Loading summary
Aaron
This episode is with Adam Sadelik of AIM Intelligent Machines. Adam is the founder of aim, a US based company that automates ground engaging machines such as dozers and excavators using AI and machine learning. Now a few years into business, they're scaling across construction and mining operations worldwide. This was a awesome conversation. Adam has a wild story and what they're doing is just remarkable. I'm learning a lot about the new, new technology coming to earthmoving and hopefully you learn something too. So with that, enjoy the episode.
They have. Do you, do you, do you like reading?
Adam Sadelik
Yeah. Especially old books.
Aaron
Old books. Have you ever been to Powell's bookstore in downtown Portland?
Adam Sadelik
No, I have not.
Aaron
It is my favorite bookstore in the entire country because it is, it's a, it's a full city block and it's multiple stories and so you could spend a week in this place. It's unbelievable. And they have every subject, you know, every section on every subject there is. And there's the new books and then there's all sorts of old books. And it's just this really eclectic collection. And then they have like the really old book section with the really, you know, expensive fancy books. But they have, is the only bookstore that I've been to with like an engineering and infrastructure section. And I'll go there. And I go there. You know, I'm there maybe like once every two years. Not that often, especially nowadays. Portland's a little rough, but especially downtown Portland. But I'll go there and I'll have a basket and I'll just buy every book that is remotely interesting. I won't even, because there's just so many there. I won't really spend a whole lot of time on what's what. I'll just buy whatever I think is cool and then I'll ship it, ship it home. And one of them I was looking at the other day, I was like, I should read this. And it's spectacular because it's, it's a little bit on the Hoover Dam, which was built by the six companies.
Adam Sadelik
Yeah.
Aaron
Which Bechtel was involved with and others. And Morrison Knudsen and like some very historic infrastructure contractors. But then the bulk of the book is what these contractors went to do after the Hoover Dam. And like Hoover Dam was really the very beginning of this infrastructure boom led by the Americans around the world. And because the Americans had become like the premier dam and tunnel and highway contractors, pipelines, et cetera, they started to get, then pulled into South America and Australia and up into Afghanistan and Saudi Arabia and it's the Canada. I mean the stories are just unbelievable what they did back in the day. So. And Kaiser is one of them. He was involved in the Hoover Dam. And then it goes into even just like. I didn't really realize that Kaiser and Bechtel built a majority of the Pacific fleet, the maritime fleet, during World War II, because it's like, oh, I guess, like who would have done that? And they were the ones building it all. It's just unbelievable. I didn't know that either. Yeah. Out of California and the West Coast.
Adam Sadelik
So these books are from there.
Aaron
Yeah, so. Yeah, exactly. Yeah. So I mean, they're all over, but yeah. Here's Planning and Design of Airports, second edition. I mean, why would I ever have something like that? There's no reason, but I think it's cool. So here we are.
Adam Sadelik
Yeah, I love these old books.
Aaron
And it's from 1962.
Anyway.
That was my story on. We were talking about the permanente quarry in cement facility. And that was actually.
I feel like they produced something out of there for bombs as well. In World War II.
Adam Sadelik
Yeah. I don't know.
Aaron
Yeah. Anyway.
Before we talk robot dozers, you are from the Czech Republic originally.
Adam Sadelik
That's right.
Aaron
And then you ended up in America from education standpoint, studying artificial intelligence and machine learning.
Adam Sadelik
That's right.
Aaron
But well, before it was the hot topic. Yeah, this was, this was kind of.
Adam Sadelik
The AI, you know, version.
Aaron
Yeah, like 2008.5.
Adam Sadelik
Yeah, 2008 started. Yeah.
Aaron
So what did it look like back then?
Adam Sadelik
Well, it was kind of a special purpose set of models for recognizing certain things in images or understanding speech generating speech. Kind of specialized version of what we have today. But a lot more statistical, kind of a lot more manual.
Engineering went into that to make the models to work.
Aaron
How'd you even get into that to begin with? What drew you in?
Adam Sadelik
Somehow I was always drawn to this general concept of automating, like adding leverage to what humans can do from early age. I don't know why.
I was exposed to construction. My dad had a small construction firm. So I think all these things have influenced me to try to do something that's automated but also physical. Not just something that helps you virtually, but also get stuff done in the physical world.
Aaron
But as you were studying originally, it wasn't anything to do with vehicles or anything like that.
Adam Sadelik
That's right, yeah. It was just statistical machine learning.
Aaron
Okay.
Adam Sadelik
Yeah.
Aaron
A lot of math numbers.
Adam Sadelik
Yes. Yeah, that's the foundation. Yeah.
Aaron
I can't imagine. So you, you get, you have this this background then in early machine learning and then you go apply that for vehicles for quite a while for self driving cars in the. And you were involved in that too in the early days, as it was starting to become a fairly early days.
Adam Sadelik
Yeah.
Aaron
Google X. Yeah, yeah, yeah. So you, you end up at Google doing what now became Waymo. From, from the early days.
Adam Sadelik
From relatively early days. Yeah, yeah, yeah. I joined right after school.
Aaron
Right after school. And then you did that for, you said about 10 years. 11 years.
Adam Sadelik
Yeah, I was at Google X for about 11 years.
Aaron
Okay.
What.
Google.
It'S a very interesting company.
What were like your takeaways from that experience?
Adam Sadelik
Yeah, it was a really interesting life stage.
Aaron
Yeah.
Adam Sadelik
To be there and like see it grow. Also while I was there, the company something like 10x in, in size across, you know, all the metrics including headcount. And so you know, when I joined and when I left, it was basically a different company. Sure, that's, that's how it goes. And so it was a formative experience for me, like being surrounded by people who were smarter than me doing like very ambitious projects, but also kind of seeing the dynamics of how the context changes and so you kind of do similar things but the context around you changes.
We see that for example now with defense, like it used to be like a big no, no for, for tech people to work on defense things but nowadays became more of a norm.
Aaron
Sure.
Adam Sadelik
It's an interesting dynamic.
Aaron
It's changed.
Adam Sadelik
That happens super fast.
Aaron
Yeah, super fast. Just a few years it seems like. Yeah, yeah. Why? What was the appeal for a technology company to get into self driving cars?
Adam Sadelik
I think the original project chauffeur is my understanding. I wasn't there at the time but it started as kind of a visionary idea. So Google was very successful in a number of digital domains and I think there was this feeling like let's also try to do something that's physical and I think the cars became the poster child for that.
Aaron
Sure.
Adam Sadelik
For that vision.
Aaron
Okay.
And then you.
You leave to then go start your own company.
Adam Sadelik
That's right.
Aaron
And this is, it's. We talked about this last night but I just wanted to cover some of the background before we get into aim. But so you, you do self driving cars for a while and then you go into.
What's even the term for it? What were you doing after that? Your first company?
Adam Sadelik
You know infrastructure. Infrastructure. Telecom.
Aaron
Telecom.
