per day or, per week, sometimes even per day. And it's really hard to keep track of it all. I mean, it's it's like a constant flow of is is this interesting to us? Like, does it have something that we might be, worthwhile looking into? Part And of what we've done is really helped with that go, no go process the way it used to work. Like, I remember when I was an estimator, somebody would be like I would get, like, five emails, they'd be like, look at this job. Look at this job. And I'd to, like, download the plans and go, like, look at it, and the plans are, like, sideways, and you're, like, trying to click through and be like, alright. Is there anything that's of interest into us in this job? With these agents, we can just upload the documents and ask it a question, or even better, we can get a structured output that gives us a summary of the bid that says, yeah. This is what you normally bid on.
Tristan Wilson [15:07]: This is definitely gonna be of interest to you, or, this job has nothing. It's a, medical building or something. Like, this doesn't have anything that you might be interested in. So even, just getting those cup even that, thirty, forty five minutes back really adds up over time. You look at the volume of bid packages that people are going through it, AI is really, really good. Our AI specifically is really, really good at helping kinda assist that. So you may not necessarily bid more jobs, but you're able to not waste time on the jobs that you're not bidding. Right? So really really important time.
Sam Nadler [15:44]: You made a comment at the beginning, something about we're not here to replace estimators. And, we've pursued here at Ryz Labs some AI solutions that I think for the, like, ideal ICP or the ideal kind of customer, they're you know, initially, even though, like, we think it saves them tons of time and generates tons of efficiency, there was a little bit of, I guess, fear, trepidation, maybe even a little bit of resistance until they kinda see the product and see how it, like, really benefits them. Are you experiencing that at all with the estimators or the clients you're working with? Or are they like, oh my god. This is the greatest solution. Like, I'm you know, this is the best.
Tristan Wilson [16:27]: This is a great question. I mean, I think we see a broad spectrum of different reactions to when we say AI. I know I've, like, present it to large groups of PMs and project managers and estimators where there's a lot of fear in the room. And I think it's important to understand and define, like, what AI is. Right? How it's there, how it can be there to help us, and then also how it's just a tool and it doesn't necessarily replace the person who is doing the important work of figuring out what's it gonna cost us to go build this job. It just makes that person a heck of a lot better, a heck of a lot more efficient, and able to spend times on the stuff that matters. Like, hand entering a 120 items does not make you a better estimator. All it does, it basically makes you spend time on stuff that doesn't matter. And if you compare two companies and one company has the power of AI and they're using it, they're able to spend a lot more time on the stuff that really matters compared to somebody who isn't using it and is doing it, the way of manual entry.
Tristan Wilson [17:37]: But that doesn't mean that we don't still need to review every page, that we still don't need to review every single document, and that we're still not fully responsible for the job as the estimator. And knowing how to do it the old school way is important. It's just now how can we apply these new tools to help us make you know, to help make us more, more efficient and more accurate and also catch things that we might have missed? Like, I had a really cool thing the other day where somebody ran a bid, somebody put a bid package in, and they had actually gone through and reviewed this thing in great detail, highlighted stuff and everything. And the AI actually caught something in the bid where the owner gave them conflicting information. The owner said, hey. The project should be completed in this many days, but then when you look deep down on the contract, it actually conflicted with what the owner said in another part. And we're human. Right? There are things that we miss, and the AI can help us catch those things that we might have missed so that that doesn't become a dispute or a claim later on down the line, for the successful bidder.
Tristan Wilson [18:40]: Part of doing this job is figuring out if anything's wrong with the documents that you've been provided and letting the owner know, hey. There's a problem here. Right? And it's really hard when you're so busy. You'd you might miss things in AI. It doesn't get tired. It and it's one great rule of thumb that I make sure to tell to everybody that we work with is that assume this is the worst AI that you're ever gonna use. Right? Because it's just going to keep getting better. Right? And we can use it to our advantage. And I think estimators are awesome. Like, I am an estimator, so I would not also, like, I've this is kind of who I am. It's in my DNA. And I don't think that estimators should be there in the 10:00 the night before the bid trying to crank and get stuff done. Why not use this great technology that's available to us just to help us do our job a little bit better and a little bit faster?
Sam Nadler [19:35]: Amazing.
Jordan Metzner [19:36]: Awesome. Alright, Tristan. Well, yeah, I think that was awesome, perfect fit for Built this week. Just, how using AI to build new products and services that, were probably not possible without AI. In the last segment here, we're just gonna jump into some AI news. It's been another super duper hot week. I'll first jump into this news. I don't know if you guys saw, but, xAI is raising money and NVIDIA is investing, I think, 2,000,000,000 into the round. So we've got Jensen now investing in xAI, OpenAI, CoreWeave, the money's sloshing all over the place. But, yeah, just wanted to get your guys feedback here as well as just watching the market continue to rip, as these hyperscalers continue to grow.
Sam Nadler [20:29]: Yeah. I mean, I thought, my immediate take was, interesting from the deal with NVIDIA and OpenAI, and now seeing with this kind of circular financing with NVIDIA and xAI and all these other examples coming out, within the last two weeks. I guess I'm a little concerned. Is it, true infrastructure build out that makes sense, these relationships, or is it really this speculative bubble that, could lead to some major problems for the space in the near future or a little bit farther down the line. I think one or one of these examples makes a lot of sense potentially for NVIDIA. But, is it just a huge signal for speculative action?
