From Earthwork to AI: Mark Schwab on Building CivilTakeoff.ai & Fixing Construction’s Estimating Bottleneck

Rishi Srivastava (00:41)
Today our guest is Mark Schwab. Mark, welcome.

Mark Schwab (00:45)
Thank you. Good to be here.

Rishi Srivastava (00:46)
Perception is about your story. Tell me a little bit about yourself. What’s the journey from cinematography to construction tech founder?

Mark Schwab (00:56)
Well, it’s a bit of a interesting chain of events. So I studied film and art in college, just a passion of mine, because I’m creative. And I got roped into a construction company. So you had to pay the bills. And I got married. And my boss at the time, he asked me if I wanted to interview. I interviewed.

Originally they didn’t want to hire me because they were like this guy knows nothing about construction. I was like, well, you’re the one that wanted to interview me. So I slowly grew through the company. They gave me more and more responsibility and I realized I have a passion for technology. And so they put me in charge of IT and HR. So I was training people how to do their jobs, which means I had to learn all the jobs. So I learned most of the jobs in the company.

Rishi Srivastava (01:25)
Yes.

Mark Schwab (01:44)
and then wrote the procedures on how to do the job. And then just basic kind of IT troubleshooting, a bit of a nerd. So it was just an easy avenue there. And now learning coding, jumping into creating websites. And yeah, it’s been about two and a half years working on Civil Takeoff and it’s been quite the journey.

Rishi Srivastava (02:06)
Yeah, you also work for a construction company,

Mark Schwab (02:09)
Yep, I still work for that same company. It’s been about seven years now, seven years and a couple months with that same company.

Rishi Srivastava (02:16)
Cool. You’ve worn a lot of hats, HR, payroll, AP, dispatch, building SOPs. Which of those experiences most shaped how you think about building software for contractors?

Mark Schwab (02:30)
That’s a good question. I think the biggest one is probably just the HR side. If you think about it, learning SOPs, putting together procedures, because I had to look at every single one of the positions at the company, I had to learn how to do their job. And through that process, I learned all these inefficiencies that were happening.

and me being creative from my background, I was thinking, we can fix this, why can’t we make this more efficient? And so I say that was the most influential is learning all the problems and then writing the procedure, even though I’m like, this is not efficient, there should be a software that makes this better. So that would be probably the most influential.

Rishi Srivastava (03:11)
You know, a lot of companies do not have written procedures on how various jobs are done. It’s someone’s head.

Mark Schwab (03:18)
Yeah,

especially in construction. When we talk to other contractors and they find out that we do have procedures, they’re like blown away. So they’re pretty detailed. have screenshots and arrows and all the things. I think getting people to read them sometimes can be a challenge because everything’s in their head. But new hires really appreciate having good procedures. Certainly.

Rishi Srivastava (03:38)
Yeah, yeah.

Next section here is on construction operations and finance. You built reports on payroll, equipment profitability, and vendor performance. Which financial metrics do you think contractors consistently get wrong or overlook?

Mark Schwab (03:55)
I think the number one thing they get wrong isn’t necessarily a specific report. I think it’s reporting in general. I think that they believe that they just need to see a report for one reason or another, but it’s the wrong reason for seeing a report. A report should drive you somewhere. It should be some sort of key performance indicator that you’re on the right track. So I think in our own company, we’ve made this mistake where

We’re wanting all this data, we’re super data heavy company. We’re putting in inputs all the way from the field to the office. We’re tracking everything. And so you could report on anything and everything. And the reality is there’s only a very specific amount of information you need to know as a company in order to know that you’re on track, if you’re going to be profitable. And that’s where people get wrong is they just plug in the data, but they don’t get a good report.

They’re not using the report. They’re not actually looking at it. It’s like a good example of like a good report is equipment. So in the grading industry, where I come from, you live and die by equipment. It’s the most expensive part of the business really. And your burn rate every single hour needs to be quantified. So you need to track how much gas you’re putting in or diesel.

mechanics hours, the parts to repair, how many hours did it work that year, what were the insurance certificates, blah, blah. Then you can run a report on the rental. How much did you rent that specific equipment? And based off how many hours you rented a machine, you might justify purchasing a machine, but you can’t make sure that your purchase is more expensive than if you were just to rent it. So it’s very important to

balance, like that sort of report is extremely helpful to determine what you do. Do I buy another piece of equipment or do I rent? And so you want to have good data and you want to be able to report on it, but it should have a purpose and action after the report is done. It shouldn’t just be for report sake, if that makes sense.

