Google Maps for Scheduling (feat. René Morkos)


Are you training? Feta. Randy Collins Construction Brothers

Listen now or check us out on one of these platforms!

Apple Podcasts

Spotify

Google Podcasts

Overcast


SHOW NOTES


We welcome René Morkos of ALICE Technology to talk about the next step in digitizing the construction industry. René inspiration came from seeing the underutilization of construction sites and a desire for it to be done better. He is doing to scheduling what BIM did to design. ALICE is parametrizing scheduling down to three algorithms that finds processes which will be part of the critical path even if the schedule order changes. ALICE, at its core, is beneficial to any site that is constantly in flux… so all of them.


If you have anything we missed, feel free to text us! - 478-221-7009

RELATED LINKS


René’s LinkedIn

ALICE Technologies


Please consider subscribing to our podcast!


SUBSCRIBE!      

Like us on LinkedIn!      

Like us on Facebook!    

Follow us on Instagram!

Eddie's LinkedIn      

Tyler's LinkedIn      

(Our day job)   



TRANSCRIPTION


Tyler: (06:02)

So translating this back, we're talking about automating your scheduling and that's what Rene is specializing in currently. He's got this incredible way of feeding in a CPM, critical path, and pulling it apart and saying, well, if this happens, then take this route. If this happens, then take this route. But it breaks it down into millions of different ways that you can get to the project's completion in as short of an amount of time as possible. And there's flexibility in it. You can push and you pull and you figure out what the best plan is for you and your team.


Eddie:

Yeah, he's given you, the ability to play, to explore is what he calls it.


Tyler:

Exploring is a great way to put it too.


Eddie:

I liked that that's the way that they set this up. It just made me want to go and explore. It made me want to take a project and like take it apart, put it back together, take it apart, put it back together and watch the algorithm work and watch the different options. Because, it's the iterative process of design where one little thing gets broken and then the ripple happens and you're trying to figure all that out. He's covering that. Like I got the ripple. I know what the ripple is and I'm going to cover all that ripple and I'm going to put it back together because this is all parametric. It's not just going to break in that one spot and then I got to figure out where the ripple happened. The ripples covered.


Tyler:

Yeah, so your concrete sub shows up late or your steel subs shows up late. You can account for that. It gives you a roadmap around that and says, all right, well, if that happens, then do this.


Eddie: (07:50)

What if I apply a little more crew here? What about if I have the crew, but I don't have the resources? Like that was the kind of stuff he started breaking down. And I hope that we've done justice to the tool that he has created. And that you guys hear exactly what this means for the industry. This is something that is going to change the way we build. It is going to change the way we build.


Tyler:

We often steer clear of being like, "Oh yeah, this is going to be a game changer" and all that stuff. I hesitate to even say that, but at the same time guys, we see a lot of different apps. We see a lot of different tools out there and we legitimately feel like this one could really, really change the game.


Eddie:

Think about the ramifications of this in comparison to the way that we're doing scheduling. Now think about the capacity you now have to think through problems because of the way that this is solving them.


Tyler: (09:00)

It's an extension of you. It's an extension of your brain in a lot of ways. And it's taking in some of those variables that you forget. If you go to sleep for two seconds, you're going to miss it. It's picking up on that stuff.


Eddie:

It's automation, it's machine learning. It's the artificial intelligence aspect. This is so much we got to get to it, man.


Tyler: (09:24)

Yeah, we do. We do. So here is our conversation and we know that you're going to enjoy this with René Morkos.


Eddie:

All right. Well, Renee, thanks for joining us today, man. Can you tell us who you are and what you do?


René: (09:51)

Hey, Tyler & Eddie. Good to be here. Thanks for inviting me. Name's René Morkos. I am a second generation civil engineer. So, I've been in construction my whole life. My dad would take me to construction sites when I was growing up. When I graduated high school, he gave me a good piece of advice. He said, do anything you want, just don't do civil engineering. I was like, great. I know what I want to do. So, I basically liked to build things. I would actually cut class and go and build stuff.


