Welcome to another episode of Data Driven where we put the hard hat on

data and get our hands digitally dirty. Today, Frank

dives into the world of construction. Yes, actual buildings with

Amir Berman, VP of industry transformation at

Builderts. If you thought construction was all bricks and

backaches, think again. Amir reveals how computer vision

and data analytics are transforming job sites from chaos to

code. Think. Fewer delays, more precision, and slightly less

swearing at blueprints. So grab your virtual safety

goggles because this episode builds a strong case for AI in

hard hats.

Hello, and welcome back to Data Driven, the podcast where we

explore the emergent fields of data science, artificial

intelligence, and of course, without data engineering, really not going

to get very far. And speaking of data

engineering, my favorite data engineer is not able to make the call today,

but we've already scheduled this poor guest a couple of times and I don't want

to push it back another time. So it's just going to be me

today welcoming Amir Berman,

VP of industry transformation at Bill Dots. And

this is going to be really cool because it's really about. He has

a passion for digitally transforming the

construction industry. Now, I know the term digital transformation has

probably left a bad taste in some people's mouth, but I think there's real opportunities

in the construction space to leverage

tools from as mundane as predictive maintenance all

the way to fancy computer vision stuff.

Welcome to the show, Amir. Thank you. Thanks for having me.

Cool. And if you, you know, we're going casual today. If you're

watching this on video, we're both kind of in. One's in a black shirt, one's

in a gray shirt. And I kind of joked, like, too bad this isn't like

a hacker call. Like, you know, gray hat, black hat, Andy

could show up with the white hat. But. But I digress.

So I remember seeing a video

from, like, build 2017, 2018

at Microsoft, big Microsoft conference, where they showed a construction site in

computer vision where it basically said, you know, hey, where's the

jackhammer? Oh, jackhammer's here and it's in a dangerous position.

It's about to fall down. Or Tommy picks up

the jackhammer and he's not authorized to do it. It'll send an alert to the

construction manager and it actually sends an SMS and it becomes this

whole chat thing. Keep in mind, this is pre, like, chatgpt big

bang moment. Tell me, how far away is that

vision? You're nodding along, so you may have seen

this demo. So how far away is the vision of

a smart construction site? I would say it's pretty

close because we're practically there. Like you know, for some

of the audience, I'm pretty sure it's gonna sound like a sci fi

movie. But bear in mind that

for once, construction is probably one of the biggest and the

wealthiest industries out there. I mean if there's like a cool

tech out there, you've got to be sure that it's being used

or has been used in this industry. Like personally, I remember

I've been dealing with augmented reality for

construction sites back in 2016, I want to say

so even then. And we were not alone. I mean we were like a few

startups back in, back in 2016 or 2015,

I want to say 2015. And we were developing

augmented reality apps for the job site and

practically Microsoft was one of our design partners back then

randomly. So if

it's, if it's sci fi or if it's like today.

So it is pretty much like the, the present

I would say. But I also, but I also think though, like construction

sites I think are also the ultimate kind of

test bed for technology. Right. Like you, you're in

these if you have to wear a hard hat.

Clearly, clearly it's a, it's a rugged, you have to have a ruggedized

equipment. You have to have, it has to be reliable. Right. Because

if, if the system goes down. Right. You have an entire crew of people that

are billing but not working. Yes.

So it has to work. Right. So like that's always, I think has that been

a tension between like, you know, we have this augmented

reality technology and I understand, I remember seeing the demos too.

We're probably have seen a lot of the same kind of marketing

material. Right. Where you know, you put on the headset and like this is where

the pipes are going to go and this is where the wall's going to go.

So you know, whoever's on site saying like, oh well you know, we need to

adjust this, how do we adjust this? But I also

know like, you know, it's always cool to

have the new gadget, but that gadget has to work. And it seems like

construction could be a high pressure environment.

Yeah. Now whenever we talk about construction, I mean

people have like tons of different kind of examples running through their head.

Anything like I've buil a shed. Like

personally I can guarantee you that I've not built a shed

that's not in my sweet spot. But whenever we talk about,

you know, construction. So people have those all sorts of different examples.

It start by building a shed or like we renovated our house

or you know, three stories high kind of building

somewhere downtown, all the way to 40 story

high, you know, hotel, you know, let's say Austin or

a data center which is like, I know 2 million square foot or an oil

rig. Construction is pretty, pretty vast. So

the cool thing about construction is that the cost

of running a construction operation is so high, it's like

ridiculously high. People don't get it like how prices like construction,

especially like the major projects and at the same time these

guys and these companies are running like razor thin margins, you

know, would you, do you want to take a guess? Like what's the margin on

construction project? I guess it would

depend on who it is. If it's the real estate developer versus

the contractor that's pouring concrete versus the guy that's doing the electrical

or the girl that's doing the plumbing. But I would say, I would say

probably on a low end, probably like maybe 2%, 1%.

You're freakishly kind of accurate. I would say like if you're a

contractor, like a top tier contractor that does like you know, a major

construction, you're looking at single digits. It really depends on the

continent, like in the States versus like Europe versus APAC and so on.

But you're looking at single digits like and if you're saying like let's say

that we're building a half a billion dollar like

healthcare facility, right. So 3%

margins means that you don't have a lot of leeway for R and D.

Right. That's fair bearing mind. So you have like folks which are like the most

talented, most devoted people I've ever met. This is like the best industry to

work for in my opinion. Personally. People are devoted,

people are like mission driven people like you know, salt of the earth.

But at the same time, you know, no matter like how good and how solid

your technologies, you have very little opportunity to prove it

to them. That's true. Yeah.

The margins are that thin. Like you have to have a solid story,

right. Like I don't know what the final price of the HoloLens was, but it

was something like three, $4,000. Yeah, right. And if I'm, I

mean if I'm on, if I'm talking to a business owner that has a single

digit, you know, profit margin number, let's say 2%.

