Keeping up with work that has been done on the construction jobsite is a never ending struggle. Documenting work that has been completed takes time and has in the past been primarily manual. StructionSite have pioneered the use of scans and photos to automate site documentation, and in doing so have created an entirely new class of construction software. Matt Daly, CEO of StructionSite, shared their product, vision and some great perspective on where the jobsite of the future is going.
Matt Daly
[00:00:00] Hugh Seaton: Welcome to Constructed Futures. I'm Hugh Seaton. Today I'm here with Matt Daly co-founder and CEO of StructionSite. Matt, welcome to the podcast.
[00:00:11] Matt Daly: Good to be here.
[00:00:12] Hugh Seaton: So let's start, like I like to by telling everybody what StructionSite does.
[00:00:18] Matt Daly: Cool. Yeah. Well, what are we doing over here at StructionSite. We are building a new category of software for construction teams. It's intelligent project tracking. Basically, you're taking some form of reality capture data, whether that be images or scans and using machine vision to quantify the progress that we see on the job site, against the plan.
[00:00:41] Hugh Seaton: That's awesome. And you guys have been at it for a while, yeah?
[00:00:46] Matt Daly: Yeah. We've been at it for awhile. I actually met my co-founder Philip back on a job site, the Kaiser Oakland hospital project here in the bay area, actually in 2013. He was working as a VDC/project engineer over at McCarthy.
He was the guy walking the job site with a set of paper plans in a highlighter mark and down MEP and framing installs. At that time, I was working at a company called Faro. You familiar with Faro?
[00:01:13] Hugh Seaton: Yeah, but let's talk about what Faro is a scanning... primarily a scanning company, correct?
[00:01:19] Matt Daly: Yeah. They make a wide variety of 3d imaging measurement, hardware and software primarily prior to 2010 for like aerospace and automotive manufacturers. But then in 2010 launched this LIDAR scanner, that scans buildings, which is part of why I found myself on a job site back in 2013 with Phillip.
[00:01:38] Hugh Seaton: So the two, well, two of the co-founders one was scanning the world. The other was working with models of buildings, is that correct?.
[00:01:46] Matt Daly: Yeah. Yeah, that's accurate.
[00:01:48] Hugh Seaton: So is that kind of the genesis, as you said, I've got peanut butter. He said, I've got chocolate.
[00:01:55] Matt Daly: I thought you were going to say jelly, but yes, peanut butter and chocolate.
That, all that absolutely where he...
[00:01:59] Hugh Seaton: Reeces, man, two great tastes that taste great together anyway.
But is that kind of the genesis of it is that you looked at what you were both doing and had a beer and said, you know, we could do something cool.
[00:02:10] Matt Daly: Yeah, we had a few, and in the meantime, sort of, it didn't just happen overnight, it was really multiple ideas that Phillip had around how we can use reality capture data, to just run a project more effectively and eliminate some of this stuff. Like, for example, him walking around with a highlighter and a set of plans, marking stuff down.
[00:02:35] Hugh Seaton: What an interesting image, right. It's funny, I worked in a factory very early in my career and did something analogous where I had to go to every station and hand data or hand paper off and check some things.
So that was sort of the lived experience that Phillip had where he's walking around and making physical marks on physical paper.
[00:02:55] Matt Daly: Yeah, exactly. You know, one of many, I would say, I mean, obviously there's a lot of different challenges in construction. He was also the guy that was managing their BIM models. Basically all this stuff coming in from the subcontractors doing coordination.
But, as far as like manual paper processes, yeah. That was one of the ways, really the best way, they had to try to track the different trades as they were making progress on that job, to make sure that things were staying on the plan.
[00:03:22] Hugh Seaton: I think some of that is because it's nothing beats a human looking at a thing, or at least that was the state-of-the-art at the time, right? Is someone had to go there, stand there, look at the context, look at the thing and say, okay it's done.
[00:03:36] Matt Daly: That's absolutely right. And now we would say definitely computers can beat a human staring at a thing. And there's some things that, uh, that are still very much in the realm of humans and will be for a while, more related to quality.