Rishi Srivastava (05:51)
Yeah, the end goal of a report needs to be clear, not just throwing bunch of numbers and hope that they will make sense.

Mark Schwab (05:58)
Yeah, I think they like they, believe that they look at it hard enough, it’s going to tell them something. And I imagine the first time it should be like that, but then you should think, okay, how can I make this data serve me? And then cut rid of the noise, get rid of the noise, all the numbers that don’t affect don’t influence because if things are interesting to look at, you’re like, ⁓ interesting. But if it doesn’t drive anything, there’s no reason to track it. There’s no reason to report against it. But

Rishi Srivastava (06:02)
Mm.

Amen.

Mark Schwab (06:24)
That is something interesting that AI can do is you can just plug in an AI and say, hey, what are good things for me to take away from this report? What can I use going forward as a key performance indicator that I can report against?

Rishi Srivastava (06:35)
Yeah, actually, in fact, I used AI to help me get to the bottom of a medical report recently. My sister got some medical issues right now, and I was not able to understand the report very well. So I gave it to the AI and asked him, give me summary of this report.

Mark Schwab (06:53)
Yeah.

Rishi Srivastava (06:53)
From your time at Elite Earthworks, what are some real-world workflow pain points that finance teams feel that outsiders typically miss?

Mark Schwab (07:03)
I think, well, if they’re coming from outside of the construction industry, I think there’s a confusion with, you you get an invoice and you think, a project manager can look at it and you can just say, pay, and it goes to this code. But if you’re a big construction company and you’re making a lot of purchases, you have to tie that to the field. And so I think that’s something they get wrong is there needs to be an audit paper trail from the foreman all the way up to the office.

And then all those steps need to be in agreement for you to pay the invoice. Now, certain contractors are a little bit lazier or less paperwork and they do have a project manager that just signs off on it and they kind of blindly just let it go. But I think a good contractor is the foreman is responsible for ensuring that the activity, the material, the equipment, the hours, they’re all reported accurately digitally or I guess paper, but we’re digital.

And when the invoice comes in, it can be tied to that data and they should be in agreement. I think that’s what people get wrong is there’s a lot of many steps of everybody has to do their job right in order for it to work. So it’s complex, very complex.

Rishi Srivastava (08:09)
Yeah, definitely. How did working inside a grading and Earthworks company shape your understanding of job costing, estimating, and cash flow?

Mark Schwab (08:18)
Well, I mean, they taught me everything I know as it pertains to accounting. So, I mean, they taught me how to read a budget, how to look at a cash flow, debits and credits, GL counts. So it definitely shaped everything I know. And because I don’t have a history before, there’s probably, I’m just an empty vessel. ⁓ But yeah, I mean, with our industry specifically,

Rishi Srivastava (08:37)
Running.

Mark Schwab (08:42)
Job costing is just so incredibly important because cash flow can be tough. We’re paid 90 days later sometimes and our retention is held until the end of the job. And since we’re the first on the job as graders and we’re also the last people on the job, our retention is held the longest out of almost any scope. So we have to be extremely careful with our job costs in order to make sure we’re not going down below budget.

Rishi Srivastava (09:08)
Yeah, sometimes the retention amount, it’s more than the profit you’re to make on that job.

Mark Schwab (09:12)
Yeah, yeah. And so you really don’t hate you see your profit until that retention comes through sometimes. Which is pretty rough. So I’m trying to get into software faster, faster turnaround times for your your strike payment to go through.

Rishi Srivastava (09:17)
Mm-hmm. Ha ha ha ha. I unitoo.

Yeah, yeah.

AI and estimating and takeoff. What convinced you that there was a big enough gap in the market to start CivilTakeoff.ai?

Mark Schwab (09:35)
I think when we looked around as grading contractors and saw there wasn’t a solution for us, I mean, that was the obvious sign. just, we talked a little bit earlier before we started about, just start working and you think, AI, could AI do this? It should be able to do it. Okay. Yeah. We can efficient, make this super efficient. So why don’t we make it? And, you know, takeoffs are such a time labor intensive product.