I have to come in with a really good explanation

of you buy this and you're not just buying one, right?

You buy this, it's going to save you X amount of

money. Yeah, yeah, right. It's you need to

come with a few things. First of all, at some point we'll need to probably

educate our, you know, our audience because we're not doing augmented reality.

You know, we're doing something completely. Right, right, right. I'm just, I don't want to,

I don't want to get fixed. I don't want to fixate on that. But. No,

no, no, don't worry, don't worry. I just wanted to make sure that the audience

are not meeting us instead of the. In case. But I mean, I, I would

imagine that, I guess that depending on what solution you're selling. Let's,

let's. Sorry about that. This is what happens, kids, when you have too much

coffee in the morning.

I mean, obviously predictive maintenance is probably

an easy sell for the construction industry.

I don't know if it's an easy sell. Like, nothing is easy.

Nothing is easy. Let's go back a few steps. So we said

it's like a high volume kind of, you

know, monetary wise. Like it's, it's capital dense, right?

Margins are super low and all the capital in

constructions are in within construction projects.

I mean, headquarters do not have a lot of money, not a lot of capital.

Why is that? Because all of their capital projects are yielding like low

margins, you know, let alone like, we're not talking about developers. Developers are doing

a whole different kind of ballgame. But let's say that you're a general contractor,

top tier general contractor in the states. You don't have a lot of, you know,

free money to throw an R and D. And at the same time, because

construction is such a vast and, you know, major

industry which has that kind of vibe of

being late to the party, even though it's not late for the party. From tech

stack perspective, it means that if I'm coming from a

contractor side and if I'm the person, you know, if I'm the CIO

or the person responsible for developing and implementing technology,

I'm being bombarded by people pitching me constantly.

So it's not a case where the industry is underserved, but we need

to have that responsibility as technology vendors that whenever we're

hitting the market with something, we need to be responsible and respect

the fact that the other side doesn't have a lot of margin

to invest in R and D. They do not have a lot of time. They

need to deliver project asap. So it means that we need to come to the

market really, really mature and we need to make sure that our

solutions actually work. And when they do it's terrific. It's

like magic. It's amazing. Right. I think of that old

triangle, you know, good, fast and cheap. Right. Like yeah,

time and money are both constraints, it sounds like in the construction industry.

So it has to be good, right?

It has to be good. Money is not always an issue. I

mean there's some money to invest just because capital

is huge and the opportunity to gain something

is vast. I mean if you can take a gc, like a gc, sorry for

the audience, short for General Contractors.

So if you're taking a GC and you can kind of help them pave

the way to break away from the single digits like

margin, the opportunity is endless. I mean

those companies are making billions of dollars in revenue, not,

not profits revenue. So if you can turn like a, let's

say theoretically take a 5 digit, a 5,

5% margin company and make it like a 6 or 7%,

that's, that's tremendous. They're going to be leaders.

Huge. Yeah, they're going to be leaders in their industry with that type of,

you know, so, so

our decisions in the field obviously built a

construction. So I, I had done some home renovations. My wife is always

knocking down walls or doing something. So I kind of know

I would not call myself a construction expert but when we did call in somebody

to build on an addition to our old house,

I saw how much that would cost and it was, it was only three stories,

right. It wasn't like a, you know, a skyscraper or, or data center which I

would imagine data centers are completely different animal in a lot of

ways. But are decisions based on

intuition, right? Because somebody, somebody has a plan,

right. They have the blueprint, right. And the blueprint seems like,

you know, if everything works out perfectly but where the rubber meets the

road, so to speak, is going to be on, on the job site.

So I mean I would imagine that a lot of the decisions

historically have been like, you know, the foreman

or the GC

superintendent has kind of like an intuition. But like are there

ways to use data, capture data and make the

decisions, you know, know where the

problems are going to be as well as making it more, less intuition based

and more data, data, dare I say data driven

type approach. What sorts of tools are there for that?

Now I think you're hitting the nail on the head because like, you know,

I think it was like one of my last flights. The reason we

postponed the, you know, the episode from earlier this week because I caught

yet another fluke which I'm constantly catching on planes came

back from London And I think it wasn't that flight, but the previous flight I

read Thinking fast and thinking Slow. Have you read?

Yeah, I have, yeah. Yeah, really good stuff. Shout out to. Who

am I to shout out like Daniel Kahneman. But you know, if you haven't read

it, go and purchase this either online or read the paperback.

But you know, he talks in the, in the, in the

book about, you know, system one, system two, right? Like two kind of system within

your human brain. I'm far from being expert, but basically you're

talking about intuition, like the way that we manage ourselves using intuition.

And what does it mean to have an intuition versus like a deep kind of

line of thinking and you know, the way that you typically

would analyze the more complicated, slow thinking process.

So if you take this back, like this system one to

construction projects, what does it mean to run based on intuition or

hunch? Let's take your veteran superintendent, superintendent,

like the person who really runs the job on the job site

from the general contractor side, and let's say that he

or she will have like 20, 25 years of experience. These

guys can, you know, can sniff, can sense that

something is wrong. But in reality, you know, without having the technology

on their side, typically what will happen is that their intuition

will kick in when it's a bit too late. Why is that?

Because let's say that you're doing like a 20 story high, you know,

let's take that 40 story high somewhere building in Austin, Texas, right?

That's going to be, I don't know, half a million square foot

of a building. I'm going to have a crew of in between 10 to 20

people from the contractor side. And there's

literally hundreds of people working on my building

installing ductwork, electrical wiring and you know, drywall and

Sheetrock and you know, you name it and everything changes

each and every day. And you as a superintendent, even though that you

have the best kind of experience ever in the job and you have a really

good intuition, your threshold, right, to

noticing that something is off, you're only human, so it's

natural for you to sense that something is off at some point. But what

technology can bring to the table, and sorry for the very long explanation but

technology can do, is to lower the threshold for you to be

sensing that something is off. Let me give you an example. Let's say that

in, within that same building, you have a crew of people that installing

ductwork, you know, there's going to be, let's give it like

an even number just for the example. I'd say like 100,000

of linear footage of, you know, duck work.