Adam Sadelik
Telecom, Yeah, I think that's the closest.
Aaron
And so the root of it is sending information from point A to point B.
Adam Sadelik
That's right. Through a variety of methods. It can be kind of standard comms, like between two towers, for example. But it can be also more unusual where you use the planet itself as a mechanism to transfer the signal. Like super low latency.
Aaron
And that's what you were doing.
Adam Sadelik
Yeah, So I got really interested in those kinds of things. And as it turns out, to build that, you need to move a lot of dirt to adjust the terrain.
Aaron
Yeah.
Adam Sadelik
Build roads, you know, bring utilities and so on to make this happen.
Aaron
Yeah. So the way a lot of information is. Is.
Transferred from, say, California to New York is through fiber optics.
Adam Sadelik
Right.
Aaron
Which is, in theory, the speed of light, but it doesn't actually work out that fast. It's really fast in the grand scheme of things. But there's still a delay, which is latency, correct?
Adam Sadelik
Yeah, that's right. So there's bandwidth and then there's latency. And fiber typically has very good bandwidth. You can send a lot of packets, a lot of data.
Aaron
Sure.
Adam Sadelik
Per second.
Aaron
Yes.
Adam Sadelik
But it tends to have low latency compared to other methods.
Aaron
So it's like when it comes to, like, flowing water, your bandwidth is how much information you can send at any given time, but then your latency is how fast that water's flowing.
Adam Sadelik
Yeah, that's right. That's right.
Aaron
I'm trying to put it in.
Adam Sadelik
That's right. So bandwidth is like the diameter of the pipe.
Aaron
Sure. Exactly.
Adam Sadelik
Yeah.
Aaron
Yeah. And so you figured out. Well. And were there other people doing it this way?
Adam Sadelik
I haven't figured it out. Like, I just actioned it. This principle is well known.
Aaron
Sure.
Adam Sadelik
Like certain radio waves bounce off of the atmosphere in different ways. And if you build it right, then you can transmit signal super fast. In fiber. Well, the speed of light in. In plastic, in the fiber line is lower than in air or vacuum. So that's one source of delays on the fiber line. Second one is that the fiber optic line follow roads. So they kind of zigzag through the continent. It's not a straight shot.
Aaron
Yeah.
Adam Sadelik
And a third one is that you have to have switching costs, so you have to amplify the signal. It's not like one continuous fiber that goes from your home to somebody else's home on the other side. But with this radio wave, you can do it as one hop. And so you cut away all those sources of delay.
Aaron
Yeah. You just go direct to wherever you need to go. And so it's one of the fastest, if not the fastest way.
Adam Sadelik
Yeah. I'm not aware of any other way to do it faster.
Aaron
Okay. And this was for like financial and that kind of thing. Anything that needs very fast information.
Adam Sadelik
That's right. Also government comms.
Aaron
Government.
Adam Sadelik
Those became some of the customers.
Aaron
So while you were doing this, you started to then buy pieces of land and develop them for these arrays. And since the angle of, of it is so important, that meant the earth moving was very important.
Adam Sadelik
That's right. So I became a construction operator by accident.
Aaron
So you end up as an accidental earth moving contractor.
Adam Sadelik
That's right.
Aaron
In the most roundabout way I've ever heard.
Adam Sadelik
Oh, really? Okay.
Aaron
I've never.
Adam Sadelik
I thought it was relatively straight shot. That's.
Aaron
No, no, no. That's. There's, there's, there's nothing straight about this whatsoever.
But the, the earth moving was really important because if the angle was off, it wouldn't work, correct?
Adam Sadelik
Yeah. And you want to do it fast, like once you get the land, you want to get going.
Aaron
Sure. And so what was the frustration with the earth moving?
Adam Sadelik
Well, like, I think what basically any other earth mover has seen. So now I have started to see it firsthand because I start to employ these operators, run these machines, you know, run bigger and bigger projects. And so I started feeling all the pains, like not enough operators who can do this.
Machines get broken by bad operation, not sticking to plan, delays due to weather, low visibility, a long tail of issues that then slow down the projects. And I was really pulling my hair looking at this. Why is it so difficult? From first principles, it's pretty simple. You have a certain landscape, you have machines that can move a lot of dirt. You can just deploy them and make that adjustment. But for all these reasons, it became quite challenging to execute. And I noticed among my peers, other construction operators, that they had exactly the same issues. And so I decided after a while of kind of pulling my hair, to transform that frustration into a solution. And that's how AIM was born.
Aaron
And that was five years ago?
Adam Sadelik
About that.
Aaron
Yeah, about five years ago. And so where did you, where did you start?
Adam Sadelik
I started by pulling some of the machines off of my fleet back then.
Aaron
So you owned equipment at this point?
Adam Sadelik
That's right. That's right. I took some of the dozers and excavators and brought them to the workshop and brought in the original team to start prototyping on these machines and see how attractable it is to turn these vehicles into more autonomous robots.
Aaron
So it was the desire was autonomous to begin with?
Adam Sadelik
Yeah, that was the goal.
Aaron
That was always the goal.
Adam Sadelik
That was the goal because like seeing now both sides, right, Being involved in Robotics and now running these machines in the wild. Like, I could see the gap between. I could see the bridge that I needed to build between those two worlds, and that's really now aim. And so like. Right, and so like understanding the frustrations and the delays and the safety issues, like all of those things, but also being aware of some of the solutions that can be deployed here. And like noticing that there was a lot of activity around autonomous trucks at a time, and there still is. And I think it makes a lot of sense, like those products are successful, but to do ground engagement the way we do it, you know, dozers, excavators, vehicles that actually shape the ground very intentionally, I didn't feel like you can deploy the previous versions of automation and autonomy into this particular application.
Aaron
Can you explain that? Because I hadn't really thought about the difference between the two since we talked about it earlier. To me, it's like automation's automation. It's all kind of the same thing. So you can apply the system that you use for trucks on whatever else you want. It all does the same thing. But your point was. No, it's. They're two completely different. Two completely different applications.
Adam Sadelik
Yeah, that's right. Like, you have to tailor the medicine to the patient, basically. And like the stuff or the technical approaches that are very successful for one task don't necessarily apply to other tasks. And so the difference between the mining trucks and something like a dozer, well, the truck drives.
As an embodied vehicle that drives on the side across GPS paths, waits to get loaded, and then goes to a crusher or to a dumb zone. So it's like a very well defined task. And the map is given and the track knows roughly where it needs to be. So you can enumerate a lot more rules about what the vehicle should do. It doesn't need to be as aware as a human. It doesn't need to have a full range of IQ to do that. But for other tasks like ground engagement, where the terrain is amorphous and the vehicles have really no map because they dig it up anyway. So they have to build and maintain the map themselves. And they have to be very intentional about each action, what kind of reaction it produces in the terrain, like as you are pushing with a dozer, the machine needs to be aware of all the minute differences the blade change will affect on the terrain. So it's about moving the terrain rather than just the vehicle. And that's the principal difference, which, hence.