But as we talk about, is it done, is it not? Is it there, is it not? The more sort of binary view of how we look at progress? Computers are very rapidly, if not already surpassing what a human could do there. Which I think is totally fine for most people who've ever had to do that work.
[00:04:09] Hugh Seaton: Well that's right. I mean, it sounds like a good idea in intuitively and anecdotally you might say, well, obviously humans going to be better because they can do this and that. Say “fantastic have them do it 30 times.”
[00:04:20] Matt Daly: Yeah. I had a customer say something other day along the lines of, I need to hire builders who are great at building and then they also have to be good at tracking.
And these are really not the same skillset, if you're a great builder, that doesn't mean you're great at tracking stuff. Tracking is almost like a form of accounting, right? It's not a builder. So it's kind of like the message is, hire the best builders you can find. And let's offload some of this tracking work to the technology.
[00:04:48] Hugh Seaton: I think that's a fantastic point, right? Cause we're constantly hearing about labor shortages and the need to train people in skills and so on and so forth. And if you can take away some of the drudgery and the inevitable mistakes that happen when again, you know, if you do anything 30 or a hundred times, or even 15 times in a row performance really kind of falls off, right? And it doesn't with computer vision.
[00:05:12] Matt Daly: Yeah, exactly. The computer doesn't get tired. You can throw as much data at it as it, as you want. And it's not going to get tired. It's not going to decide that it doesn't enjoy this work or not. It's really removing an aspect of the construction management process that you're just not going to find anybody's sad about.
[00:05:32] Hugh Seaton: Yeah. And do you find that, that as you talk to people and they've used it for a little while, that there is, I mean, I'd be surprised if anybody's done numbers on it yet, but is there a worker retention impact on this as well that people enjoy the job or at least they complain about the drudgery a little bit less.
[00:05:49] Matt Daly: Yeah, it's absolutely anecdotal at this point where we're just, you hear the stories of folks who have been subject to this type of work. In fact, there's a project engineer that I'm thinking about in Southern California that works at a company I won't name, but it's still working at that company happily now, but, but early on in his career as a builder, You know, what happens is you come in as a younger person, don't have a ton of experience. So what do they do? They give you this type of work.
And this particular individual almost left the industry because he was so disillusioned by this type of work that he was doing. And it just, it wasn't exactly what he had in mind. And I know that there's like another side to the story here.
You're more experienced builders will look at that and say like, but that's how you learn, you know, that's how you learn the job. That's how you learn the ins and outs. That's studying the documentation. I think the truth is somewhere in between where it's important to learn the basics and go through that process as a young builder to learn how building happens.
And then there's a certain time where it's like, okay, well now let's offload this and not make this someone's full-time job.
[00:06:53] Hugh Seaton: Well, there's also having spent some time in learning and development, I can say there's more than one way to learn something and paying your dues the old way often has value, but isn't automatic and there's more than one way to teach somebody how things can go wrong and how processes work.
So if you were to put your finger on the problem that you solve in the market, let's talk a little bit about that.
[00:07:15] Matt Daly: Yeah, the problems that we solve.. I think there's a few: there's the business need and then there's the human need. And we've touched a little bit on the human need, which is simply to remove this type of work that isn't really adding a ton of value.
It's very important, the data itself is super important, so there's that side of it. And then there's the business need and I think it depends on whose lens you're looking through, right? When you talk about tracking progress on a construction project, that means first of all, that data is extremely important in order to have a clear view of how things are progressing, but it means different things to different aspects or different parts of the market.
If you're a GC, not having a clear picture of project leads to inconsistent flow of trades to the job, which leads to a lot of schedule uncertainty. And if you're a GC or an owner, nothing is more frustrating than schedule uncertainty, because there's an actual cost, the business cost to not having that building an operation.
For a trade partner, it means something a little bit different, right? Not having a clear view of productivity. It means project teams might have a false sense of security about how the job is going. It might mean they think they're 50% complete, but they're only 40% complete. And suddenly at the end of the job, they're going to blow through their budgeted labor hours and they've still got three or four more weeks worth of work to do.
[00:08:37] Hugh Seaton: And how, as you're implementing this, do you find that there's a process of getting used to having this new kind of data that StructionSite provides?