It just takes man hours. There isn’t like a tool that really makes you faster, like clicking faster. It just takes a body tracing around contours and shapes of the plans. And since that’s all visual, we are thinking there has to be an opportunity here for us to slide in now that AI has gotten good enough to create a vision model that can do this for us. So like just there as like feasibility that

we saw an avenue there and we looked around and no one was doing it. There’s a few competitors have entered the space as far as earthwork is concerned. There’s still unique in that. But we saw that a few AI takeoff softwares are doing quite well in different scopes of industry that we’re not planning on approaching. And so it’s kind of like ours for the taking until someone else comes in, I guess. But yeah, so there’s a big enough market.

And really when you look at other takeoff software out there, there’s a lot of different options and they all do quite well. And I think a lot of it comes down to preference of the tool layout and how it interacts with your system. So yeah, we see a little bit of a market there.

Rishi Srivastava (11:06)
Yeah, and

definitely. So takeoff in, let’s say, grading, is it different compared to takeoff in other specialty industries?

Mark Schwab (11:15)
Yeah, so grading takeoffs, they’re going to be volumetric. And so it’s almost like you’re building a 3D model of the job site. Whereas any other takeoff is typically going to be like a two dimensional plane, but you can apply depth to it and it’ll calculate cubic yard. for the most part, it’s just planes. And right now we only do planes. We’re working on the earthwork side. So we’re inching our way towards a solution that’ll be just geared for earthwork.

while still servicing those that want to take advantage of the service.

Rishi Srivastava (11:45)
Why is estimating one of the most manually painful parts of construction. And how do you see AI changing that?

Mark Schwab (11:54)
Well, I think I talked about it a little bit where it really is a person identifying features on a plan, either paper or digital, and they’re just drawing around the contours of the curve or the building pad. And all they’re doing is tracing. And so it’s just a bunch of manual effort. You have sometimes plan sets that can be 100 pages long, and each one of those pages takes 10, 15 minutes.

So you can imagine if you’re trying to a thousand projects in a year, a hundred projects in a year, depending on your size, that’s just a ton of man hours. So that alone is super laborious. And the problem as someone in construction is a lot of the times it’s how many bids can you get out? So if you can capture 10 % of the bids that you send out and you send out a hundred bids,

you won 10 jobs that year. But if you want to win 20 jobs next year, that means you need to do 200 bids. Well, an estimator probably can actually just do another 200 bids. It’s for them. That’s not the time intensive part. The time intensive part is that takeoff technician. And so to do another 100 takeoffs, you need to now bring another person body in, you need to train them up, and they’re working in your company full time. So it’s limiting because it’s just a

a time consuming task. And I think it’s, also prohibitive for that person to be able to grow throughout the company too. They can’t become an estimator typically, because all they’re doing is tracing plans all day. So they’re not engaging with clients. They’re not learning the trade. They’re just tracing. Yeah, that makes sense.

Rishi Srivastava (13:27)
Yeah, definitely. So manual, you know, take off and you needed more people to be able to double your volume. Yeah, makes sense.

Mark Schwab (13:34)
It’s

not really scalable by itself right now. Yeah.

Rishi Srivastava (13:39)
What’s the biggest misconception contractors have when they hear AI for takeoff and estimating?

Mark Schwab (13:45)
Well, right now where AI stands, and I think this is kind of the AI bubble that’s been popping, is there’s been a lot of over-promising and under-delivering when it comes to AI models. And so I think contractors, like to see, they like to touch, very tangible people. And so when they get in there and there’s a lot of big promises about AI being able to just make their workload 98 % faster or something.

And the reality is they can’t see any tangible increase or efficiencies. I think they just distrust it in a lot of ways. But at the same time, I think they’re tentatively like excited about the next thing. I can see the excitement when I talk to people there. The buzzword of AI, it’s immediately deflating, but they’re also like, does it work though? I kind of hope it works. So there’s a dichotomy of, I want it to work, but I also don’t trust it.

Rishi Srivastava (14:33)
Mm-hmm.

Mark Schwab (14:37)
So I think right now there’s ⁓ just a misconception of just accuracy and what it’s able to do, which there’s some truth to that. We’re still in progress. I mean, even our own company, we haven’t rolled it out because we don’t want that to be the case. We don’t want it to get it 60 % accurate and for them to go, what’s the deal, man? So we’re waiting until it’s…

a strong enough model that we can release and then we’ll release it when we feel like we’ll blow away people’s socks but until then we’re just keeping it internally because we don’t want it, we don’t want that to be our first opinion for people.