Aaron
The complexity of it, because you're modifying the environment around you rather than Just navigating the environment that remains the same.
Adam Sadelik
That's right. Think about it this way. For the autonomous haulage truck, imagine the rules you have to write for the truck to do a meaningful set of tasks. It's imaginable. You can enumerate the things and there's a lot of complexity still, but you can prescribe it. But for a dozer, imagine how would you write a recipe for the dozer to do 8 hour shift? How would you describe it? Move the blade by half a degree this way and then tilt it and try to go faster. I don't think you can. I cannot, for example. And so therefore we made a bet in the early days that learning and in particular end to end learning is the solution.
Aaron
Here.
The top question I'm asked is what does Bill Witt do? Our purpose is to build the dirt world's next generation. The dirt world is the companies and people building the critical infrastructure and supporting those who build our critical infrastructure that we need to live the lives that we do. Our business is much bigger than me. I run around the world building our brand. But the business itself does two things. One, we help develop the next generation through our product called billwhit Improve. It's a daily training and development platform at about 300 civil construction companies are using to not just make their people better workers, but better people. And of course we have the 2026 Ariat dirt world Summit. The best opportunity to develop yourself and your teams as, as leaders. So check us out. Billwhit.com book a meeting with us and we'll talk to you soon. So the to to talk just a little bit more about the trucks. The.
So automation in mining has been a thing for quite a while now. The trucks started, I mean they were pretty mainstream by like 2010 maybe deployed in Western Australia I would say. Is that correct? Yeah, yeah. And then now they're deployed in South America, in North America, in Australia, everywhere basically. Everywhere. Yeah. Now and then the Chinese now are starting to deploy it like everywhere with much smaller trucks as well. But the two front runners are.
Komatsu and Caterpillar and both of them have. It's pretty impressive, just the hours and the haulage and I mean they've been doing it for 15 years. So while Google and others have been making all these headlines over the past maybe five years in like the mainstream environment, they've been at it for quite a while. But what you're saying is it's more so it's not simple but it's prescriptive. You can Just tell the truck what you need them to do. What, what the truck needs to do. And then the other thing too about trucks is the, the two points of complexity at either end of the cycle at getting loaded and then dumping, they're managed by humans. You have a human in a dozer telling the truck where to dump over here. Or.
You can, you can actually. You don't have the dozer. The. They'll just create points on the paddocks. Yeah, yeah, yeah. Because it's the same pattern that. But, but you have a dozer there to babysit, you have a person there to babysit. And then the, the loading unit tells the truck where, where to be for every pass. Um, and then between that, that is just between A and B. Just navigate there and just look out for obstacles as you're doing. It's kind of.
Adam Sadelik
Yeah. And there's still a lot of complexity.
Aaron
There is, but still a lot of complexity. I could not automate a truck, mining truck. But it's.
It'S, it's a lot simpler than another machine.
Adam Sadelik
Yeah, there's a different set of different, it's different challenges.
Aaron
Yeah, it's different. So when you were considering autonomy, you were never thinking about trucks. You were always thinking about the other side of things.
Adam Sadelik
Yeah, I've intentionally skipped over the trucks. Like I felt as a market it's fairly saturated. Like you said, you know, Mindstar and Modular have been pretty successful.
Ad
Yeah.
Adam Sadelik
There are other new players coming into that ecosystem. And the overall market is not as big. Not as big because we started a company on this premise that we want to terraform this planet. And terraforming mostly happens through ground engagement. So the machines we are doing now. And so to deliver on that mission, we knew that we had to go to these machines and focus less on the other vehicles like the trucks, because for one, we knew that others have either solved it or very close to solving it.
Aaron
Sure.
Adam Sadelik
And so we can apply that capability through others. But nobody has been seriously working on or deploying the other vehicles, the dozers, the excavators, the shovels. And so we decided to create that kind of capstone complement the trucks too.
Aaron
I think the hang up is that you kind of need to be all or nothing. Like there's a huge capital cost and you need to design the overall operation for autonomous. And so the amount of money that needs to be dedicated to an autonomous truck operation is enormous to actually get it right. And most of the successful ones are like greenfield operations now, like out in Iron Ore, for example. Whereas dozers are kind of working on their own in a lot of applications. So you don't need this huge. And maybe I'm incorrect in my assumptions here, but you don't need this huge capital outlay to start automating some of these other machines. You don't need to go design a whole operation around autonomy. You can. You can start much smaller.
Adam Sadelik
That's right. That's right. And that's how we structured the business in the early days. That's why we started with a retrofit kit that. That is compatible with a wide range of these vehicles. So it required no changes on the fleet, like no new machines, kind of plug and play on existing vehicles, even pretty old ones. And so giving our partners that optionality has always been one of the tenets.
Aaron
When you say terraform, what do you mean by that?
Adam Sadelik
It means to very intentionally adjust the planet in a way that scales with the challenges that our civilization faces. So whether it's lifting up a shoreline so cities don't flood anymore, or building a levee in the right way so there's less flooding, or building a firebreak against wildfires, either preventatively, you do forest management to minimize the risk of wildfire, or when the fire has started, you can still deploy these machines to contain it.
Aaron
It's an interesting thought because I, I did. I read a book years ago now, and it was talking about. His whole argument was like, listen, all right, the climate's changing. That's great. We're good to go. Because we can just modify our environment to adapt to climate change, which is what we've done already for hundreds of years, very successfully. Like, we have people living in deserts already. They shouldn't be in deserts. There's no reason for them to be in deserts. But because we've modified the environment, largely thanks to fossil fuels and energy, we've become much more successful as a human population. And it's like, that's pretty compelling because that's kind of what we've already done. And I think you're making a bet that we need to do more of that in the future. And I'm in agreement there. It's like, I don't see there being less modification to our environment in the future. I only think there's more and more and more even especially as.
Like in the United States, there was a huge amount of modifications made Post World War II in the decades after that sometimes had negative consequences. Like what we were talking about with the Everglades Restoration. Yeah, For. For example. That's one of the examples, is like, well, let's just drain the state of Florida to make everything better from an agricultural standpoint to for people to live here, etc. And then now we're going in the other direction, which is, oh wait, we've drained the whole place. Hence we've started to kill off the Everglades, which is a pretty important ecosystem that we want to keep around. And so now we have to spend tens of billions of dollars on just Earth moving. Essentially. That's all it is. It's just a giant. One of the biggest Earth moving projects in America currently. And for, I mean it's like a 20 year project or whatever it is to then undo, to build these new channels, to build these new levees, to build these new reservoirs, to reconfigure this whole system to kind of get it back to how it was in the first place. It's all just an Earth moving project, which is really interesting.
So the whole terraforming thing, I don't think people talk about it a lot, but it's like.
For there to be a future. That's kind of how you do it.