[00:08:50] Matt Daly: Yes and no. Some of the best companies we run into have a rigorous process around the way that they track the work. In fact, you can almost draw a straight line between companies who have a really robust tracking process and the ones who are larger and more successful than others. Obviously there's a lot of gray area in between there. So some that have a process where it's really about us fitting into that process and making sure the data and the formats are correct.
I'd say the overwhelming majority of builders fall into a different bucket. They're just now building out their strategy around data and how they use it. And they're getting this information maybe in a spreadsheet today. And the way that it gets to them is probably pretty manual. And then it might die where it lands when it finally informs something.
So I think there's a huge spectrum of customers where we're fitting into an existing process, and this is nothing new. And others who we're kind of helping them build the strategy around this data where they may not have had it before.
[00:09:50] Hugh Seaton: That's great timing too, because you're seeing across the industry just it's bubbling up everywhere, this idea of data and data strategies and understanding how to use data and integrating it into how you make decisions and think... I mean, the MCAA had their conference and that was like the main theme.
And Procore, obviously for obvious reasons has been pushing it for a while and so on. So are you finding that there's a bit of a wave of respect and understanding that, "okay. We do have to understand data and utilize data and move away from paper processes."
[00:10:23] Matt Daly: Yeah, there again here. There's a spectrum of folks who are on the very far end of that on one side where they have already put together a data strategy, they have a data lake, they have a warehouse, they've thought about the different processes that are not currently digital.
And so that's kind of one end of the spectrum. And then you have this other end of spectrum. They're there they're very, very early on in their digital evolution, if you will. And there's just a long way to go there. So I would say there's a very small percentage of companies we see today have a great strategy around data, have actually begun their process of aggregating everything, putting it into a lake or some sort of repository.
I think we have a very, very long way to go there and it begins, I mean, when you think about what we do, it's one of the first digital data sets for the job site. Now we're at a point in the industry where we've got digital paper, we have digital spreadsheets.
There's never really been a digital version of the physical jobsite. And so you can see where we're just now creating that digital asset that maybe can then inform other things down the road.
[00:11:32] Hugh Seaton: That's really exciting. And are you finding that, that you produce not just replacing what, you know, a guy walking around might've produced, but actually, you know, more data.
So instead of doing it once a day or once, you know, once every other day or so that you're actually able to produce a higher kind of density of data.
[00:11:52] Matt Daly: Yeah, I think we're about to see an explosion in that. I think there's something really exciting happening in exactly that area right now that, I'll plug one of our friends SmartVid.io (now Newmetrix) here for a minute, cause we're finishing our integration with them and, and like, it's a great example of how you can get more from a little bit of effort, right? Like one of these sort of core tenants of lean is reducing waste, but also getting more value from the activity you're already doing.
And, so right now, if a customer captures data with something like structuring site, they might get a series of photos. As we launch more tracking tools, they get data around on that. If you integrate this information with something like a SmartVid.io, for example, that same exact effort you put into capture the job site can also feed your safety analytics.
And so what you're seeing is...and what you're about to see, I think even more of, is when companies do have a strategy around this, that you can repurpose a lot of what you already have and what your field teams and are producing in the field to do multiple things from safety to quality, to progress and in so-and-so.
[00:12:59] Hugh Seaton: That's really exciting and yeah, you're right. That's one of the points of data that takes a minute to really understand, right...Is it's a freestanding thing that you can make a million copies of and it doesn't lose value, but also it often can be grouped in different ways to do different things with it.
So apart from SmartVid.io what are some other integrations that you guys have.
[00:13:24] Matt Daly: Yeah, we've done quite a few. I mean, the, probably the most notable ones are you have a Procore, you have Autodesk. We have others with Egnyte, Box, any number of cloud storage providers that we've done.
I'm sure I'm forgetting like certain ones off...
[00:13:42] Hugh Seaton: That's ok it's not the academy awards. It's all right.
[00:13:44] Matt Daly: Yeah, I'm sure there's others, but those are some of the major ones and it usually just comes down to... Plexus is another great one. Like where do customers want their data? And, that tends to be their ERP or their project management software.