I would knock on the cool construction projects that are happening in the city and just be like, "Hey, do you guys want an extra hand, I'll work for free? They were like, okay. So I got involved in that and really liked it. Worked my way up from there. Was an assistant site foreman then became a site engineer. My first job was as a project manager in Afghanistan. I would design, build, and procure from scratch my own projects, like from A to Z, which was an incredible experience professionally and personally. And then I ended up doing PhD at Stanford and AI applications for construction. Through the PhD I developed tools first, a generative construction simulator. So that's my story.


Eddie: (10:58)

So what is that called now today as you have iterated it out and made it something more marketable, right?


René:

Yeah. It's called Alice, Alice technologies. Alice is an acronym. It stands for artificial intelligence, construction engineering, that L is key, but we thought it sounded better than ACE. So Alice is the name, Alice technologies. I thought we were going to find something way better. And then it kind of stuck and we were like, I kind of like it.


Eddie:

Yeah no it's catchy. I like it. Easy to remember. It sticks in your head. So what does it do? Like tell us about how you took that experience you had in Afghanistan and kind of applied that forward. What's the problem you solve and the problem you saw at first?


René:

So there's two specific moments I think that really kind of burst the company. The first was, I was in Kabul building landings for F16s. Really exotic location, exotic project. Well the truth is, it's a slop. I mean the actual stuff that we're building, like nobody would blink an eye at it. And I remember my boss saying, "Here's 30 people, go build it." And you know, the sun's rising, freezing cold, our breath in the air and I'm sort of looking at these 30 men and going, well, if I put five people to do rebar over here and then I switched them to do that...And I was like, seems really hard. You know, I'm probably really dumb. I can't really figure this out. I was like, I'm going to go and learn how the pros do it. So I did a masters UFC.


I kept looking for this tool that was gonna solve that problem. I kinda put that idea on the back burner a little bit. The other sort of aha moment was in Amsterdam. I was looking for a topic and we're in a meeting and the project manager and everyone is yelling at each other, cause they're six weeks late. And the subcontractor is saying I can't work any faster. It's 50,000 euros per day, it was a cruise ship terminal in Amsterdam. And I look outside the window and there's a hundred thousand square feet of empty space and six people standing in it. And that's when it hit me. I was like, wait a minute. They're empty. Look at any construction sites, literally just drive down, pick any construction site and it's empty. There'll be like one or two pockets of work and the rest is empty space. And so I started doing research where we would take pictures every 20 minutes and measure the space utilization on construction sites. And we found, and I even like to get people to try to guess what percentage of space is actually used for construction at any given moment, that the number is 3%.


So, 3% of construction sites space is used for construction. So, that's crazy. That is such a tiny number. Anything you do is going to increase that and all I got to show is that increasing space usage is going to reduce duration. Then I got my PhD topic. So, we started looking at that and to increase space usage, we developed an algorithm and that algorithm suddenly solved that big problem that I'd seen in Afghanistan. I started to kind of piece it all together. So, I think that's kind of how the theoretical kind of part of the company came about.


Tyler:

This is an algorithm that's helping us solve for scheduling. You're fleshing out and getting more site utilization by using this algorithm to solve the complexity of schedule. So what are you doing there?


René:

Thanks for sort of setting that up or asking that question. There's a couple of things that are really interesting about construction. There's really a huge misconception, which is that construction is not innovative. The reason that I believe that it's not innovative is not because folks in construction are not hardworking or smart or innovative. It's just that I really, really believe that the problems we solve in our field are way harder than problems solved in some of the other fields. Take finance, for example, it's really easy to digitize finance.


How many dollars do you have in this account? Then you move it to another account. Whereas for us to digitize construction, you first have to digitize our input, which is design. And that's only happened in the last 10 years. And so that's what I always thought, so it's interesting when you said "you guys are developed an automated scheduler." We used to think of it that way, but we sort of realized that it was never about creating a schedule. It was about solving the problem. What's the problem? How do I build it? The schedule is merely one viewpoint of that how do I build it? But there's also an estimate.