Adam Sadelik
Yeah, I agree that that's one of the big premises for why we exist, why we are scaling the way we are. And it also formed the past. Like if you look at Netherlands, for example, a big chunk of the country is below the ocean. It was terraformed like very intentionally from the Middle ages to today. Maintaining those structures.
Aaron
Well, in the Netherlands is not alone in that. I mean, you look at any major port facility is probably on reclaimed land, like Rotterdam, like even New Jersey Port of New Jersey, Long Beach, Singapore, you can go down the list of Dubai, you know, the biggest ports in the world, some of the biggest facilities to facilitate global trade. It's all made up Earth. And then major airports are oftentimes on reclaimed. I mean, even San Francisco.
Adam Sadelik
That's right. Or Tokyo, you know.
Aaron
Tokyo. Yeah, yeah, yeah, yeah. You can go down the list. Seoul and, and the list goes on and on. The making new land thing is not a new concept. And I don't think it's going anywhere. It's. It's incredible to watch.
Adam Sadelik
The desire is there.
Aaron
Yeah.
Adam Sadelik
And then people have done their best in manual ways to make it happen. And that's why these are kind of mega projects that take a long time, but they don't have to be considered mega projects anymore. Sure. Like if you can do them five times faster, a lot more safely and so on, then.
Increases the envelope of what we can build as a civilization.
Aaron
So.
All of this is a Great big picture, how do you make a machine go on its own? Like, where do you even start? Were you able to lean on your past experience in some of the things that you knew, as you said you had to hire people as well?
Adam Sadelik
Yeah, yeah, some. I mean, this is a, it's a very high complexity project because you are modifying a big ecosystem and you have to do that responsibly. So you're interfacing with machines that, you know, amazing machines that were made by others, by the big OEMs. Usually you're coming in and trying to turbocharge them with modern software driven AI. And the field of AI has made a huge amount of progress as well, even since AIM was started. And so we're constantly honing our process, how we deploy the technology, both in a physical sense, like how do you actually get the equipment there and install it and retrofit it, how do you support the site to transform from very manual operation to an autonomous operation? So it's a whole process that we have honed over the years, how we do that transformation in collaboration with either the mine site or the construction company or our government partners. So in a nutshell, it involves going in and understanding what their needs and desires are. There's usually a mix of productivity goals and safety goals that they are dreaming about, but without technology it's very hard to achieve them. And so we come in and install the system on the fleet they already have. It's a very straightforward process.
And then we work with them to graduate to an autonomous operation where every former operator is still there. They just have much higher leverage.
Aaron
Before we go into that, from a, a nuts and bolts standpoint on how you make one of these tractors go on its own, it's. You are leveraging the existing information coming off of that tractor. You are tapping into. It's probably has to be somewhat of a newer machine. It has to have a computer.
Adam Sadelik
Not necessarily, not. We have done machines that are 40.
Aaron
Years old, but if they're 40 years old, you have to put sensors and such on the machine.
Adam Sadelik
I thought so too, but we actually are constantly reducing the number of sensors we need.
Aaron
Okay, but it's. So there's. But there's, there is a component of understanding what's happening with the machine itself.
Adam Sadelik
That's right. And that happens mostly through perception. Okay, so like you don't realize so much about like what's happening under the hood on the machine. Just like a human doesn't have like access to the RPM number on the engine, but they can still operate really well, so in a similar way, the system is able to get the gist of what the machine is doing without going too deep into the underlying system.
Aaron
Sure. So you're. That's. That's a major part is understanding what the machine is doing at any given time, and then the other part of it is. Is understanding where the machine is through primarily lidar and then a camera system of some sort.
Adam Sadelik
That's right.
Aaron
And so that's essentially it from a hardware standpoint.
Adam Sadelik
That's it. Yeah, that's it.
Aaron
Okay.
Adam Sadelik
So you plug in a compute box.
Aaron
Sure.
Adam Sadelik
Into the. Into the machine. And like you said, there's a. There's a lidar and then there's a set of cameras. And that's the bulk of the. Of the outfit.
Aaron
Yeah. So computer box, which is kind of like the brain.
Adam Sadelik
Yes.
Aaron
Gathering information, etc. And then all you really have is. Yeah. LIDAR, camera system. And you can put it on any type of tractor.
Adam Sadelik
Yeah, pretty much. Pretty much, yeah. I mean, at this point, we have done all the major OEMs in all.
Aaron
Kinds of different ways, but the difference is between a truck. You are.
You are. There's machine learning at play here, which I think is quite interesting. And so the machine is getting better as time goes on, as it learns. Because like, okay, an AI model, like a. An LLM, for example, you have to train it on data. So you take data, basically the Internet, you give it the Internet, you let it chew through it. I'm painting in a very broad brush here. And then it learns based on all that information. Oh, okay.
Adam Sadelik
And then.
Aaron
And then you train it further through inputs and through, you know, smaller, smaller adjustments as time goes on.
Adam Sadelik
That's right.
Aaron
Additional data.
Adam Sadelik
That's right.
Aaron
And it's the same principle here with the dozer. It's in the early days, you.
Retrofit. And I'm saying all of this so that I, like, I can.
Make sure I understand.
Adam Sadelik
That's really cool how you're describing.
Aaron
Well, because, yeah, my brain does not work this way. So you see, you retrofit the machine and then you have that machine's operator in the early days teach it and just teach it by just running the machine as they would.
Adam Sadelik
As normal.
Aaron
As normal. And as they are, it's gathering data. So, okay, this is good, this is bad, so on and so forth. And then once it's learned up to a certain point, you can then remove the operator from the machine itself and you can then allow the machine to learn on its own because it's at that point good enough to do the job. And then the more it runs, the more hours you accumulate, the more data you're building, which means the better the machine gets over time.
Adam Sadelik
Yep. That is the flow. That is the flow, yeah. And you know, one of the key components there, it's called reinforcement learning. So this ability for the machine to get better based on its own experience. So it takes an action it can grade itself on, how well it did, what was the impact, how much rocket move was it per plan, how much diesel or fuel it has burned. And so it's able to then adjust its own behavior based on that experience. And that's the critical input to making the system perpetually better and go above what humans can do. The human reaction time in the best circumstances, maybe quarter of a second between an impulse happens and a movement happens. So if somebody has a hand on a joystick, there's a pretty significant delay even in the best circumstance to do that. Sure. But the robot doesn't have that constraint. And so it turns out you can then run these machines at fundamentally different clock speed, if you will, that you just run differently. And that's the primary source of value creation.
Aaron
Yeah, we.
