Episode Transcript
[00:00:00] Speaker A: This is constructive voices.
Hi, and welcome to building conversations. I'm Steve Randall, and this is a podcast where I speak to the movers and shakers in the global construction industry. We'll talk tech, materials, sustainability and climate change, architecture, site and building management people, and much more. These short episodes are designed as conversation starters, and I encourage you to follow the links in the show notes to find out more. So, let's meet today's guest.
[00:00:41] Speaker B: My name is Rene Marcos. My dad gave me a good piece of advice when I was growing up. When I graduated high school, he said, it's funny, anything you want, just don't do civil engineering. So I went and I did civil engineering. I built around the world. I really like construction. I like coming home with concrete in my hair. I built one year in Afghanistan as a civilian. I did underwater pipelines in Beirut, a cruise ship terminal in Amsterdam. I was involved in all kinds of projects around the world. And I did my PhD at Stanford in AI applications for construction, which turns out was a relatively relevant and useful topic. And then I founded a company called Alice Technologies that developed the world's first AI construction schedule.
[00:01:24] Speaker A: That's a pretty impressive background. I love that visualization of concrete in your hair. I think everyone in the construction industry, certainly on sites, will be able to recognize that one. Rene, let's talk. Before we talk specifically about Alice and some exciting things that you've announced recently, let's talk about AI in the construction industry overall. I mean, I think a lot of people think or traditionally have thought about robotics and how that's being used, but that, of course, is just one side of it. There's the whole bit before you even get on site. There's the whole planning and visualization of everything that AI can do so much faster and, I guess, so much more accurately than has been able to be done before.
[00:02:06] Speaker B: Yeah, I guess a conversation I had with a large governmental agency comes to mind, and they were building, I forget it was 10,000 residential units or something, or 100,000. I forget some crazy number. And they basically were like, well, we don't want AI. We want robots. I remember thinking, well, what do you think the robots run on? To answer your specific question, when you take that kind of compute to an industry that historically has an incredible amount of data that is highly fragmented, and then you bring something like neural networks, which are very good at dealing with incredible amounts of data that are highly fragmented, I think you have an extremely exciting period in construction history, because this.
[00:02:47] Speaker A: Is going to make things so much more streamlined, so much more efficient things will be able to be done quicker. And whenever things can be done quicker and more efficiently, there's a cost saving there, which everybody in business, particularly in construction, loves. When we're talking about such big amounts of money and being able to run lots of different options and looking at how different possibilities could work and tweaking them slightly and having that incredible power of AI. And we don't know just how far that may go, but we can certainly imagine that this is just going to continue to be a game changer for goodness knows how many years to come.
[00:03:27] Speaker B: I mean, take a look at what we're doing with Alice. Imagine you could build it before you build it. Like, think about that for a second. What you do today is you get a contract and you say, hey, I'm going to go build whatever $300 million hospital. And you go out and you actually do it live for the first time. Imagine you went to a concert. This is an example we'd like to give to our students at Stanford. Imagine you go to a concert and the musicians have met for the first time, and they're trying to sort of figure out this concert as you're sitting there. The question to you, right, like, what would you expect to see, Steve?
[00:04:01] Speaker A: They would be trying to learn off each other, who plays in what style, who should come in at what point and trying to work out an arrangement as they go along. But to start with, it would be pretty raw.
[00:04:15] Speaker B: Truly raw. Exactly. I mean, you would not be expecting to hear the fifth symphony, right? And any, any sort of quality that was listenable to. And that's what we do in construction. We meet for the first time and we literally go live. We go build the darn thing, life, for the first time. And so with Alice, you can build it 6 million times before you build it. You can build it with one cranes, with two cranes. The software simulates that for you. Just sort of fascinating. And that's the power. Like that. You were sort of referring to the ability to be able to run millions of differences of ways to build a project. And as we often stated, imagine the cost of exploring an option is zero. Imagine, like, if you asked a question like, hey, what would it take if I resequence. What would it take if I built a crane? What if I added crane? Sorry. What if I tried overtime? What if I tried any of these things? Imagine the cost of asking any one of these questions was zero. That's what you can do with AI. So when you look at that from a perspective, what's going to be available to government entities, to large corporations and so on and so forth. It's crazy. If you took a human and you said, hey, I want you to build a ten story mid rise. I said, okay, then they build it. And then you say, hey, I want you to build it again. Well, they probably got better at it, and I want to build the same thing the third time around. Probably also figure out some tricks. Imagine if you could build a 6 million times, how much efficiency could you unlock that way? And that's just one example from our company and what we do.
[00:05:48] Speaker A: And I guess you could have templates, couldn't you? You could say, okay, I want to build this type of building to these specifications. And on this site we have these parameters. This is the space we have, this is what's around. This is particular specifications for this locality. But actually you could then make a few changes and say, same building, we can adjust it using the AI capabilities and put it in that different location that has slightly different parameters and slightly different requirements.