Rishi Srivastava (15:10)
Yeah. So are you actually training these AI models for estimating and takeoff, or it’s just something out of the box from somewhere?

Mark Schwab (15:18)
Yeah, we’re training it. We’re doing it. We have some developers overseas that are working on it. So we’ve been in contract with them for two and a half years. So I make the website, the tool, the backend, the frontend, and then they specialize in just training models, custom models for us. So we’ve been just feeding them additional data as we make them. And then they’ve been working on it.

Rishi Srivastava (15:40)
So overseas developers, are they good quality? And you think they’re delivering good work for you?

Mark Schwab (15:46)
I think so. I think there’s, yeah, it’s tough to say sometimes that, you know, they just say, Hey, we need more data in order for it to be a more efficient model. And I think someone that’s not necessarily technically skilled and training models, ⁓ besides just generally understanding how it works, you can’t really fact check them. You’re like, well, I guess, I guess we need to give you more models or data sets. So it’s tough to say sometimes.

Rishi Srivastava (16:04)
Yeah.

Yeah, because AI is still very cutting edge, right? Not a lot of developers in the world can do the AI model training.

Mark Schwab (16:19)
Yeah,

yeah, we definitely spent a long time interviewing and making sure it was the right fit. That was one of our more laborious tasks.

Rishi Srivastava (16:29)
Yeah, that makes sense. You lead both product and technical decision making. What’s the hardest part about building an AI based estimating platform from scratch?

Mark Schwab (16:40)
I think the hardest part is you just don’t know what to go after. There’s so many different features and you have a lot of ideas. So you’re like, Oh, it’d be really nice if this thing existed. And then there’s this thing that you wish would exist. And so you start to switch what you go after. Like one thing we want to make in the next year is a PDF viewer. a, like you download it as an application and it kind of would be like an Adobe reader, but just has editing features.

And we want to release that for free for the world. A for some authenticity for our, our software people see our brand and recognize it a little bit more, but also it helps us compete with other brands that have PDF viewers built into their software. so it’s, it’s like, do we go after that or do we switch to go after, another development item that we, think would bring a lot of value to our customers. so I think.

As you’re creating something from scratch, it’s like, which one do we go after? kind of get the, uh, there’s a bunch of rabbits in the field and you’re a dog. Which one do you go after sort of thing? So that, that, that, just, you’re always learning every day, there’s something new and you think, you know, how, how to get a customer. And then like we were talking earlier, you find out that the best way to get a customer is this way. And, uh, this is more important to a paving contractor than.

Rishi Srivastava (17:41)
Yeah.

Mark Schwab (17:57)
This is more important to a land developer. You start to learn the nuances of customers and why they want something. Yeah.

Rishi Srivastava (18:04)
Yeah, as a founder,

it’s so important to be able to understand how people get convinced and what we need to build for them so they start using our mind.

Mark Schwab (18:15)
Yep.

Rishi Srivastava (18:16)
The next section here, is on contractor mindset. You mentioned technology built by contractors for contractors. What makes contractor-built software fundamentally different?

Mark Schwab (18:29)
Well, think contractor built software that’s by a contractor. We understand that being able to get from point A to point B is incredibly important. I think there’s a lot of like steps that someone that doesn’t come from the contracting world would add to a process that slow you down. as a contractor, you’d go, I just want to get to B. I don’t want to have to go to B, C, D, E to get to F.

⁓ and so I think just efficiency workflows, big buttons, just understanding like the UI that you want it be really clear messaging. like right now the user experience for Apple is kind of like hide text, make Apple icons, dynamic where you hover and then things expand and,

While I love that, I think the typical contractor is going to say, I want it to say whatever it’s going to do so I can go click that thing. So just getting in the mindset of a contractor, how do I want the software to look? How do I want it to function? And also I want it to have features that actually address my business needs, not like a buzzword. I think right now one of the buzzwords

actually I think you were mentioning it about voice to text is for Procore they’ll say like, just speak to the software. Like they’ll say something like you don’t ever have to type. Like it’s one of like the buzzwords for contractors. But in the reality, it’s just a microphone button that’s built into your keyboard. So it’s no special. It’s like every app has that. why are you using like, so I think contractors, we see those buzzwords and we go.