We were out at a operation. It was a mindstar system. It was D11s, like five D11s and it was two guys. It was semi autonomous. Two guys. I'm sure you've seen this system where it's two guys in an office at the mine running the five 11s, slot dozing. And people. There's a lot of people that like to chime in on this stuff that don't know what they're talking about because they run a dozer finishing pads over here. They think they're a dozer expert everywhere in every circumstance. But it's like guys like a slot dozing application is pretty repetitive, pretty different. And it's a production based. It's purely production based. You are just trying to move. You have this amount of material that is here and you need it over there as fast as possible while burning as little fuel as possible. That's it. That's kind of the method and to do it safely and being safe. Yeah, yeah. And I was talking. It was the first time I'd seen. It wasn't the first time I'd seen the system. I'd seen it in Wyoming before, but first time I'd been able to ask questions and really get to get to know it a little bit. And one of the things that stuck out with me was these guys. One they liked sitting in an Office much better because you get the shit kicked out of you in a, in a big dozer. And Rock especially, it's just. That's a hard day. Hard day, yep. But two, they said they want to put as little human input into things as possible. They just want the machines to do what the machines do because the machines are much better than they are because it's just consistency. Like you can't be more consistent than a machine and consistency is key to productivity and to managing fuel burn and wear, etc. It's just like the machine is just better and people don't like to admit that that are outside of the world. But when you see it, you're like, yeah, that to me this makes perfect sense. It does do better because again it's.
It'S production based and a human can't be that consistent. One, and then two, they can't be available for that many hours. They can't, they can't at all compete with the machine when it comes from, when it comes from a 24 hour block of time because they can only do 12 hour shifts. And then in that 12 hour shift they've got to eat, they've got to get in and out of the machine, shift change, they've got to go to the bathroom. And so you're not getting your, you're not getting 12 hours of time as well.
Adam Sadelik
Yeah, that's right. That's right. So when we run D11s.
There are a couple of things that come to mind. Like one is to your point, like it's pretty miserable experience to be an operator in a dozer, right. Like you're sitting basically like if folks have a million dozer, right. You sit kind of like at a 45 degree angle because you spend about the same amount of time going backwards as forwards.
Aaron
Exact same.
Adam Sadelik
And so it's like really, really annoying to do that the whole day for like 10 to 12 hours. And so it's not a human condition to be in. It's kind of first principles and then the productivity. Exactly. It's about moving this volume in the most productive way from one place to another. That's really all we do. Like move this volume A, to a position B in the most effective, safe way. And what we are seeing is that why do you move that overburden? Why do you move all of that? To access some kind of material that's down there, that's valuable, that's useful.
Aaron
Yeah.
Adam Sadelik
And so what we are seeing across our D11 fleet, for example, you get to access over $14 million worth of ore each year for each D11. And so if you look at the economics of this, now you have a machine that costs somewhere between 1 and 2 million dollars. Let's say it lasts a number of years in the field, but the value it creates when turbocharged with autonomy is an order of magnitude different from that. And so when I was coming into this from a construction standpoint, I thought like, if we automate this, it's going to be mostly about kind of efficiency of the workforce and the machines doing more per unit of time. And that is the case, but the savings actually are fairly small compared to the top line improvements. And that was the light bulb moment. Like if we can get more of the or out of the ground thanks to this, then that is the business.
Aaron
Yeah. So it's, it's, it's less about removing a person and it's more about just doing more.
Adam Sadelik
Exactly. Getting more output and not putting anybody in harm's way.
Aaron
Sure.
Adam Sadelik
And getting those benefits.
Aaron
And so the first commercial applications that you went after were mining, Correct?
Adam Sadelik
That's right.
Aaron
What? Slot dosing. Which is why I've been talking a lot about it.
Adam Sadelik
That's right.
Aaron
In the early days or even now, when you guys show up to a new operation and say you're going to retrofit two D11s, since that's been the example, we're going to make these two D11s autonomous and there's operators that are running those machines. How do you, how do you approach all that? Sure, you put the stuff on the machine to let it go and then what, how, how do you talk to everybody? How do you set up the team for success, Explain it all to the miners.
Adam Sadelik
Yeah. So there's a whole program that we deploy for the site where the operators participate, the ground staff participates in this and they start running the site in this new mode where nobody's in the cab, nobody's remote controlling the machines, nobody's nearby. The machines are really good at solving the shift, but the operator is there to do the other things that are still required. They might run other support machines, they might refuel the vehicles. We don't have autonomous refueling yet for sure, so there's a lot more work. So the point is that you give the staff a lot of leverage and now with the same bandwidth they used to have, they get to amplify it through automation and get a lot more out of the ground and not hurt anybody.
Aaron
Yeah, I. And so people would think, like from my perspective, I am all for careers in the trades and so people would be like, well, shouldn't you be against automation? The only way I see careers in the trades long term, and the only way I see it as a viable long term career, is the industry becoming more productive. The industry has to become more productive because if you do more, you can make more money, which means you can raise wages, which means in theory everybody's better off. Which has to happen. It has to happen. Wages are too low.
The only way to raise wages is by producing more with the same, the same number of individuals and the same amount of equipment, the same amount of capital expenditures.
And that's why I'm very excited about this. Also.
I'm basing my opinion on real world conversation and anecdotal evidence of me talking to operators at each autonomous operation I've been to now around the world and all of them have been very positive. Because even like the truck example, driving a truck 12 hours at night, Saturday night, 12 hours, and you've got to stay alert the whole time like that is hard. You're just coming off days because most of the time they're swinging you back and forth.
Adam Sadelik
That's right.
Aaron
So you're not like a nurse. You don't just stay on nights and.
Adam Sadelik
Your family is like four hours away.
Aaron
Families four hours away. It's, it's hard, it's, I mean physically, that's really, really hard to be swinging back and forth every few days, it's, it's tough. And even, even every other week, I mean that's still tough, it's still inhuman.
So that's, that's one like minds have to be a 24 hour operation because it's production based. So you need humans right now to run at night. But that is a world in which not many people enjoy nearly as much as days because again, it's, you're working against your, your humanity, your body.
Ad
The Uniswap wallet makes it easier and safer to own and use crypto. Created by pioneers of the crypto economy, the Uniswap protocol has powered over $3 trillion in trading volume and it's trusted by tens of millions worldwide. With the Uniswap Wallet, you can discover, swap and manage your crypto all from your phone. Buy your first crypto assets in just a few taps and start exploring the freedom of decentralized finance. With Uniswap, tap the banner to get started.
Aaron
So there's that and then two, there's a lot of jobs that are very repetitive, that are not all that dynamic and like some people might just love driving A truck. And there will always be operations that need people to drive trucks. A lot of people don't though. And because these mining companies need people, they're talking up some of these careers, quite a few of them, making them seem like they're just absolutely incredible. You get to drive this amazing gigantic truck and it's like it's fun for.
Adam Sadelik
The first 20 minutes.
Aaron
Exactly. Yeah, yeah. It's fun for a little bit. Like how many, like how. Wait, how many hours Mr. Mining Executive have you spent driving a truck? Like get in the buddy seat for 12 hours and then we'll talk about it. Just do one shift. Just do one. They probably haven't even done that.