And they need to do it at a certain pace and to work at a

certain sequence. And let's say that they're like, by week two or

week five, they're off by 7%,

right? They should have done like X and they've done like X minus 7%.

What is the probability of that veteran super to

kind of miss that? There's a high chance for them

to be missing that point. Why is that? Because someone else is yelling. Because

someone else is like, hasn't been delivering as they should

be. And the gap is not 7%. They're missing by 50%. Or

there's a truckload that was supposed to get to the job site that day and

it hasn't gone there. Or like there's like a design change. There's so many

moving pieces on the job site and for them to be missing

the fact that that team is lacking like 7%

and the week after it's going to be 8%. So you're looking at the kind

of a snowball effect. So at some point, I know

by week 20, if, if the ship is like off track,

right, someone will notice it. But the trick is that you're

noticing too late. Using

technology is like, you can combine the system one, the intuition,

which is basically intuition if you don't know it. Intuition is like experience,

his knowledge, expertise. It's like how your brain is being, you know,

rewired as time goes by. But if you combine that

intuition with technology, that lowers the threshold all of a sudden. You

don't need to wait until week 20 to sense that you're off by 7%.

On the second or fifth week, I'm going to say like, hey, you know what,

you've been doing tremendous work, but bear in mind that you're under delivering by just

like a tiny bit. Let's go to the root cause of that and figure out

what we can do together as a team in order to get better, back on

track before it's being too late. And what I sense that the biggest

opportunity for construction with technology is exactly that is

like one of the opportunities is like lower the threshold so we can

let humans do what they do best and we can let technology do what they

do best, which is like the heavy lifting, the long tail, like the all that

kind of boring, quote unquote analysis of this

situation so that the pros can be like, you know,

do whatever they do best. Right. And I would imagine too,

like, I mean, it's Probably a lot easier to, you

know, once it gets to 7%, right. It's probably

one level of effort, but if you catch it at 3% or 2%, it's probably

a lot, you know, like if a concrete shipment, I don't

know, you know, misses its deadline or is late,

the downstream effects probably AI is better

at figuring that out than a person would be. And it's not a, it's

not, it's just you. Every human on earth is limited by

human perception. Right. The gateways of that. Right. And, and not that.

That's. I think, I think you said it best. Like I'm a, I'm a big

believer in the idea that AI is meant to be an augmentation technology

for humans because there's things that AI can do better

and it's, you know,

and there's things obviously that humans are going to do better than machines for the

foreseeable future. Right. But I think

it's interesting is that when you think about, you know, AI and construction, right.

It's probably, you know, everyone I, you know, immediately like I went to the

computer vision demo, right. From a few years back, right. But

it's probably this is. It sounds to me that construction is a very logistics,

heavy business, right. I need to get people in a place, I need to get

gear, I need to get equipment, I need to get

material there and that. And there's probably a certain timing of it, right. It's

probably very heavy on the waterfall process where you can't put

ductwork if there's no, you know, I guess the iron skeleton

on the building or whatever technique, right. If there's no floor, you can't put the

flooring down. If there's no walls, can't paint them. Right. I mean, it's like from.

It kind of goes this and I would imagine

that that creates a very complicated

web of interconnectedness that.

Just thinking about it gives me a headache. Oh yeah, yeah.

I think you, you're exactly right. Like it's the knockoff kind of

cascading effect because everything in construction is sequence. Like

the most, you know, the easiest kind of example is like you need to do

the groundwork in order to do, to, to erect the structure. Right.

And once you have the structure, you can start pouring the slabs, which are the

concrete kind of floors and ceilings. And once you have the structure in place, you

can start, you know, to, to install all the fit out, you know, all the

internals. So that would be like the facades and windows and

externals and guess what? You need the building to be

what we call wet ready before you can install

elements which are sensitive to weather. Right. I wouldn't go install

my precious kind of sanitary work before

I know that no damage will be caused by weather. And, you

know, when we're talking about mechanical and electrical and plumbing

equipment, there's a certain sequence. If you look up, you know the audience. If you

look up and you have those kind of.

You can see the ceiling scheme. You know, in office areas, you would

see that there's, like, a certain pattern in your overhead. Mechanical,

electrical, and plumbing equipment. Typically, there's going to be high

difference. So, you know, ductwork, which are the biggest kind of pieces, will

go first and then sprinklers and then, you know, and so on and so on.

Mechanical piping, electrical conduits, you typically will go

last because they're the most flexible. So you're right. There's a certain

sequence, and once you have kind of a delay or a

problem in one element, there's going to be a knockout effect to the

rest of the pieces. And you want to make sure that one. You keep the

right sequence. And if there's something that isn't ticking the

right way, you need to fix that asap, because everything that will

follow will be impacted. And not only that,

sorry. You want to make sure that you

keep a certain flow. Like, it's funny, but in

construction, it shares, like, a bit of, you know, Zen kind of.

Right, right. Elements. Because bear in mind, there's like,

dozens of different trades and contractors and supply chain elements that are

working together seamlessly, and one depends on the other.

And if I come trade number one, let's say I'm doing ductwork,

and the next one after me will be the sprinklers guy. If I'm

late, that's going to affect the other team. And if they cannot

pull their people to the job, guess what? At some point, they're going to pull

them off from the job and you, you know, divert them to the next one.

And me, as a superintendent, is like, the current project will suffer from that.

So you want to make sure that everyone is working according to pace, according to

their sequence at a certain flow. And it's really hard.