[00:14:01] Hugh Seaton: And is that the integrations are you creating a new kind of data. So are you finding that the integrations are primarily about access and account management or are you actually starting to integrate with some of their internal processes?
I mean, like Procore's massive, right? So there's all these different things you can do with it. So are you finding that you're able to integrate, like with SmartVid, for example, you talked about your data or your output, providing a key input to what they do for a living. Is that, is that unusual or are you seeing that happening more and more?
[00:14:36] Matt Daly: I think it's the beginning of a trend that we'll see continue on for the next maybe decade or so. So you could say it's unusual today because I think we're at this very early stage of connecting tools. Yeah. I think now in this next phase, what you're going to see is more like what we're doing with SmartVid, where yes, you're connecting these tools together, but you're extracting value from the same dataset, all along the chain, again, from safety to quality to these other things that you can do. So yeah, it is a little unique, I think at this point. And I think some of the frustration the industry expresses at this point, is like, "yeah, sure. We integrate things and I can stick all this information in my project management tool, but now what," right? What is that really doing for me other than aggregating? And I think, we're sort of moving past the aggregation phase and more into the let's call it the analysis phase, where we're crunching a lot of this information with different systems to produce different business value.
[00:15:36] Hugh Seaton: Yeah, I love that. That's actually why I kind of dug in there is that you're, you know, you put your finger on it earlier in the conversation where you're creating, for the first time jobsite data, and it's consistent and it's obviously it's persistent, but there's a constant flow of it.
So it's almost like a living picture of what's going on in the job site, which you can just imagine how different entrepreneurs will be able to take that and say, great, well, I'm going to combine this with weather data. I'm making it up, but, and create some new value that makes the job easier or schedules more accurate, or predicts the delivery of materials.
I think there's just, it's really cool to see the beginnings of what you create being used as an input for someone else, because I think that's the promise of data, isn't it? Is that you get out of it, which you were supposed to get out of it, but other people can innovate on top of it as well.
[00:16:32] Matt Daly: Yeah, I think that's one of the more promising and exciting aspects of we'll call it sort of the bucket of reality capture as it we'll call it, a type of data. I think in this last wave, we saw all the different things that you can unlock. We're still seeing all the things that you could unlock when you digitize plans and spreadsheets, right? Like digitizing plans, brought this last wave of innovation where like, you got plans on the drawings and now you have people automatically extracting things from plans. We're still in, you know, we're still in that phase.
I think the next major wave is really going to be what can be powered by reality capture data. As this like really core feedback loop to the entire construction process. Cause it can touch all, it can touch every core aspect just about, of the construction process. Maybe not all of them, like in pre-construction, but, but even then, like, I guess even as I say that there's, there's aspects of almost every part of the construction process that that reality capture data can touch and inform and improve.
So I, I think you'll see in the future companies building more of a platform on top of reality capture data, where you have applications and sub applications that are like, look, we don't need to build the capture tool like StructionSite did, but man, we could really use that data to do this. And it's a completely niche application that look, I don't want to have to build the base for, but if I could lean on this dataset, I could do something really useful with it. That has nothing to do with what StructionSite does or some other company that does something similar. It's one of those things that I think you're going to see a lot of interesting things happening in that world as well.
[00:18:13] Hugh Seaton: The mind immediately goes to artificial intelligence and the need for a lot of data to be able to train a model, to do anything useful. Even if it's a pre-built model that you're... you guys use machine vision all the time so you know this better than I do, that having a lot of data that, especially data that's been tagged and, categorized.
You can imagine that that people will find ways to be... back to pre-construction. Right? Is that maybe one way that you do impact pre-construction is people are able to create predictive models based on big sets of data that you've produced over the course of some years.
[00:18:48] Matt Daly: Yeah. There's you know, what you can do with computer vision and deep learning with reality capture data is so vast. There's no world where we, as one company will be able to extract all the value that could exist in the reality capture data that's being captured. I mean, if you think about it, it's basically just this very accurate visual record of what actually happened in the construction process today.
You know, the pre-construction process is largely gone digital, right? You have BIM, you have lots of parts of that process that have embraced digital technologies and have been using them for years. Right? You could hear people maybe disagreeing with how effective any of these tools are, how well they work together.