It's not all that dynamic. It's pretty mind numbing. And so by letting machines do some of that more mind numbing, physically abusive work either at night or in rock, for example, going back to the dozer example, you can allow humans to go do more, to add, to add, create greater value within the value chain in different ways now. Which I think is great. And then also I would be like, well we don't need autonomy because no, everybody is staffed up and everybody is well trained and everybody is as productive as possible. We don't need machines. But it's like that's not at all the case.
Adam Sadelik
Yeah, yeah. And we have a perfect safety record.
Aaron
And we have a perfect safety record.
Adam Sadelik
Yeah.
Aaron
And we're not breaking machines, you know, so on and so forth. It's like that's not happening anywhere. So to me it's like if we had a whole workforce banging the door down. Yeah, maybe let's slow this down a little bit. But we don't. So we're. And if we don't, we're jeopardizing our ability to.
Produce the critical resources and build the infrastructure that humanity needs long term. And then second, if we don't get better at what we're doing, more productive, we can't make this sustainable career for the people that are still necessary well into the future. And so it's like a double whammy. And so for me it's like we need something, somewhere to start making this better. Because the way I see it done right now, it's just not, it's not feasible in this conventional, conventional method.
Adam Sadelik
I couldn't agree more. And we see these patterns already of changing across the world. So now we have production operations around the planet across big mining, construction and work with the government and every single place. Irrespective of the location or the details. We see these patterns. They Always do more with the staff. Plus or times autonomy that we allow, we bring to them. And so they start to build more, they start to mine more. So it's not a subtraction of somehow like labor minus AI, but it's rather a multiplication where. To your point. Exactly. You have the same experts on the side. The ground staff is most informed about what needs to happen, but they don't particularly want to be or should be in these environments. And so you create a motion where you add this leverage factor where now the mind side gets turbocharged. Sure. Or a construction site. We're helping to build a data center, for example, a lot of earthworks. So you turbocharged those operations. They still, all those people still have meaningful things to do, but at a higher level. And then robots does the dirty work too, so to speak.
Aaron
Well, and you know, for anybody that wants a sustainable future, they should be. They should be very interested in the most reasonably priced commodities possible. The only way you do that is by producing the commodities more efficiently.
Adam Sadelik
That's right. And like everything comes out of the ground.
Aaron
Everything.
Adam Sadelik
Almost everything. Maybe some stuff comes from asteroids, but like, for practical purposes, everything is out of the ground. Like this Mike. Right. Like your glasses. My shirt.
Aaron
No, Every. Every. Everything is either. I mean, there's some forestry products, but very small, majority of materials. It's either from the ground in. It's either mined or it's oil. That's it.
Adam Sadelik
That's right. It's grown. Right. Like, but for growing or you also need a lot of earthworks to actually do that.
Aaron
You can't grow anything without mining. Yeah. Because. Because you wouldn't have water. Because water is transported through infrastructure, largely created by steel and.
Adam Sadelik
That's right. And that needs to be.
Aaron
Yeah. Powered by, you know, electricity, etc. Yeah, yeah. So it all has to be mined. And then fertilizer. You. We can't grow anything without fertilizer. All fertilizer is produced by, again, stuff from the ground.
Adam Sadelik
That's right. That's right. And we're helping to mine phosphate, for example.
Aaron
Phosphates. A great example.
Adam Sadelik
Like stockpile and mine it. And to make the field itself, the agricultural field, you need to grade it.
Aaron
Yeah.
Adam Sadelik
Remove the rock, make sure it doesn't get flooded, make sure it doesn't get burned by a wildfire. So it's kind of a recursive situation where the same machines are necessary across the board. We. When you.
Aaron
So going back to rolling this out to a new operation, what's the conversation like with the operators there?
Adam Sadelik
Well, it's Very simple is that, you know, there's a new technology that allows you to have higher leverage in everything you have been doing. And you won't be exposed anymore to the vibration, the noise, all the nasty things that happen in the cap of these machines. And so you will have this tremendous amount of added output because you will have many machines under your management. And so you still get that satisfaction. The part of the job that is satisfying is that they look back and like I dug up this giant hole and it's going to house something meaningful or it's going to get more iron ore out of the ground to make more steel. Right. And so that part remains. So they get to do all of that still. But with this gearbox of lot more power.
Aaron
And they're pretty receptive to that.
Adam Sadelik
That's what we have seen across the board. Like once, once we sit down and explain it and, and talk to them, you know, as humans, then it's very straightforward.
Aaron
Yeah, I have. You had situations in which there is resistance.
Adam Sadelik
I haven't seen any resistance. There's resistance sometimes kind of at the outset as you walk into the door because there is that reaction like these robots are going to destroy everybody. That's kind of a doomsday default sometimes. But then as you start to explain it and as you start looking at the first principles like there is shortage of operators. The machines everybody knows are not used to their fullest potential.
At the same time everybody wants more product, more of these commodities. High pressure environment.
Once everybody understands that there is this dichotomy of relatively low supply but high demand and autonomy is one powerful tool to level that you up the supply to meet demand.
That's when the conversation becomes very productive.
Aaron
It's I think unfortunately, business and capitalism's been hijacked by the evil empire in a lot of ways.
And it's been made into something because it has been in a lot of cases, especially in the States, something that is just extraction based. We just want to extract as much money as possible. And I don't know what it is about money. People just, let's just stack it up more and more and more. It's like how many billions do you need? What are you going to do with all that?
Let's just extract. But in a business that's thoughtful. The more money you make, the more profitable you are. The more you have to then reinvest in your business and reinvest in your people and reinvest in your equipment and technology to then make your operation better, which then makes you more profitable. Which then gives you more to reinvest. And like, that's true at our company and true at your company. And a lot of companies out there is, hey, if we can do this better, that makes you better, that makes you better off because it makes this operation more sustainable. Because now we can. Now big picture, we're a little bit more, if we can produce at a lower cost, we're a little bit more insulated from where the commodity market goes. And most of these are commodities that are dictated to buy a global market. And so if I can produce, you know, just, just dumb numbers, you know, iron ore for $50 a ton, and it's $100 a ton if it goes to 60. And my competitors are producing it at 70 now, they're losing. But I'm still, I'm still down here. I, I can still have this operation go. So my workforce is, is more secure. But then if I'm, if, if it is at $100 a ton, say, and I'm making that, you know, huge sum of money, sure, some of that's going to go to shareholders. I don't want to be too big picture here. Yeah. Too naive. But that also does give me more capital to go reinvest in my existing operations and then to explore future operations to ensure then we still have, you know, a future as well. There's other ore bodies, there's other mining operations, there's. There's other places that we can go here as well. Which I again, is, I think, in everybody's best interest in, In a perfect world.
Adam Sadelik
Yeah, I think you're right. It gives you power to create a more enduring organization.
Aaron
Yes.
Adam Sadelik
More enduring business, maybe More responsible business.
Aaron
Yes.