It's really hard because, like, you plan your job perfectly

on day one, right. But the minute you started, you're being thrown with

everything possible, like weather, supply chain issues. The

owner will change the design because of reason. You know, the

marvel that, you know, you kind of. You

wanted to get from Italy, stuck in somewhere in the ocean, like Everything

will be thrown at you. And you need to have that really

good data collection system that, you know, keep

tracks of everything for you so it can raise up all the risks

and all the kind of flags you need in order to make the right decision.

So this is kind of the story. You really want to make sure that you

keep up with the sequence because every kind of grain of, you know, dust that

goes into that mechanism will

probably. You know, it seems

like you can, you can, like you said, like a Zen thing, like it has

to exist in a certain flow state and the universe is going

to conspire to make you get out of that flow state.

Right. I, I imagine weather probably plays into it, you know,

and, and it always fascinated me to see when people would

build homes. I used to live in this big suburban development

in New Jersey. As they were building it, we had these huge blizzards

that winter. And I just remember seeing like the entire frame of the building

was exposed to, you know, the snow and the ice. And I'm thinking

to myself, how is that going to impact, you know,

you know, in a one story townhouse or building? It probably is not that big

of a deal. But like, I just wonder like, how do the bigger projects deal

with this, right? If it's a hurricane, if it's this, if it's that. And

I could just imagine a logistics nightmare, especially the bigger the project, because

the bigger the project, the bigger the mart. I mean, the bigger the,

the crews and the bigger all of this. And, and I think you're

right. Like if, if the ductwork guy gets delayed by a couple of days,

I would imagine like the sprinkler contractors, the

plumbing and all that, they probably have, are working multiple

jobs, right? Like, so they're probably like, oh, I have, and I have

Bob and Tony working on that. But if you're for this week, but

if you're delayed by a week, I got them over here now that screws up

your schedule even further because you can't get those people. And I would imagine

it's logistics nightmare. Yeah, it is, it is,

it is. But, but to be honest, I would say, you know, that industry, this

is how it operates. So it knows how to handle the

unpredictability and how to kind of

change plans at the floor level and

adjust itself. But the key is, and I think that

what's really happening over the past few years, and it's

not just because of AI and technology, I think it's mostly about data structuring

and ability to really represent the project

digitally. So you can represent it

digitally, all the moving pieces. So you can start simulating, you can

start predicting using predictive analytics and so on.

What it offers is, like, it offers. Like, the.

People on the project to kind of work with different options and say, like,

hey, you know what if I'm 7% late? You know, that previous example

on the ductwork, what does it mean for me? Like, what's the end

date for me for that activity? Let's say that I need to have all the

ductwork installed by, I know, December this year.

That's my plan. That's my schedule. Now I'm off by 7%. If you

extrapolate and say, you know, if we continue the same pace, you know,

relatively the same pace, am I going to finish that

on January or February or, you know, what's. What's the knockout

effect? Because once you know that, you can start plan the remedy, and

you can say, all right, you know what. What happened? So far, it's in

the past, but we need to get our, you know, our stuff together. You know,

sorry, keeping my language and, you know, back on track. Sorry, I was

almost there. And you can start having, like an adult conversation with your

supply chain and say, like, hey, you know what, guys, let's go to the root

cause of that. We need to amp our game by,

you know, by that amount. Do we have enough labor on

site? Do we have enough materials? Like, can you. Can you increase

manufacturing of the missing ducts? Maybe? Can I change my

sequence? You know, I have, like, the most amazing example from, you know, a

year and a half ago, we

launched a new product, a new feature about 18 months ago, which

is like a predictive analytics for delays, which is tremendous for

a job site. I remember launching it. And like

everything in life, when we launch a product, we, first of all,

you develop it in the background and you

have early versions of it. And I remember working with

mine, my. My first kind of beta

user for that, one of the projects in the uk, And I told

him, like, hey, there's a coming conference, you know, how about we get on stage

and present together that example from back in the day? And he was like, you

know what, I'm all good, you know, presenting with you, but I have a

new example. I was like, what are you talking about? He was like, you know,

he just released a feature using predictive analytics. And we noticed that my

electrician, he's actually six weeks behind schedule, and it makes zero

sense because he has his whole crew on site every

day. I was like, how the hell are you kind of six week behind schedule.

And the electrician, you know what he tells him, he was like, you know what?

I'm waiting for the elevated floors, right? If you know

what I'm talking about. Those like, elevated floors. I'm waiting for the elevated

floors trade to be finishing in that area for me to getting on there

with my, you know, ramps and everything to be working. And they

stayed together. And I think he was telling me like, why the hell are

we waiting for the elevated flows to be completed? Can we just like have the

electrician go there instead? You know, change the sequence. That's it. And

they change it immediately. And the only reason they could have, you

know, add their kind of experience saying, like, you

know, we just change the sequence. That's it. The only reason they could have done

this because something raised that flag and said, like, hey, you know what? You're

going to be six weeks behind schedule electrical work if you don't

do something right now. So it's, you know, once you're off

track and once you miss something, it's not the end of the world

as long as you kind of address it.

I just, I just love that story because it represents so much of

the industry and its ability to make the best, like,

decision in the split of a second. No,

I think that's a good example of the AI flag something and people kind of

like sat down and talked through and I guess one of the other things

you kind of said was the ability to represent a building

digitally. I would imagine it helps a lot to

have that and then test out different scenarios. Like if

we switch the order this way we'll save two days, right? We'll get back two

days. We change it this way, we'll get back four days. Right. Or,

or something like that. And again, I think the, I think the

construction industry has always had to be resilient for a number of reasons,

right. I think that's something

I don't think people would necessarily appreciate from the outset. Right? Because you always

see, like people always notice when something goes wrong, right? Like, oh yeah,

that building. That building was supposed to go up, you know, in the spring.