There's a video, there are analytics, there's lots of different things that Alice can do that are not possible today. And so what we developed is a parametric or a generative scheduler or simulator. What does that mean? The example I always like to give is from design, because design really is probably sort of 20 years ahead of where construction is. And it's not because it's our fault. I really believe it's because we have to wait for them to digitize design so that we can use that input so we can then digitize construction. The example I always like to give is, imagine you had a cylinder. And on that cylinder, you draw an ellipse and a plane. If you want to increase the size of cylinder, you've got to redraw the object. And if you want to decrease it, you've got to redraw the object every time. Somebody came along and said, "Wait a minute, I'm going to make this tool parametric. You're going to input a height and a radius, those are my parameters and I can change them and the tool redraws that object." That, as you can imagine, is really useful.


Let's say you're designing a car engine, right? And someone says, Hey, in pipe 24, make it half an inch, not one inch. So, you know, back in the day, you have to change all the elevations, all the cross sections, all of that stuff. And I know because I was designing something, three buildings at a compound complex, you know, with guard towers and whatnot, and the boss wanted to move, the staircase which rippled through 23 drawings. With parametric you can just change that parameter and everything kind of... It's the ripple through effect, everything changes for you. And so, as you can imagine to get that done, you've got to really know quite a bit about design.


How is a door modeled in a 2D drawing? How does a 3D object, what are the transforms? It's never been done in construction and deemed impossible, which is what we did. Here's the trick. So there's kind of several reasons why people are like, "It's not doable." And one of them is like, "Oh, construction's too complicated." It's got too many rules or too many variables. Yes, but we have created ALICE and it is three phases. Which is, you set up a rule or set a plan, you send it to the simulator which crunches those rules for you, and then you analyze the results. The setting up of the rule set is something that we invented from scratch. There's nothing else like it in the world. And it's a translator. It allows you to translate your complex construction rules. The exact same reason that people are telling you that it's not doable, it gives you a way to encode that into algorithmic terms. It sounds very abstract.


When people hear that, they're like, "What does that really mean?" It's actually relatively simple rules, like calendars, labor, equipment, resources, consumable, reusable, and crane location. Those are the rules, right? The regular stuff that if you woke up any construction person at three in the morning and said, how do you build a column? And they'd say you put the rebar, put the form work, pour it, cure it, and remove it. That's a rule set. And so we basically gave people a way to encode those rules, put them in a computer and then get the computer to crunch it for you. And the reason that's such a leap forward, is that you now have a way to explicitly model every rule that you're dealing with.


Today's technology, if you look at P6, each arrow is a representation of precedence, of calendar, of workflow, of labor, of equipment. That's a common jumble of constraints that it represents. And it becomes impossible six months down the line ... Can I delete it? Well, yeah, because you added crews, but the precedent is still in effect or, well, I changed the workflow, so that's not affecting anything. It becomes really, really difficult and convoluted. With the theory that we've invented or the approach we've invented, every rule is a specific little box. And you can look at them separately and you'll let the computer crunch them all in unison for you. And the upshot, obviously, is you can explore, which is something that our field is just pretty unaccustomed to doing simply put because the tools have not been available. And you can now explore, you can generate in the course of an afternoon easily 600 million different options.


Tyler: (19:46)

It sounds like there would be millions and billions of rules. I mean, how in the world can I choose from all of these things? It kind of sounds like it would be confusing.


René:

Yeah. That's 100% it. So you, if you were to create a button for every rule, we would need... I'd estimate probably 1200 or 1500 buttons. It's just not doable. You can't expect a human to remember 1500 buttons and you can't expect your software developers to develop 1500 buttons. So it just so happens by a little bit of luck and a little bit of design that I did my PhD on translating complex construction rules into generalized scheduling rules. And it turns out that you can convert all that complexity into three rules, which is mind blowing.