[00:06:20] Speaker B: I mean, that's what I figured out in my PhD, and that's what we spent the last decade kind of doing, which is, you know, how do you convert construction into constraint language, constraint model? And that's, that's what we did. We, we figured how to do that. You can basically take construction and then apply algorithms to it to do exactly what you're saying. So like, the thing that we figured out at Alice, we're the best in the world at, is how do you tell the software that, hey, my construction project is in London versus Lithuania or South Africa.
How do you impose those changes in a way that the software will understand it, and also that it doesn't take you five years to set it up with Alice.
[00:07:03] Speaker A: You've recently announced Alice Core, and also a new name, Alice Probe, which takes an existing software and is a rebrand, basically. And also there's new integrations and things. And this is really new because this is in the last couple of months. Tell us more about what you've been doing there.
[00:07:18] Speaker B: Yeah, Steve. So I think when you look back at construction, people will think of it as evolving sort of four phases, which is one, construction scheduling was being done with paper and pencil. The second phase was critical path methods. So all the software on the market today, CPM. Third is this 3d BIM based simulator that we built with what we call Alice Pro. And what we did is we sat down, we thought to ourselves, okay, we've learned more about AI algorithms for construction scheduling than most people would ever want to know. So wouldn't it be great if we could take that juice? We could take that power and apply it directly to an existing oracle schedule, Primavera schedule. Wouldn't that be awesome? You could literally import the darn P six and hit optimize. As you can imagine, that's tricky because you're asking to do constraint based transformations or tasks in a task based universe. Long story short, this is exactly what we released. We've been working with better customers for the better part of four months, and we have figured out what I consider sort of the holy grail magic potion, which is we have figured out how to import an existing schedule and optimize the hell out of it on day one. That's what we figured out.
[00:08:37] Speaker A: This is exciting stuff because that integration is something that isn't easy to do, but you found the way to do it. You know, you then unlock all these extra capabilities that your technology, your software can, can bring to that.
[00:08:49] Speaker B: Yeah, absolutely. And what's interesting is that you can do pretty much everything you were able to do with Alice Pro and Alice core. This surprised even, like, even me. I've been working on this for 15 years. I would have originally thought that the capability of core would have been lower because you're ingesting an existing file, you're not setting it up from scratch on a 3d model, that somehow that you'd get less optimization juice. And the answer is no. You can do some stuff in core that you weren't able to do in pro. You can pretty much do all the optimization that we were able to do before with this new system, which is really exciting.
[00:09:29] Speaker A: And what do you see happening next? I mean, I guess you can't tell us too much about what you're working on right now, but some clues, perhaps, to where you see the future of Alice technologies and AI more generally in the construction industry.
[00:09:44] Speaker B: Nobody's got a crystal ball, but here's some predictions that I'm going to make. What you're going to see increasingly is the advent of llms, large language models being applied to our industry. And so the days of you having to sift through 2 million pages of specifications and drawings to find x are over. If you take a look at Rspec IO, founded by a Stanford colleague of mine. So they basically look at lighting. So literally you've got lighting specifications for a skyscraper building and hospital. And literally this thing will not only read those specs, but actually suggest suppliers and individual lights that satisfy those specs. You look at trunk tools. Trunk tools. Also, a Stanford colleague of mine there basically have figured out how to literally read all of the data for your project and then spit out the relevant question. You can ask anything you want, like how many windows on the third floor, or why is there a dip in the concrete slab on this floor? Or whatever it is. So what I'm predicting is that what you will see is the advent of the ability to ingest large amounts of information into large language models and spit out relative answers to those questions. I think the next step is you'll see the combination of old school artificial intelligence, so things that are like engineering solutions, like structural systems, right, SAP 2000 or whatever the UK equivalent is. And what you'll see there is, you will see the combination of llms that feed these mathematical engineering solutions. That's what I think we're going to see in the next short term. Two to four years.
[00:11:28] Speaker A: Yeah. I mean, this is all happening fast and we see it in all aspects of life and in business as well. It's almost that whole thing of AI, being able to give you the solutions before you even know what questions to ask to get to those solutions. And that is. It's mind boggling, but it's exciting.
[00:11:47] Speaker B: Yeah. This is the most exciting period in construction history in the last two millennia. There's no question what's happened in the construction today is what happened to manufacturing in the seventies and eighties. It's really like, yeah, we are going to see in our lifetime construction become a high tech field. I will guarantee that.
[00:12:07] Speaker A: Well, look, Renee, with our new short, punchy format, we've run out of time to talk more. We only really scratched the surface of this. But exciting times for the industry, very exciting times for Alice technology as well. And I'm sure we'll speak again, and I'm sure you'll have some even more mind blowing innovations that you'll have made with your technology.
[00:12:28] Speaker B: There's some really interesting stuff we're doing at Alice. We're actually implementing a 2d based version where you can start scheduling off of 2d drawings. But I'll let that leave that to when we talk next. Thanks so much for inviting me, Steve. Thank you.
[00:12:43] Speaker A: Thanks so much, Renee. Been great speaking to you.
Constructive Voices is building a new podcast network for the construction industry, and you can be part of it. Whether you're an industry thought leader with insights to share or are leading sales for an innovative new solution, we can help you share your message with professionally produced podcasts, including standalone episodes, short series, or ongoing shows. Find out more at constructive dash voices.com.