Here it is the fall. Or, you know, God forbid there's some kind of

collapse. Like there was. Was it Thailand? I think it was

Thailand. A building collapsed, unfortunately.

So does the sequence of things or the normal sequence of things

change by region? Like is. Is the

US going to have a different order of things or. And like, how

much does zoning affect that? Right. Like, you know, do you have a thing where

you know, well, the local government, the local county or state says you can't

do this before that. Like is that, is that a thing?

Generally, I don't know. I'm pretty sure that there is a zoning kind of thing,

but it's not my. Okay, I was just. But I would say, you

know, think about this one. No building, like most buildings are

not kind of cookie cutter. This is kind of another challenge in

construction. It's like someone, I'm

quoting someone, I can't remember who said it, but it's like you're building

a one time thing, thing, factory. Like you're building a

factory, right? The factory is like the team and the job site that need to

kind of build that building. But it's, it's, it's a factory that

you're going to use once, right? And that factory

needs to build the building. And no building is the same as

the other, right. One will have like a

lowered suspended ceiling, the other one will not. And even like

simple thing like you know, like drywall, like Sheetrock.

Some of them will have insulation, some of them not. Some of them will

have like the two coats of paint. Some of them

will only one. Some of them will have glass walls, some of

them will have brick walls. Nothing is the same. So

sequence changes and varies according to the building that

you're building. Not talking about different verticals.

The healthcare facility is like completely different thing from

residential project. It's a different thing from

an airport or a hotel or data center or an oil reg.

It's like comparing like the, you know, the F1 or in the

NASCAR kind of car to my lousy vehicle that I'm driving

my, my day to day. It's like a completely different animal.

So there's going to be a lot of variations and differences. And this

is like one of the challenges because you only have one shot on

making that building on time and on budget. That's it. You

only have one time. Interesting.

Super challenging. Super challenging. That is industry.

Maybe it's a good example because you know, I promise like the audience just like,

you know, I'll give it like a really short kind of explanation of what we're

doing and then maybe we circle back because like

I think we kept the audience like in the dark for a bit.

Mysterious, like I'm serious about what we're doing. So build outs,

like simplistic. What we do is use computer

vision, right? We use computer vision to Compare

visuals from 360 cameras to your plans

and schedule. Right. Oh, the result is that what we do

is that we analyze the results from the computer vision and we can

programmatically provide you progress

data, like, compared to analytics, like progress

data for your job site. So at any given moment in time, I

can tell you precisely how well are you progressing against

your plans and against your schedule. And it goes down from the very

top level, saying like, you know what, you should have been like 80% so

far in your project altogether. And you just like 75. Or if you're

doing really well, like you're 82 or 80. And it goes down

layer by layer all the way down to the very specific

conduit and specific wiring. Right. You break it down by the

different activities and trades. So Electrical will be

70% out of 75. Ductwork will be so. And so goes

all the way to. On that very floor. It's going to be that percentage

complete and going down to that specific

element type. So it's going to be drywall versus block work or versus

concrete walls and specific wall pieces and

specific floor and so on. And everything is backed by

photos because it's computer vision.

And I'll finish by that. Because the way that we run this

product is that every time you start a new project,

we're going to obtain two things. Your

schedule, which is like a simple Gantt chart. It's far from being simple, but

imagine a Gantt chart. Every major construction

project has a schedule. And the other thing is that

we taking the 3D models, believe it or not, for people who are not part

of the industry. The blueprint that you remember from, you know,

movies, by the way, my first impression of blueprint, have you. Do you

know Die Hard? Yes. You remember him pulling

the blueprints. So there's still blueprints, like in

2D these days. Everything is like still working in 2D, but

major construction and even lower than that are being designed

in 3D, which is tremendous, right? Pretty cool. So we take the

3D models and schedule and we create something that called 4D.

4D model. 4D is like the 3D model that has that

time kind of factor to it. And all of a sudden we have a

digital representation of the project that you're building. Let's say a healthcare

facility somewhere in Jersey. Right. So we know how the

project should be looking like, should behave. Like, what's the sequence?

Who are the trades working there, how many walls, how many pieces of ductworks,

electrical conduits and sockets and so on. And every time,

every time someone takes a walk on the job site with

a hard hat and a 360 camera mounted to the top of it.

Turn the video on and just walk the job. You

walk the job. You then finish it. You upload the video

to our computer, like our servers to our platform. And

we use computer vision to do two things. One,

we precisely locate each and every frame in the video.

You don't need to tell us where have you worked, just walk the job.

We'll figure out the exact positioning of each and every frame in the video.

We're accurately positioning it against the model and against your

plans. The second part is that we use computer vision to

automatically annotate and extract data from

that frame. Let's say that you walk across

a block kind of wall that has an opening. So we know that

that walls in your camera, in your footage kind of is

compared to that wall in the model. Right. We can mark this

as done and we can know whether it was like it has

plaster in that, whether it was coated and so on, so on, so on. So

this is basically what we provide. We provide progress

data, which is equivalent to analytics to the people on the job

site. So that that super. Remember from the previous example,

they know on each and every day whether they're on track or

not. And if not, like, what is the reason for that? Which trades

are behind schedule, what activities are problematic, do they have any quality

issue, what's the predictive analytics says about the end date and what should

they change and to what extent in order to get back on

track and finish the project on time and on budget and

obviously as safe as possible. That's interesting. So

you have this computer vision solution that can be very granular.

It's almost like you have like. What

did you call the person who's in charge of the project? It wasn't foreman, it

was superintendent. Superintendent, yeah. If you're typically.

It's like you have that person on every floor at

all times paying attention to everything all at once, right?

Yeah. And you know what? You can't have this.

You can't have. People are people. People are people.

Yeah. But to be more fair than that is that one. Remember

that 3% or 5% margin, I don't have money

to have enough superintendents on each and every

part. Is that there's a huge shortage

in professional

sophisticated talent in this industry.