When we started five or six years ago, I kept thinking, "Hey, we have three, right? So like, maybe there's a fourth or there's a fifth." And I thought to myself that if there's seven or eight, that's probably what we could pull off before running out of money. If there's more though, if there's 10, I don't think the problem is crackable with today's technology. And that question is the question that would keep me up. That was my like 3:00 AM when I was like, "Oh my God, are we ever going to pull this off?" That was the one thing that was stressing me. But it turns out that we kept looking for the fourth one, the fifth one. And it was like, no. You can literally translate all of construction complexity into algorithmic terms with three rules, which is kinda nuts. But we've done it and at this point it has been over 18 months where there is not a single constraint we can't put in the system. There isn't anything anyone tells us where we're like, Oh, we can't put that into ALICE. We can tell ALICE to go figure that out for you.


Eddie: (21:49)

So I'm dying to know what are the three,


René:

The first rule is physics, or which tasks have to happen before which tasks, which is encoded in the current technology. That's what CPM, precedent, P6, all of them have that. All of construction thinks in those terms so that's done. The next rule is grouping and splitting. The technical term is changing the batch size. And so when you change the batch size, it's like, is one of my tasks all the columns in floor two, or is it zone A zone B on floor 2? And so you can change the batch size analysis. And the third rule is what we call recipes. And a recipe is how do you build a given element. So a column requires rebar, put the form work, pour the concrete, cure it, remove it. And rebar requires some resources. So recipe is which tasks and which resources. Those are the three rules. We, at this point, spend zero time on the core algorithms. And our engineers, I think probably spend like zero time on the core algorithms. The thing that we spend the most time on now is automation. Make it faster, make it easier, validation. Let me make sure that the input is correct. Integration. I want to input a large 1000 task or something into it. So that's what we're spending our time on. But the core stuff is done. Those are three rules.


Eddie:

You mentioned exploration. And so I want to define that. You mean, and I'm going to try to get words in your mouth, if I want to say, what if I, what if I, then now I can iterate and I can understand what if.


René: (23:40)

In my dissertation, part of the title is visualizing, navigating, and manipulating the construction scheduling state space. So what does that mean? The state space is all the possibilities that are available to you to build a project. Most people don't realize that there's in the order of like half a billion different ways to build something. And sure, you can knock those down and you can say, these ones don't make sense those ones don't make sense. I know that there's going to be two different zoning options there's going to be three different resourcing options... So even if you look at two times three times four, it'll still give you in the tens of options. Let's say that that's the way you want to analyze a problem. Algorithmically, you've got analyze it the other way and reduce it to those tens. But, what I really pushed forth, was the idea that you'd have a spaceship. And that was really it, that we were building you a spaceship and you could sit in that spaceship and it would take you off into space and you could explore all these things that you haven't been able to explore before. There's three pages in an atlas, one of those pages is the explore page. And we've really spent a lot of time thinking about, what do we call it? Is it the navigate? It's actually like, you're exploring. And that was always it. It was always a spaceship where we're going to give you that tool where you can start to explore what's happening and know what is available to you from these options.


And so you could visualize it as lots of little dots. Each dot is a different schedule on this cost time diagram. And if I add a crane, it will reduce duration but increased cost or whatever the trade-off is. But you can now visualize the fact that there's lots of options. That, I thought, was a big step forward because Hey, we shouldn't be thinking in terms of one option, we should be thinking in terms of a portfolio or a state space. Here's the options available to me, let's have a conversation about which ones we should select. And it really is a conversation enabling tool at its essence. And so you can visualize all the options that are available. You can navigate those options, which means you can generate new ones or say I want this one or that one. And finally you can even manipulate the options that are available to you.