But the fact is that part of the process is farther along in its digital journey than the part of the process where we build and we learn. And just like capturing what happened. There's just not a ton of data-driven learnings that come out of jobs because there's never before been a really good way to capture the work as it has happened, in the past. So yeah I think things you can do with computer vision, deep learning are, like I said so vast. I think it's going to be pretty fun to watch the different applications for all that pop-up in the next decade day.
[00:20:11] Hugh Seaton: For sure. And it brings to mind what happens with technology generally is first things become digital and then you start to get more and more, almost derivative value, right? Where it's the outputs start to kind of build on themselves where people invent things that you could never have thought of. The great example of that is that if you think about most of computers and computer...not really computer vision, cause that's an AI thing, but digital photography is just arithmetic, just lots and lots and lots of it.
And it becomes so cheap that you can do things you never would have thought of. There's a great book by a guy from University of Toronto, Ajay Agrawal who wrote a book called "Prediction Machines," I believe it is. He's an economist actually talking about AI and his point is when things become just exponentially cheap, they create possibilities that you otherwise wouldn't have had.
And exponentially cheap doesn't have to mean that StructionSite doesn't make money. It means that the cost of gathering that data before was measured in human steps, on a job site and then so on and so on. But as your platform develops and you're able to produce more and more data and it's not perfect, but it's very, very high quality. You create data that is just vastly cheaper than, than an equivalent data set ever could have been. Certainly having someone walk around with a camera would have been a little bit more expensive and then annotate what they saw and so on and so forth.
So, um, I don't know. I think there's a lot of, a lot of great possibility coming out of that.
[00:21:42] Matt Daly: Oh, yeah. I'll have to check that out. I mean, I think in the end, it seems like if you draw a line far enough from what are we trying to do with AI, it almost all leads to prediction. Where we are, we're really just trying to be able to predict outcomes before they happen. And I've actually watched our business firsthand benefit from a tool we recently bought internally that was helping us create some AI driven predictions that allowed me to make better decisions about the allocation of resources within our company.
And I see very similar things playing out for construction companies as we get more and more into this tracking stuff. As we can not only just say great, how much of the work is done? How are we tracking towards the actual plan that we had and based on how we see things progressing now, where are you going to be in the future?
Are you going to make money on this job or are you going to lose money? Are you going to blow the schedule or are you going to make it. And make those adjustments earlier. And it's just been interesting for me both as a building a company that is working towards a more predictive future, but also as a consumer of it, it's just so obvious how valuable that can be when it's really done well.
[00:22:58] Hugh Seaton: Someone once said bad news early, as good news. And you know, you can't be earlier than a prediction. And you're not really saying in five weeks, you're going to have a problem. What you're saying is when we've had jobs that look like this, we've had problems like that. So put some resources behind it or go look into what's going on and head it off pickup before it becomes expensive reality. Which is huge. Like if, and you're seeing that in other realms...where there was a Procore story that they actually talked about at Groundbreak from a GC in the mid Atlantic, where they had just pulled together all of their data across in a harmonized way, across all projects.
And they had about 50 going on at once. So it was a good sized builder. And what they would do is, now they had a dashboard where they had 10 indicators and senior management would be able to look at those indicators and decide where they would sit, put, you know, send resources to get things back on track.
And, you know, they're lagging indicators, a bunch of reasons why it's a good step in the right direction, but it's not what we're talking about. Where they were, again, lagging indicators and they may or may not have been right. But it's a whole lot better than finding out at the end of a job.
[00:24:11] Matt Daly: Yeah. It boils down to the same thing. Well, first of all, hat tip to that company for managing to bring their data into a single place in a harmonized way, anybody that's ever attempted to do that, I'm sure knows how painful that can be and how heavy of a lift that is.
But that's an investment that is going to pay massive dividends for that company in the future as they sort of get past the, like you said, the sort of diagnostics of okay. Or maybe the, you know, the reverse looking like what happened? Yeah. More about like, okay, what is going to happen? And then, and then that final piece of what should we do?