The industry is lacking so many people, like all of

the people in the industry are extremely talented, really

smart, really voted. But there's not enough people out there.

And the sad news is that more and more young

professionals are leaving the industry. So you're not only fighting

to recruit people, but also to retain them because

it's a hard physical labor job.

So I wish we could have had like so many superintendents on the job

site, but honestly we can't. But

it's not the end of the world because if you harness technology, you

know, when you combine that technology with the system, one kind of, you know, those

people, all of a sudden you turn them like to, to be more

superhuman in a way. They control more square, square footage

of project. They can know more. They can be,

God forbid, live, you know, early, you know, to be. Right, right, right, right.

To keep their kind of mental health in place because really

kind of it's, it's a hard job. I mean, you need to respect those

people. They working so hard. It seems like it would be very stressful

job, like, especially if when things go wrong and it sounds like things

almost always go wrong a little bit. Yeah, I

would say it's not for me to be speaking about this because I'm. Even though

I've been serving the industry for the past more than a decade, I'm

excellent. So I don't have, I haven't heard. Earned the rights to talk

about this. Right, right, right. Yes. It is known in the industry

that, you know, mental health is an issue. And I think

that if we technology vendor can help just a bit, you know, to

let them go back, spend time with their family and you know, to

decompress for a bit and to be less stressful over the weekend and over,

you know, nights and everything. That's. I would love

that for it to happen. No, I think that's really cool. I think

it's an important. People don't people, I think

under underestimate mental health and things like that. And

to your point, like if there's going to be a

skill shortage. Right. Even if we solve the skill

shortage today. Right. To get that level of experience

that a seasoned like superintendent would have is

going to take. I mean, even if we fix the pipeline today, the

downstream effects and the shortage in the pipeline are going to

cause problems for, you know, a generation potentially. Right.

So how do you, how do you, how do you mitigate that? And I think

this seems like it'd be one way to mitigate that where you could have,

you know, and it's really using AI, I think,

where it's good at. Right. Paying attention to every detail at all times,

everywhere at scale, and then collating

that data and getting to the point where, you

know, AI does a really good job

of, you know, doing the Slow thinking system very quickly.

Right. Like so I think, you know, if we, if we kind of

leverage it that way, I think it's. And I also think too like it's a

very practical use of computer vision. Oh yeah, right.

And I would imagine as time goes on, you'll learn more

about what you said would happen in your system versus

what actually happens. So you have like that training loop probably in place.

There's this training loop and I

would even say, and this is something we're doing already. So

one thing is to optimize the existing project at hand. Right.

Go back to that half a billion healthcare facility

in Jersey. Right. About the next one. I mean

one thing that we've been doing because our computer vision

generates so much data about plan

versus actual, about how actual progress happens on the job

site versus how it was planned. What we're doing right

now is that we look at future jobs and we look at their schedules

and their models and their plans and we can say, well, what is

the probability of different types of risk to happen

on that scheme based on previous historical data that we

have? So let's say that we're not building a healthcare facility in Jersey,

but rather like in, I know, in Indiana. Right, right.

And how many healthcare facilities have we built so far?

How much information have we gained in order to

validate future plans and to de. Risk future plans. Right.

And to build. Right. For the first time. And this is the other example of

how technology and AI can, you know, can kick in because we're not just looking

at the one time factory that we're trying to build, but rather

optimize all the current and future pipeline of our business, which is

tremendous. And all of a sudden you can schedule better. Right?

Because if you look at project scheduling, just to give you like an example,

I've seen construction project schedules with

more than 2 million rows. Right. Think about

a project schedule that has 2 million rows. I've never

seen anything like that personally in my job, you know, for tech company.

So it's beyond human scale. So if you use

historical data and again AI and computer vision and everything

else to kick in to do the stuff that is really hard for human to

do. How did you say it? Like allow computers to do system

two things really fast. Which by the way, I would buy that

T shirt if you get this. I think I'll make that T shirt

make it black.

So all of a sudden you can, you can leverage technology to do other things

as well, like better planning, better scheduling and look at all the other

parts which are heavy lifting tasks that we can

kind of take it from humans not because we want to replace

them, but rather we want to keep their abilities and experience

to do the really hard reasoning and decision making

and, you know, what if

scenarios and so on, and to let technology to kind

of lead the way on the repetitive kind

of hard job. So it's not just about project

control analytics, it's about predictive analytics and better schedulings and

better planning and better kind of de risking for the entire industry, which is pretty

cool. Cool. How did you get into this? How did you get into

it and construction? Oh, that's, you

know what I've, I just 40 about a month ago

and I've been playing with my, you know, lifelong

decisions for the past few years, you know, thinking I'm happy with everything

that I have. But you know, I have been thinking about stuff. So originally I

came from technology, you know, pretty young, about 20 something.

Started in advertising tech back in the day, which then. Still cool.

Yeah. My first kind of role, I remember

I was a product manager for an advertising tool, believe

it or not, as an add on for Flash. Wow.

Yeah, I was like doing some product management for an add on for

Flash and at some point I kind of fell in love

with data analytics. That was my sweet spot, kind of. I know why.

I love numbers, I love reasoning, I love logic. And

I worked in a company called Datorama, which later on was

acquired by Salesforce, which is pretty cool.

Not a lot of credit for me in the acquisition obviously, but you know, it's

just a part of the team. And then I remember getting

a phone call from a friend and that's pretty cool. He was like, you know

what, there's a young startup in

construction tech, you know, looking for the first product manager. Do you want to join?

Believe it or not, my first response was like, no, forget about it.

There's nothing to do there. You know, it's probably going to be boring. But I

took the meeting and you know, eventually I joined the team.

And I remember the first few months I was flying like hell. I was

flying to, you know, Indiana and Boston and New York and Turkey

and Thailand and you know, the UK and France.

And the reason I fell in love with it was people.