Rishi Srivastava (19:53)
Mm-hmm.

Mark Schwab (19:58)
Is this real? Do you mean it? there’s not like you keep kind of doing like a fluff hoodwink thing situation. Like, I think one of the biggest buzzwords is synch or something, our program synchs to this program. And what ends up happening is it’s a CSV export import. And they hide it and it’s, it’s just you go through the process as a contract, you’re like, yes, I want them to talk, I wanted to synch

Rishi Srivastava (20:08)
Thank

Mark Schwab (20:22)
But there’s actually not an API. It’s just a sync or sorry, there’s a, you have to download a CSV, you have to upload it, you have to select the headers, and then you have to import it, then you have to validate it. And you’re like, that, that’s not what I was promised. So I think like as a contractor, you recognize what’s important and the messaging stays consistent. Well, we’re not using buzzwords that we don’t mean. We’re using true statements that’ll actually follow through, if that makes sense.

It’s a very long answer.

Rishi Srivastava (20:50)
Yeah, definitely. You know,

on the syncing part, the older technologies in our space, lot of times reading from them is easier. But it’s very difficult to write to them because they have their validation rules on all these fields. as a separate software, you cannot write on these older software, say, Procore or even.

Mark Schwab (21:03)
Mm-hmm.

Rishi Srivastava (21:13)
⁓ Foundation or other site you have to use their import tool where they can validate what you’re writing it. So.

Mark Schwab (21:19)
Yeah,

yeah, no, that’s very, that’s very true. Like in a lot of these legacy software that hopefully will have some replacements in next 10 years. ⁓ They some people have developed a middleware software that handles it. And I think when they have a middleware software that does handle it or an AI bot that goes in and it completes the final steps, I think you’re allowed to say, you know, we have a synch if that makes sense.

Rishi Srivastava (21:29)
Yeah.

Mark Schwab (21:47)
⁓ I think you can say sync also for read. Like if it reads through an API, like I think you guys do it. I would say that that’s a sync. I think it’s when they claim they have a two-way sync and I’m like, no, that’s not necessarily true. It has a one-way sync and then there’s a push. But just being clear with the messaging as a contractor so that way I don’t scare any other contractors away, if that makes sense.

Rishi Srivastava (21:47)
Yeah.

Yeah.

Yeah, and some of the times the software companies, they’re trying to get somebody to use their software so they start lying too, right?

Mark Schwab (22:19)
Mm-hmm. Yeah.

Yeah.

Rishi Srivastava (22:21)
How do you balance the desire to innovate with the practical reality that construction teams often don’t want to change their workflow?

Mark Schwab (22:29)
I bet you run into this a lot.

I think you just have to convince them that it’s faster. And maybe that’s you hopping on a demo call and walking hand by hand through doing a takeoff. And they could do it their way and you could do it your way and you can show them whose way is faster. I think there’s going to be a generation that leaves. There’s just a high level of just more tenured staff that is

about set to retire with the baby boom generation. And so I think we’ll see a little bit more adoption with technology and tool changes. I’m just hoping that the wisdom continues from that generation to the next generation so they can continue to bid and be an estimator. But that one’s tough. I think that’ll always be an uphill battle in construction. We’re set in our ways. We want to look a certain way, but we want a different outcome.

And so you have to plead to their, you have to show them and convince them that it is better, it is faster, and this is a better tool to use. And a lot of that is just answering objections that they might have. So asking, what are you currently using? What do you don’t like about it? What do you don’t like about it? And just convince and answering all their questions and convincing them otherwise.

Yeah, so I think that would probably be my answer there. I’m sure you have you run into it a lot with accounting workflows is very precise workflows of how I mean, I know I have many questions for you.

Rishi Srivastava (23:43)
Yeah.

Yeah,

I think change is tough. Actually, I had another episode just on change management with a change management expert in construction. How do we ensure that people are willing and understanding to change? Because the way we’ve evolved so much and we’ve such a prosperous society, it is because of changes, technological changes mainly.

Mark Schwab (24:16)
Yeah. Yeah.

Rishi Srivastava (24:17)
What

early customer feedback surprised you the most while building CivilTakeoff.ai?