And so I think there's, it's going to be fun, Like I said, to watch the construction industry go through the same transformation that many others have, that is being supercharged by AI right now with which is making that journey from just collecting the data, digitizing it, getting it in a spot where you can make sense of it together and then starting to make some of those predictions. And in the end, we see ourselves as a piece of that puzzle, right? We're not going to be able to tell the entire story for any one builder. But we can give them a really clear view of progress today in a way that they've never had before.
And I think it's going to be fun to watch some of these other tools come, come into their own as well, that focus on different areas like safety, quality and things like that.
[00:25:38] Hugh Seaton: Normally I end with the future, but we've just spent 15 minutes talking about it. I want to take one second to talk or one minute or so to talk about an example of how somebody using StructionSite today has improved what they do. I mean, and any version of that, but normally it's nice to give an almost a case study.
[00:26:01] Matt Daly: Yeah, it's, the thing that we hear the most is...well, it really depends on who you're talking to. If you're talking to a general contractor, the thing we tend to hear the most is actually, there's a bit of a delay here where at the time of the project, we hear things from the project team about how it saved them time, how it saved them headache. There's just constantly questions that arise. The electrician comes in and says, "I think there's an outlet back there that got buried by the drywall or, you know, can we take a look right before we would be punching a hole in that wall and doing some rework, today with StructionSite, we just answer the question. I got a text message the other day from a trade contractor that was like, oh man, you guys just saved us 50 grand. We were going to have to knock this wall out and prove this, condition behind the wall for an inspection. This guy was going to have to fly out from Arizona to do it.
And like, Hey, great, thanks. You guys just saved us fifty grand. These are pretty regular stories for us, I wish we had a better way of aggregating all of them, but those are some examples of like the two most common, I think things we hear are these phases of construction, where you're about to bury something.
You're going to pour concrete. You're going to get one shot at looking what's in there before you. And then it's gone. When you're going to close walls, similar situation. Those are almost always the stories we hear after the fact when, when people have avoided rework, or just saved a lot of time and money with better data.
[00:27:26] Hugh Seaton: That's awesome. So, I mean, to two classes of benefit. I'm oversimplifying dramatically, but one of them is just day to day running the job better and getting better data to map to schedules and so on. And then the other, the other one is, oh my God, did we, did we do that right? Going? So one of them is running current jobs better.
The other one is going back in time to just check on something. Is that a fair assessment? That those are two big groupings of, of where StructionSite helps?
[00:27:55] Matt Daly: Yeah, really the last bucket is just the cost. The most common thing we hear that is honestly very few... I won't name any names here, cause nobody wants to talk about this, but we hear these stories all the time, where there was a change order that came in after the job was over.
And, we're about to write a check for, you know, 20, 30, 40 grand. And it turns out actually that never happened. And we just avoided that change order altogether. it's all the stuff that happens that ends up eating away at the profit you were supposed to make on that job.
It's usually something related to a change order or a delay claim or a warranty issue, something that allowed them to pull up StructionSite and say, Okay. That is actually what those field conditions look like on that day. Sure enough, we were delayed and like, you know, we're owed that money. It's just having that source of truth.
The efficiency stuff is what you see sort of day to day. It's the claims delays and things like that, that affect the cost to the company that, that I think we hear the most.
[00:28:54] Hugh Seaton: Yeah. You hear that all the time, even outside of StructionSite, right? It's the best documentation wins.
And what you're saying is now we've got a whole other dimension of documentation. It's not just what someone said in a daily report or what, you know, but somebody put in some other form of report, but it's actual, you know, video or photo imagery of what it was like that day.
[00:29:18] Matt Daly: Absolutely. I heard, we had a pretty funny story from a customer who was talking to us the other day about how there was, you know, there's this meeting at the end of a job and you all have to get together and kind of work out all the different who knows what on changes.
And, there was a contractor who we were not, who hadn't started using us that showed up to one of those meetings one time. And the electrician had been documenting with StructionSite. Guess who walked away from that meeting with the most money? Yeah, it was the electrician and, and that's because he had the best documentation.
[00:29:52] Hugh Seaton: Matt, this has been really great. I really enjoyed thinking through with you where this all might go, but also hearing about how you're helping contractors on the job site today. So thank you for being on the podcast.
[00:30:03] Matt Daly: Thanks a lot Hugh!