Eventually you meet that superintendent and you meet that foreman and you

can see everything in their eyes. It's not just, you know, you're not optimizing

that additional impression on that Google

search ad or whatever, full of respect or everything

dealing with this. But that's not my cup of Tea. I'm a people

person. And you remember, you know, you could have seen everything on their eyes. And

I remember that, like, you know, back in the day, working in the augmented reality

kind of app that I told you about, I was working in a project, not

working. I was like, you know, demonstrating a technology app in.

I think it was Lebanon, Indiana, if no one

knows where it is. They were building a veteran

healthcare facility. I was, like, demoing the app, and

I can't remember why, but I think it was like, they told me that everything

that they build in Diana is built in swampland, that

they need to dig a well into the basement. Like, I know

70ft of well and need to constantly pump the water.

And remember I'm holding, like, a device with augmented reality. And they tell me,

like, hey, can you. Can you come in for a second? I was like, yeah,

sure. And they were telling me, like, hey, you know what? We

believe there's a problem with our well. Maybe it's dislocated or

something. I was like, all right, let's check with the app. Obviously,

spoiler alert. It wasn't working perfectly.

And I'm trying to locate the well, you know, in the model,

in the. In the. In the. In the plans. And

I couldn't see anything. But I had a weird

intuition. I told them, like, guys, what's the probability? What's.

Is it possible that the well is positioned well, but the

diameter is different? Maybe, maybe. Maybe the diameter thing is wrong.

Because they were trying to kind of coordinate the position

of a wall against that well, a kind of a pit in the floor.

I was like, maybe the diameter is wrong, so this is why the wall is

not working against it. And they checked it, and I

don't know how I had this intuition, but I got it right. And the diameter

and the story is that I remember the face of that superintendent.

It turned white immediately. And I

could see that everything is personal. Everything is, like, very human. You're dealing

with, eventually with human that devote their life to. To this industry.

And I just fell in love with this. So I know I'm sold for

the industry. And looking back at my childhood, my

parents, they had this family kind of business for printing 2D

sheets for construction. So my.

All of my summers from age six, probably, I was spending, you know,

folding huge 2D sheets for construction. So maybe, maybe

if, you know, you're looking psychologically, maybe like it's kind of

something that brings me there, but that's definitely my passion. This

is how I got there. So it's A mixture of data analytics, AI and

construction. That's cool. That's cool. Obviously

you mentioned thinking fast and thinking slow.

Audible is a sponsor and there is an audiobook version. So if you go to

thedatadrivenbook.com, you'll get one free audiobook

on us. And if you

get a subscription we'll, we'll get a little bit of

kickback and help support the show. Any other

audiobooks you recommend? I'm not an audiobook

person. I tried it once. Are you doing audio or paper?

I kind of like, I have printed books, I have audio books

and I also recently got a Kindle Scribe which I actually kind of like.

I like it. I like it. If you look at a lot of the.

I've been a big tablet PC fan, like pen computing fan since like

Windows pen in the 90s and even I had an

Apple Newton if you. That's really. Oh yeah, yeah. So I've been a

big believer in that tech for a while. So my,

I saw there's something called the Books which is

basically the actual E ink screen

is A4 size. Oh. So you can drop

PDFs into it and it's like you know, PDF books and it's like

perfect but it's like 6,

$700. So I was like, I don't know if I like it but

I look at the remarkable because I want to be able to take notes in

meetings without being distracted by notifications.

But when I saw the Kindle scribe I was like well I need a reader

and I need a note taking platform and it happens to be the least

expensive of the three. So I'm going to try it out and I like

it. What I really like about it is the

screen's bigger than my other Kindle. Right. I like

the E Ink display because it feels there's no glare, there's no

nonsense like that. I also not staying awake

until like 2:00am or something. Exactly. And you don't have to light on

because it's backlit. You can read it outside but also you can take

notes in the margins. You can open up a different notebook and

kind of write out, sketch out ideas. Yeah, I mean

I'm, I'm a big fan if you're not a big.

And I already have a lot of stuff in the Kindle ecosystem so like it's

not a big loss. I know some people militantly hate the Kindle ecosystem

and that's why like they would go with remarkable or books or something like that.

But you know, which I probably Will end up getting one

if you know when the price comes down. And I go

everywhere now with this little like Kindle and I've only had it like almost a

week and a half. I should probably buy one because I fly

a lot and. Yeah, if you fly a lot. Yeah, I fly a lot and

I, I love reading on planes. This is like the best time usage

ever. You know, if you don't need to work in presentation or to work on

planes, read on plane because it makes you fall asleep faster. Now

that's true. One, another kind of recommendation that I can

give to the audience. First of all, read books, kids. It's important.

Two, have you read the Innovator's Dilemma? No.

So the Innovator Dilemma is like Innovators Dilemma is kind of

one of the best kind of business startup books in my opinion.

It's written by, sorry if I'm not pronouncing it right, I think it was

Clayton Christensen. Look it back. It talks

about why do large enterprises

are late in adopting new technology and

should, should they adopt the new thing

on tech or should they wait? And why are they late

in adopting certain technology? And don't want to give you spoilers but you know,

every time you hear about something new, you know, choose your

current hype, whether it's like vibe coding, I know mcp,

whatever, you know, knocks you out. But sometimes

you think about like why do, why don't Amazon or Google

or you know, Apple or you know, all your top hundred

Fortune 500 companies do not adopt it immediately? You know, why is

that? And there is a certain dilemma. Should they adopt it really

fast before the market, you know, demands it, or should they wait? And

I don't want to spoiler read the book. It's tremendous. It

goes through like research from the 80s and

90s and explained flawlessly like the dilemma of

developing and adopting a new technology right away or should they

wait? There's kind of balance in the middle. I

really recommend it. Awesome.