For example, you could change the design, right. That completely changes which construction options are available to you. And so that was kind of the thinking, which was like, Hey guys, you know, there's lots of different ways to build it so let's start thinking about all those different ways of building and selecting the best ones. So that was kind of the thinking and it really unlocks this incredibly cool theoretical adventure, because you start to realize that everything in construction today, all the theory, is based on this one option. Like somebody created a P6, like that's the option. Whereas you suddenly start realizing that with algorithms, you can start running analyses on what is the most resilient group of schedules.


Here's a simple example. What's the critical task? But if you change sequence, that task is no longer critical. So what you actually care about is what task is critical across multiple sequences. If I build ABC CBA, BAC, CAB et cetera, which of the tasks are always critical. And you'll notice that that doesn't necessarily line up with one of those sequences. When you start thinking this way, you'll start to see that you start to think of the problem in very similar terms to the way that other fields think of their problems, which I think it's tremendously exciting. Because what it will enable is it'll enable the next generation of smart and hardworking folk to go and start developing the next frontier of things that will sit on top of this. As we slowly digitize this field that I think just unlocks this incredible potential. Finance reached the point where a smart student who knew some algorithms could join Capital One whenever they founded five years after and try a bunch of these really realistic algorithms and be like, Hey look, I'm earning money. And I actually believe that's where we're going to end up in construction. Where people will be able to plug into this ecosystem and apply lots of cutting edge, AI, ML, vision, all of these things, to add optimization to our field.


Tyler:

One of the things that blew my mind when I was doing the research behind this episode was the ability of your software to adapt and the idea of adapting. You had mentioned certain things being on a critical path on your project. What happens if something occurs in that critical path item doesn't happen? Can you adapt to things like that and solve for that problem?


René:

It's 100% exactly what you're saying. And the way that I've started to explain it is think of Google maps. Imagine today you go walking somewhere they go, well, I want to drive from San Francisco to New York. And they pull out a whole bunch of papers and they got maps and they got drawing. They highlighted it. And they've got all the calculations and they've told you how much fuel they're going to burn. So on and so forth. And you're like, okay, well, but what happens if there's a blizzard on I-95 or some highway. Or like, what happens if there's an accident or a fire, or how does your...and there's just no way to do that. Whereas, for us in the Bay area, if I drive South I'm going to take the 280 or the 101 and, I will always put it into Waze and it will tell me, Hey, there's an accident. The 101 tends to be five minutes shorter for me but there's an accident today. Well, I'll take the 280. That's really natural to people that use Google maps, it makes complete sense that it would work that way.


That's exactly what ALICE is doing because it's building it for you. And that's the thing that a lot of people don't realize, it's actually shuffling those resources around and building a project for you. So when something changes, it shuffles around that delay. It's a hundred percent and I mean, for us, as we've gone through the adventure of building it, it makes complete sense. It's how it works in architecture. It's how it works in structural engineering. It's how it works in every other field. When you go to a structural analyst and you say, Hey, I'd like to move this column from here to here. They don't say, Oh yeah, it's going to take me three months and I got to redo the whole thing. They say, hold on, and they go through SAP 2000, they do some adjustments and hit the recalculate button. And that's exactly what's happening with us in construction.


Eddie:

It's incredible to me that what you're doing here is I mean, the way that you talk about parametric modeling and the digitization of design information, how there's a parallel here that I can draw that this is the making of parametric information for scheduling. Like you're doing to scheduling what BIM did to design


René: (30:44)

I would just say, we're doing it for simulating. And the reason that I'm pushing for it is that this took a long time for us get together. When I figured this out, it was like I got hit on the head with a brick because, you know, as a civil engineer, I'm trying to think linearly. So you create, you got the design, you create the estimate, you create a rough schedule, you create a detailed schedule, right. That's how it works. Suddenly, instead of this line, imagine you take that line and bend it into a circle, and that all points into your central model. Your design, your schedule, your estimate, your 4D model, and all those analytics are just different viewpoints of the same construction project.