Jordan Metzner [21:26]: I don't know. I don't think so. I don't Jensen says, like, he got the opportunity to invest, so they're investing. I don't know. Would you I would invest. Tristan, wouldn't you invest?
Tristan Wilson [21:36]: It's hard to say, Pan. I don't know. I think xAI's got an interesting approach. I like kind of what they're doing with, like, their philosophy of truth seeking. I think that's something that's unique that they do that I like that I don't think other people do. You know? There's a long line of people that have bet against Elon that are wrong. So, yeah, I don't know. I'm I'm not a I'm not an expert on, like, AI funding and that type of stuff, but that's those are just kinda my initial thoughts on that.
Jordan Metzner [22:11]: Okay. Cool. Alright. Well, let's jump into the next one. Just rolling this out. So, all of this adds up to just a massive investment from OpenAI. They got some money obviously from NVIDIA as we spoke about on our last episode and then today as well. And, yeah, they say that estimated spending by OpenAI is gonna be close to a trillion dollars. And I think as like, current revenues are, like, around 10 or $15,000,000,000 annually, which is obviously incredibly impressive for a start up, but nowhere near a trillion dollars. But it just shows at least their belief as well as kind of these hyperscalers belief in where AI is going and where the AI spend is going. I'd love to get your guys' feedback here, but I don't think we've ever seen an investment project of a trillion dollars before. This has gotta be like the biggest ever. And, I mean, what does this mean? Like, if AI is this prevalent right now, how early are we when, all of this trillion dollars is actually invested in these data centers and pipelines and all these other things software is built?
Sam Nadler [23:19]: Who knows? I mean, it's it is a mind boggling amount of money. And at the end of the day, I'd I'd like I said in my last comment, I think there it could be a signal for some concerns. But, taking the other side, I'm pretty much investing all of my free cash flow in NVIDIA. I And Yola. Yeah. I think I'll probably continue doing that even at $4,500,000,000,000 valuation.
Jordan Metzner [23:52]: Yeah. It's still early, I think. Tristan, what do you think?
Tristan Wilson [23:55]: I think AI is massively underhyped. That's my opinion, just as a whole. So, like, so, like, where do I you know, I do see more, like, OpenAI enterprise accounts. That's becoming very common in construction. I'd say probably, I don't know. If I had to you know, I'd say 15 to 20%, of the folks that we talk to are using, AI across the enterprise, of course, because they know that if they sign up for the, enterprise account that they can feel more secure with, like, people uploading sensitive information. At least that's kind of the basis that I've heard. And just people getting more familiar with using AI. There just aren't a lot of tools that are purpose built for our industry, in civil construction that are very specific to the industry that have AI. I mean, we're I think we're one of the few. There are some others. So I think, yeah, people they know AI is a big thing. They know it's coming. They read the headlines, and, they don't wanna miss out. I think that's the biggest concern, because like you said, our my industry is very traditional, and, we're not known for being the most for grabbing on to innovation, as soon as it's available.
Tristan Wilson [25:07]: There are, of course, just like anything else, like, there's the technology adoption life cycle. Right? Crossing the Chasm by Geoffrey Moore, where he talks about there's, like, the 13 and a half percent that are innovators and early adopters. There are those, in this industry, but the majority is in the you know, early majority, late majority, there's they wanna see products that are more proven before they jump in. But I think everybody that I talk to, like executives, they're all using ChatGPT, maybe Claude, Grok, those types of tools on a daily basis, whether that's for personal or for professional use.
Jordan Metzner [25:44]: Yeah. Makes sense. I mean, yeah, you can see when, more AI is implemented inside of construction and more chips, more OpenAI, more hyperscalers. I mean, that's where the trillion dollars is gonna pay itself off. Right?
Tristan Wilson [26:00]: I saw an article yesterday or maybe Dave Hoy yesterday about one of the management consulting firms. I think it was maybe I don't know if it was Ernst and Young or just, one of those management consulting firms. 470,000 people got access to with Anthropic Claude, the kind of an enterprise academy. And just imagine, like, the size and scale of a deal like that. You know? I mean, 470,000 people with, a license to an app. Just It's amazing how big some of those deals can get. Right? Hundreds of millions, right, per year.
Jordan Metzner [26:36]: Hey, guys. Overall, I think it was a great episode. It was really awesome, Tristan, just to hear your side of talking about an industry that, we don't think that much about since we're focused so much on technology. It's taking industry like construction and, roadside highway and how AI is, being adopted and being applied. And, it does definitely seem like the early days and just shows how exciting it is. And, hopefully, we'll we'll get AI implemented on the other side as well from those governments. And, they'll make, better contracts and better diagrams. And, it'll be AI talking to AI, and at the end of the day, trying to, make a process that that becomes smooth for everybody. So alright. Well, I'm Jordan. On behalf of Sam and Tristan and everybody else, this was an awesome episode of built this week. You can check us out at builtthisweek.com or on YouTube or Spotify or all your favorite podcast platforms. Built This Week Podcast. So thanks everyone. Thanks Tristan.
Jordan Metzner [27:42]: Thanks Sam. And hope you guys all have a great week. See you soon. Built this week