Awesome. I will definitely check that out. What

about you? What good recommendation do you have that you

read? There's a really good audiobook I'm listening

to now called like 48 days to work. You love

the work you love of

and it's basically idea. I like my job but like you know, it, it, it

really, it's. Anyone from work is listening. No, I

actually do like my job. But like there's like, you know, as you get, you

know, because I'm. I turned 50 not that long ago. Right. And like every time

you have a Birthday with a zero on it. You always have this kind of

how am I doing? You know, tell me about this.

And you know, when I turned 40, I had this crazy idea I was going

to become a documentary filmmaker and long

story, and I went and I really studied up how to do

filmmaking and stuff like that. And then I

realized like how little documentary filmmakers make.

Oh yeah. And I realized, you know, maybe I should because

I was, you know, I was very invested in the Windows

Mobile, Windows phone platform, Windows 8. And then when that kind

of hit was a thud, I kind of realized like, you know,

whatever, you work in technology and like a particular field

kind of flops, you know, that particular niche that you're in kind of flops,

you kind of reevaluate. How did I get here? Right? And it

was almost by chance that I attended a

Microsoft research conference like

over 10, 10ish years ago

where, you know, they were talking about,

you know, AI and like what this is. And at that point I just thought

of, you know, data as SQL and you know, Power BI

dashboards. Like that was my, that was my impression of it. But

when, when I saw that there was an actual engineering discipline to it and

math that will make you go crazy. Like it was

a good technical challenge to get into. And you

know, at the time I was at Microsoft and they were talking about how they're

going to add AI to every Microsoft product, which in 2015 sounded insane.

Yeah, right now, I mean now we see it and like

everybody's adding everything to AI, even if it needs it. Whether or not it needs

it is not really a concern. But

it's, I don't know, like I just. And you know, fortunately that

was the right choice. Obviously people thought I was crazy because I was, you know,

walking away from, you know, years of

like front end development on Windows into a completely

new space and everyone thought I was crazy. But I'm like, nah, there's

something here. And it's, it's fun, it's challenging, it's exciting

and that's that. That kind of explains my current fascination with

quantum computing. Right. Like it's like, you know, it's, it's not quite

there. It's not quite there yet. Right. And people will

argue. Jensen Wong says it'll take 20 years, Bill Gates says

shorter. Some people say three years, five years. It's such in

an stage of a technology development that

we're really barely at the transistor stage. Oh yeah,

here, right. So like it's really like an opportunity to get in and the Math

is hard. The math will give you headaches for sure. But

you don't have to understand all of it to build systems on top of it.

Right. Like, and to understand the impact it's going to have on the industry.

And like, everything. And like everything, the. The smart people will build the

infrastructure layer, and on top of that, you'll have the operation system, the application

layer. And, you know, before you know it, you will build application in an

abstract way without knowing everything that's, you know, underneath the surface. A

hundred percent. You know, at one point, if you were building a computer, you needed

to have an, you know, electrical engineers on staff. Oh, yeah, right. And you

needed to really use those bytes, you know. Well. Right. And how.

How, you know, memory works and how, you know, everything. Efficiency work.

There was one of the mythbuster guys, had

a thing where he talks about a bit from an early computer, and it's about

the size of this water bottle. No way. Something like that. It was. It

was a little smaller than that, but I mean, it was like. And he was

like, you know, it was about that big, and it was somewhere between the size

of this and a spark plug, but it was big. Right. So, like, if you

just think about that, like, and then. Then some other YouTuber did this whole visualization

of what does this look like? What would this look like to have a gigabyte

with those? And it was turned out to be like a skyscraper size thing.

And it was, I don't know, like, to your point. You're

right. Like, the infrastructure layers that we're used to in technology today

are not there yet in quantum. Right. But that also means an

enormous opportunity for those to get in at this level.

You know, whether or not it'll pay off in five years, 10 years, 20,

I can't really say, but it's definitely. I know. It's definitely happening.

Yeah. Well, fun fact. The audience know I know zero about.

Right, right, right. Well, every time I think I understand it, I learned there's a

whole other thing behind it which is both fascinating and, you know,

fun and annoying, but shameless. Plug. I do have another

podcast called Impact Quantum, where we do take.

We do take a look at what Quantum is, where it's at and how it

means, what it means for people's careers and stuff like that. Who knows, Maybe

we'll meet again in decades, talking about. Absolutely. Machinery and construction. There

you go. Well, we'd love to have you back on the show if you're interested,

and maybe talk more about the individual solution, but I really enjoyed our

conversation. Me too. It was a pleasure. Like, thanks for having

me. Thanks for the audience for staying until now. The people who

stayed. Oh, no problem. Oh, one last thing. Where can people

find your company? It's called Bill dots. Yeah. So

buildups.com. like, go to our website, go through everything

that we offer. There's tons of education,

you know, case studies, webinars, you know, we're talking, we're all

the way in social media. Go through LinkedIn to either build

out's profile or to my profile. We're happy to chat

and we're happy to geek out. I mean, eventually we're construction

geeks. Love talking about technology, love talking about

construction. So reach out. We'll have to chat. Awesome.

And it's build. Ots.com, right?

No, it's Build. Like a build. Like to build something. Dots.

Oh, build dots. So two Ds. Yeah. Yeah. So it's got.

I'll make sure that the correct link is in the, in the

description and thanks for your time. And we'll let our AI

finish the show. And that brings us to the end of another episode

of Data Driven, where today we learned that even construction

sites can be smarter than your average smart fridge. Huge thanks

to Amir Berman from Builderts for showing us how computer vision isn't

just for spotting cats on the Internet. It's for keeping billion dollar projects

on track. If your idea of a digital twin was a dodgy sci

fi plotline, well, now you know better. Don't forget

to like, share, subscribe, and maybe send this episode to

the construction manager in your life. Until next time,

stay Data Driven and maybe wear a helmet just in case.