Mark Schwab (24:23)
The early feedback was interesting. realized that, I mean, we talked about this also earlier, is just that people can get stuck pretty easy on the onboarding process, especially in construction. And so identifying the different steps that they do get stuck, whether that’s just not understanding the next step of this is where you click the create project button, or this is how you use this tool. So that early feedback was really surprising to us just because we’re a little bit more tech savvy, even though we’re contractors.

And so for us, it just felt right. You you click here, you go here, you do this tool. And for a lot of users, they get stuck on how do I move it? How do I zoom it? How do I switch pages? How do I get to that tool? Which was really educational for us because we’ve been adding a lot of material to help with those. And we still are working on additional like guided click here, use this tool sort of training.

And I think that’ll help, know, with user retention and you want your user to be as intelligent as possible in your tool. You want them to feel like an expert. You don’t want to have them feel like, oh man, I had to schedule a call with so and so to understand how to use this tool. I want them to feel like by the end of the workflow or the work onboarding tour, I’m an expert and I can do a takeoff on this and I can be faster than I am on plan, Swift or blue beam or all these other software competitors that we have.

So I’d say that was pretty interesting for us to find and figure out.

Rishi Srivastava (25:44)
Yeah, intuitive user experience is so tough to design.

Mark Schwab (25:48)
Mm-hmm.

Rishi Srivastava (25:48)
Everything seems like natural to your users. Seems flawless when it’s in the hand of the customer, but when you’re designing as a software builder, that’s so hard, you know?

Mark Schwab (25:58)
Yeah, yeah, definitely. Yeah, you have an idea of how it should look in your head. And then someone comes in and goes, this is laid out just completely wrong. And you’re like, tell me more. What ideas do you have? What did I miss? I’d love to hear. But you know, the nice thing about being in early software with early users is they have a big say in how it looks that our company

Rishi Srivastava (26:08)
Mm-hmm.

Mm-hmm.

Mark Schwab (26:22)
We have strong beliefs it’s going to be a massive company with great success, we pray. And we think that these users have a big influence in how it’s going to look. So if they want to affect and change the UI and the UX, then they have great say in that with their feedback.

Rishi Srivastava (26:38)
Yeah, know the technology adoption curves, the early users, they are innovators. They are looking for something new and unique that you’re building. They’re not the common people from the street, your early users.

Mark Schwab (26:45)
Mm-hmm.

Rishi Srivastava (26:52)
Because most of the industry doesn’t want to change. You and I know that, right? They’re like, this is so hard. Let’s just keep doing the way we are doing things.

Mark Schwab (27:00)
Yeah, definitely.

If it’s not broken, don’t fix it. The amount of times I hear that is often.

Rishi Srivastava (27:05)
Yeah, yeah. The last section here, is on startup leadership and growth. You are CEO, product owner, dev team leader, and customer success. How do you prioritize what gets built next?

Mark Schwab (27:19)
Well, me and my ⁓ co-founder, AJ, we sit down and we plan out the roadmap for the next month. So we get all the feedback that our users have submitted to us. So, hey, I want the button bigger or this smaller. I want this feature. When you hover your mouse over a polygon, I want it to call out a square footage, something like that. And we put them all down on a board and we just select what we think is going to be the most beneficial. And we weigh that against

are bigger features that we have been planning out because we have bigger roadmap plan features that we have. But usually it’s month to month that we plan. And the biggest thing is bug fixes is number one. You and I both know that. So you got to get that going as quick as possible and roll those out. And the second one is going to be, if there’s any like critical oversight, that would make way more sense and would fix the onboarding process for you user. Let me do that.

because I’m fixing their problem, but it also fixes future people’s problems as well. So those are their first two priority things. And then the third one is just going to be what’s going to provide the most value to our customers? What’s an additional feature we can do that’ll just increase their love of the software and make its usefulness even greater? So usually those are the things we balance in our mind.

Rishi Srivastava (28:33)
Yes, strategic. I love it. What are some scrappy outside the box space you’ve grown civil takeoff.ai so far.