It is exactly what you said. What Revvit did to design, ALICE does for construction, post-op. You know what's really made me happy is for the first time, and it's been in the last two or three months, we've had people come to us, and I've heard this twice, someone said, I never thought I'd say this, but I actually like scheduling. Which I was like, that's it you know? And it's more than scheduling, right? It's simulating. And I think that the relative importance of construction people in the AEC industry increases, because what you're saying is, as construction folks, when the owner says, "Hey, I want to change the design," the designer can go in and do some tweaks. And it depends on what those changes are, but it can be pretty quick. But the decision latency in construction has been incredibly large for you to create a schedule or for you to create an estimate.


Another thing that I always sort of point out is like, does how you build it affect how much it costs to build. And people think about, and go, yeah, of course. Well, I mean, I'd hope so, otherwise we'll be out of a job. But I'm really interested in is, what happens to my cycle times, what happens to my idle times, the efficiency of resource utilization? And what's really interesting is that I can actually give you theoretical maximums. So if you remember that 3% that I talked about, what you start seeing, and not for space as you can get it as high as 50%, the best, and the worst solutions will...Let's say, you're looking at how much time are my carpenters actually working? And the worst solutions will have like say 54% and the best solutions will have say 63%. That resource cycle time, the reason for utilization, is the difference between the solutions. And absolutely, what ALICE does for construction, is what Revitt did for design. That's exactly it. And it's the exact same use case. It's like, you suddenly start seeing that...and I wondered when we were building it, like is it going to reach a point where like, you don't need a construction person?


It turns out, hell no, the thing is that, that part of the encoding, the rule set, even if you gave me a hundred million dollars today, wouldn't know how to pull that off, which is kind of a cool thing to think about, right? Like a human, with a regular sort of construction salary right out on the field does something that I don't think a hundred million dollars could achieve, which is kind of a cool idea. The thing is that when we get pulled into jobs, we'll send folks over, and even though we have smart, talented people, but they'll sit there and they'll be the notepad, and listen to the construction team. How are you guys thinking about building this project? What is in your head? Give me what's in your head and I'll put it into the machine. And it makes sense. There's so many similarities. When you think about it you're like, of course that's how it would work. You can just code something in the machine, let it do all the crunching, for you interpret the results.


Tyler: (34:35)

I think this is a great time to break off and ask you our megaphone question. So if we gave you a megaphone that the whole industry could hear, and we gave you around 60 seconds, what would you want to say?


René:

What I can tell you is I am absolutely convinced right now that there is a new ecosystem that is coming at us in our field. There is no question in my mind, and this new ecosystem is going to get decided by a smaller number of companies that are currently at the top of their field. It is not a one company game or a one person approach. But the digitization of our field is the biggest thing, in my mind, to happen to construction in the last hundred years. It is going to change everything. The best general contractors today are the ones that know how to identify, analyze and implement cutting edge technologies. And the name of the game is going to be about the creation management of data. The supply chains will get digitized just in time delivery. And I think that's so exciting because it means that the people that are putting the hours in and pushing and striving are the ones that are going to see the reward. If you look at Alice technologies, we reduced the effort required to explore an option by, you know, 10,000 fold.


And as a result we've run $20 billion of construction through the system. You're looking at average of 17% duration savings, 13% on labor and equipment. Those are the kind of numbers that you'll see. And we've seen that over and over again, if you can generate lots of options on the computer, you'll tend to see somewhere around that 20/25% reduction. And I've seen that through the research I've seen at Stanford. It's an incredibly, incredibly exciting time to be in construction. And every single player has a part to play. Whoever you are in this ecosystem, there's just a lot changing. Whether you're developing the technology or figuring out how to integrate it or figuring out how to implement it, there's just a lot going on. It's an incredible time to be in construction. Like I think the best time to be in construction.


Tyler:

That's awesome. Yeah. Well, Renee, thank you so much for joining us today, man.


René:

Yeah. Thanks so much for having me.