Adam Sadelik
I think a lot of focus has been historically on savings, you know, like, can we shave off like 1% on this particular line item in the OPEX budget? Like, can we get like a slightly better blade that does that, or can we run like 10 minutes longer and do these small adjustments? But what we are finding is that if you are running at high level of autonomy, the top line completely dominates any of the savings that you might otherwise chase. So if you get more gold out of the ground per second, if you get more iron ore out of the ground, that gives you a tremendous amount of power because now you get to choose your destiny. Like maybe you just open the mine site for three months out of the year and produce the same output as you would have in the manual world. And so you can you get to dictate how you throttle production and just control the operation much Better, because right.
Aaron
Now.
Like, every mine is kind of stuck to this one model that is, you need people to dig stuff out of the ground and to process it. And it's kind of done the same way everywhere, as far as I've seen it. And every commodity is kind of the same in the grand scheme of things. Like the. The way they dig it out looks a little different here there. The processing looks a little different here there. But it's all kind of the same thing. Iron is pretty similar to copper. Copper is pretty similar to coal. Coal is pretty similar to gold. Like, there's not a lot of variation.
Adam Sadelik
That's right. The color of the soil changes, but otherwise it's very similar.
Aaron
It's very similar. And it's all kind of just this one traditional model because that's all that's available to them right now.
Adam Sadelik
Yeah.
Aaron
And again, I'm painting with a very broad brush.
Adam Sadelik
Yeah, very broad. Yeah. And furthermore, like, it hasn't really changed over time.
Aaron
No, it hasn't.
Adam Sadelik
Like a linear, shallow, linear trend is what we see.
Aaron
Yeah. And this is where.
Yeah. When it comes to means and methods, this is where I start to go into the whole camp of like, well, you just have a few major equipment manufacturers, and they're also not incentivized to create anything all that. All that different because it's working with that existing model, and it's going to take a lot of money to create anything different. And they're really just incentivized to whatever the next quarter is, whatever the next year is, let's just keep. Keep the train on the tracks. And so the amount of innovation, even from an equipment standpoint, I think has gone way down over the past few decades compared to what it was in even just 70s, 80s. Like, you look at. I have a book out there. You look at letourno, some of the stuff that that guy was making for mining, it's just unbelievable. Trying all kinds of stuff. And it's like, all right, we could either just sit here and say, well, we just have the best stuff. You can't get any better than this. Or we can sit here and say, well, maybe we can, but the incentives just aren't quite there to get better.
Adam Sadelik
Yeah, yeah. I have a lot of respect for the OEMs. You know what they have done. The machines are just tremendous across. Across the board. And so, like, we have no ambition to try to create, like, better machines in a physical sense.
Aaron
No.
Adam Sadelik
But I think there's a huge opportunity in basically injecting strong software.
Aaron
Well, and Maybe that's the opportunity. Like, maybe a dozer is just a dozer. I don't know. I'm not smart enough. But maybe you can make the dozer a lot better.
Adam Sadelik
Yeah, I don't know either. I don't know either. But what we do know is that there's more millions of these machines deployed already. Most of them are manual. And we know that the ones that become autonomous have a lot more yield, a lot more safety, a lot more productivity.
Aaron
So as the, as the machines run, they, they, they get better.
Adam Sadelik
That's right. Through their own experience.
Aaron
So they're more and more productive as time goes on.
Adam Sadelik
That's right.
Aaron
That's what we are seeing, which is extraordinary.
Adam Sadelik
And then, you know, they, like I observed that when I was employing operators, like, by the time they get really good, they are retiring for one reason or another. So, like, you have a person that's, you know, getting better and better in their skill level, and then when they're at their peak, they basically disappear from the marketplace. Sure, it's usually a combination of some injury or just getting tired of the job, but with this kind of system, it compounds, like, so it doesn't forget what it learned before. Just compounds, compounds and never retires. And so that's how you get these curves.
Aaron
And then within an operation, say there's two or 10 dozers, you automate two of them, you're building the data set with those two. And then when you want to automate numbers three and four, you already have that information to apply. You're not starting over with three and four. You're not having to retrain three or four.
Adam Sadelik
That's right. Like, once you have one machine of a particular type, it becomes very simple to generalize it.
Aaron
Sure. And in theory, then the more machines you have automated at a site, the more data you're building, the more of these machines are learning. And if this machine over here learns something, then does number two already know that lesson now?
Adam Sadelik
Yeah, that's right.
Aaron
Okay.
Adam Sadelik
That's right.
Aaron
So you're, you're. The more you have, the faster you're learning.
Adam Sadelik
Yeah, that's one of the beauties of kind of digital AI that we have. Imagine if I could tap you on your shoulder and convey to you all my memories and you to me. And we can do it in one millisecond words of touch, tap on the shoulder. That's what's possible with digital AI, and that's why it consumes so much energy, because you have to run transistors at high power to have this high fidelity digital signal. But what you're getting for that price is that this repeatability and transferability of the models.
Aaron
But this is, and this is where to another point that is, I don't think the whole AI thing is sustainable without us producing energy in a more efficient way. And a lot of this is energy is in the physical world. It will always be. And it all comes down to mining at the end of the day and materials. And so we have to become more efficient at producing energy and producing materials if we do want this stuff to be at all sustainable. Because right now I just don't think it maths from an energy consumption standpoint.
Are you afraid of these machines taking over?
Adam Sadelik
Not particular, not particularly. They, they're, they're, you know, they're very good at what they do, but otherwise they are like extremely naive.
Aaron
Yeah.
Adam Sadelik
And so we chose to automate these ecosystems because they're very important. Mining, construction, defense. But we do it, you know, with a tool that's, that's really good at that particular set of tasks to move dirt in all kinds of ways across the site. But it doesn't have any other ambition.
Aaron
Right.
Adam Sadelik
So it's like, like blindfolded to do that.
Aaron
Sure.
Adam Sadelik
And it's really good at that. But I don't see a way how it can just like get ideas, get more ambition.
Aaron
Yeah, yeah, I'm sure you're asked that. I'm sure people, especially operators.
Adam Sadelik
Yeah. I mean, these are pretty intimidating machines and they learn new behaviors that humans haven't shown it. So there is this emergent aspect to it. It's like, well, is this still our dozer? Because now it's doing the operation quite differently. Like is deciding the ripping cycle differently. Like, learn new way to micromanage the blade. I see you have the CAD performance handbook there. Like it basically learned those rules, but through experience and then continues to build upon that.
Aaron
Does that hurt people's feelings?
Adam Sadelik
I haven't seen that.
Aaron
When the machine starts to do much better.
Adam Sadelik
Yeah, I mean there's a lot of pride in this industry, but since the operators are still involved in managing the side, they really get to enjoy the added bandwidth. It's like, wow, now these robots are like under my management. So I get the credit for that. Sure. And they have learned some of these tricks from me and invented their own tricks how to do this better. And so there's more of a compounded pride rather than a competitive pride.
Aaron
It's.
It'S all just, it's very interesting. This is so far beyond my world. And it's just what you said too is it's not just dozers, you're looking at some other applications.