Mark Schwab (28:42)
Well, we started out as a service, the done for you service is what we called it. And we thought it would be a big smash because we hire out a lot of takeoffs to be done by other companies just because we get overwhelmed and we do want to bid more projects. And so our takeoff technician can’t handle all the projects we want to bid. So we put it out for someone else to do and we pay them to do it. So we thought, what if we spin up our own service?

where users upload their plans and they select by page and they pay by page. And we got no feedback. We really, we got people to join the platform, not a single person uploaded and submitted a plan. So we, we definitely, we swung and we missed there. So we thought let’s, let’s pivot, let’s be scrappy and let’s use the tool that we’re using to train the AI models and release that. Cause it’s actually a really good tool. It’s, it’s got a lot of useful features. It

It does everything the big box guys do. And there’s even cloud collaboration features and backups and it’s browser based. There’s some things that other people are not doing that we have. So we pivoted that direction and that was it. That was what people were looking for. So, and June was when we did the done for you. And in August, we released the tool. And since then it’s been quite a bit different for user growth and retention. So that was the, the move is to pivot away from.

a service and ⁓ move towards just a tool that people can use.

Rishi Srivastava (30:04)
Yeah, taking market feedback and adjusting. That’s so valuable.

The last question here, is on what advice would you give to construction professionals who want to start a tech company but don’t have a software background?

Mark Schwab (30:18)
So this is the time to do it. mean, you’re in the gold mine rush era of technology, I think. The amount of opportunity out there is just, it’s staggering. It’s kind of like, it’s almost overwhelming, I’m sure you feel it, where you think, we should do this thing, because this would solve a huge problem and you could be billionaires, you know, because you can make this thing, it’ll be easy, we’ll just really quickly make this thing. So it’s easy to have that.

like just look around and you’re like, I wish I could duplicate myself. I just need more time. All it is is time. just, I wish I had more time to create these products that I believe they need to be made to make people’s lives better and win more work or whatever it might be. so my advice would be just use AI. I know that everyone says that, but open up Gemini, open up chat, GPT whatever you’re using.

and start asking it questions and ask really good questions like

something that’s related to what you’re doing. think I talked about this earlier. If you just watch someone work, you’re going to see an inefficiency in their work. You’re like for you, like if you were to look at my current company I work for and you’re like, wait, you guys spend how much time going and looking at this program to see if this invoice is correct and it matches your PO and you have to, then you have to manually put all these line items and then you have to attach everything and then you have to write it for approval and.

you would look at that and go, well, I should just automate that and save that person a ton of time. So you do that same lens with anything you do in life, whether it’s church, whether it’s your family, whether it’s like, I’m getting a text on like gifts for Christmas and I don’t know how to send people the right gift for Christmas. why don’t you make an app for that? I think in that, if you put on that lens and you look through life,

you’ll be able to see a lot of opportunities of software that you can make. And then using AI, you go, Hey, I want to make a program that does XYZ. How do I do it? And it’ll be like, Oh, well, you need to learn how to use JavaScript. You need to learn how to do blah, blah, blah. And then you go, okay, how do I learn JavaScript? And then it’ll start teaching it. It’s the, it’s the perfect professor of the world where it can go to every detail and extent and it makes mistakes. It does, but it

for the most part, it’s still the, and humans make mistakes, it’s still the best professor for teaching you and getting you into things. the time is ripe for opportunity right now.

Rishi Srivastava (32:42)
In last three years, these language models have completely changed the way we access knowledge. Like, GPT Gemini, they’re just so good. Even though they hallucinate, like you were saying, they answer most of the time is very good.

Mark Schwab (32:56)
Yeah, I think the answers that they are hallucinating, they’re usually really obvious. At least in my experience. I mean, I guess there’s probably things that it’s doing wrong, but it’s easy to figure out where it went wrong and correct it. You just have to turn on your brain. If you turn your brain completely off, AI is not for you. You’re not going to achieve the result you want. It still takes you being a human and thinking about the answer it gave you and thinking about how it got to that answer. Like you should be reading through

Rishi Srivastava (33:01)
Mm-hmm.

Mm-hmm.

Mark Schwab (33:23)
how it arrived at the response. And maybe that sounds time consuming or exhausting, but I mean, that’s how you learn, that’s how you catch things, that’s how you can take it to a new level. Outside of that, it’s just leading you and you’re not doing anything. So, yeah.

Rishi Srivastava (33:37)
Mark, thank you so much for coming on to the show.

Mark Schwab (33:40)
Thank you, thanks for having me. Happy Thanksgiving, it was pleasure being with you.

Rishi Srivastava (33:44)
Pleasure here too.