Adam Sadelik
Yeah, of course we're doing other types of machines. They're all hydraulic machines for the most part. But you have dozers of different sizes and shapes and different OEMs. Same for excavators, shovels and diggers. Different sizes and shapes and also loaders of different types. So basically all the major ground engagement vehicles that actually dig holes, load trucks, manage the environment in a very physical way.
Aaron
And these are primarily regardless of the machine production type environments. Like if an ex, an excavator is loading trucks, a loader is loading trucks or shifting this over here. Yeah. So it's all production based.
Adam Sadelik
That's right. That's right. We have proving grounds next to our headquarters in Washington State. That's where we developed the bleeding edge of this. But then we have production sites around the world now.
Aaron
Okay. Is it challenging to be around the world with different cultures?
Adam Sadelik
Well, it's challenging, but you also get a lot of interesting output out of that. So nature is fair in that sense is that if you are against a lot of challenge and complexity, but if you tame it and solve it then it's a lot more enjoyable. It's really cool to look back and see how many tons we have moved and the types of sites we have improved and that's really satisfying. I cannot imagine running a purely digital company where you look at a dashboard and that's all you have.
Aaron
Sure.
Adam Sadelik
To me personally it's important to have that physical, embodied aspect of the business.
Aaron
It is interesting too. I think someone asked me this the other day, what are the differences between operations all over the world? And I flipped it on its head. Then I actually said what's been most surprising to me is how similar they all are. Like most operations are very similar. Sure. You have these cultural.
Cultural quirks. You know, an American operator is very different from a Japanese operator. Very different from a Saudi operator, Pakistani operator. But.
It'S all kind of the same at the end of the day too.
Adam Sadelik
Yeah, yeah.
Aaron
They're all pushing dirt. They're all proud of what they do. They're manipulating the environment. They're using the same machine. I mean it's a D9 or Komatsu Dozer or whatever it is. It's the exact same machine here and here and here. You know, the ground conditions might be different but otherwise it's all pretty similar. And I haven't seen everything. I never will see everything. But I've Seen like I have enough of a sample size now to be like, it's actually amazing how similar it is across the board. So once you figure it out someplace, you can kind of apply it in other, in other locations, I think a little easier than people think you might be able to.
Adam Sadelik
Yep, yep, that's right. And I think that's partly an artifact of the fact that there's been fairly small amount of innovation in these ecosystems compared to what has happened anywhere else. Basically like mining construction have seen very little of the, of the transformation through technology that has happened elsewhere. Even in medicine, even fairly conservative fields have taken a lot more advantage of tech in general.
Aaron
Why is that?
Adam Sadelik
I really don't know the answer. But that's what we are changing now. Doing it in a way that plays well with the ecosystem so that they have optionality to adopt this type of change management and automation. And what we are seeing is that yes, you visit these sites and if you squint your eyes, they're running roughly the same way.
That's because they all have the access to the same tools and playbooks.
So now we are offering this fundamentally different way of running the site at high leverage. And then if you do that comparison, if you take the average of manual sites and compare to the average of the autonomous sites, there you see a big change. So for example, if you run the machines a lot more productively now you need fewer machines to get the same output, so you get these options to trade off. Or if the fuel is too expensive on a particular week or a season, you can throttle down production because you know you will make up for it later on in the year. So it gives you like a lot more knobs how to manage the site in a new way that brings with it a lot of power.
Aaron
How did you explain all this when you were raising money to people from the outside that aren't mining people, aren't earth moving people. How did you explain it? To quote unquote, like a common folk.
Adam Sadelik
Well, it's really first principles, like the dichotomy between the fact that there's just not enough people in these ecosystems. The ones who are employed in these ecosystems are not particularly safe or happy. And at the same time everybody expects a lot more from the ecosystem than it's doing.
Aaron
Which is your point about happy is I think key. A lot of them are not happy.
Adam Sadelik
Yeah, and I wouldn't be happy either. That's partly why we're trying to solve for that.
Aaron
And I don't blame them. Yeah, exactly.
First principles.
Explain first Principles.
Adam Sadelik
So first principles is just the basic, basic facts of the life. Right. So you have a situation where everybody wants to build more infrastructure, get more product, whether it's phone or car, anything. But at the same time, the industries that are supposed to supply that supply for that demand are stretched to their limits.
Aaron
Exactly.
Adam Sadelik
And so that's the disconnect where the supply is low, the demand and expectation is very high. And the only practical way I can see to solve for that is to turbocharge those sites with, with robotics.
Aaron
Sure.
Yeah. And I didn't think I'd be sitting here agreeing with you.
At this point, but based on what I've seen, again, I think that is, I don't think it's like the magic bullet, because I've seen it done terribly wrong too.
Just because you have autonomy does not mean it's going to be successful.
But I think it's a big piece of the equation. I think it's a really big piece of the equation. And I think it doesn't like at face value, it goes against again what we're working on, which is the development of human beings and of leaders within the industry. But it works hand in hand with both of those things. And again, I think the industry has to become more productive, it has to do more with less labor hours to succeed long term. I don't see another way of, I don't see another way of it working out. And we just, we mathematically, like, especially a place like Japan, mathematically we just have less people to do more work. And so mathematically it doesn't work.
Adam Sadelik
That's right.
Aaron
A fifth grader, a five year old could pencil it out. Your, your kids could be like, oh, this doesn't matter. Yeah, two plus two doesn't equal seven.
Adam Sadelik
That's right. So those are the first principles also. Sure. Right. You have this fact, what is the stat like half a million workers are right now missing in construction just in.
Aaron
The U.S. just in the U.S. yeah.
Adam Sadelik
And at the same time you have like 99 people dying every hour in construction in the U.S. yeah. And so that's another example of a big dichotomy.
Aaron
Yeah, I, you'll hear me talk about it in a few days. I, you know, I start my talk.
Ad
This message may be shocking to many millennials. If you are one, you might want to sit down right now. Loads of people are searching the following on low rise jeans, halter top, velour, tracksuit, puka shell necklace, disc belt. You likely place these in the dark of your closet in 2004. Never to be seen again. But if you can find it in yourself to dust them off, there are a lot of people who will give you money for them. Sell on depop where taste recognizes taste.
Aaron
Kind of just saying guys. And I pointed out with undisputed facts that is we're not healthy right now. Like this industry is not healthy. And I, I am the number one fan of construction, of earth moving, of mining. I am, I am so committed to these industries.
But we also need to be honest with ourselves. And so I use the example of okay, we're going to talk workforce development and let's go to a mine site. Since this is a mining conversation, we'll go to a mine site or you can go to a construction site at shift change or start of shift and they're not going to know we're there. I am with a group of parents and they're considering where to send their kid for the rest of their life. Their 15 year old, 16 year old, 17 year old, 18 year old, son or daughter. And it's my job to talk them into this being the best place for their son or daughter. And we show up, they don't know we're there. We're looking out over everybody. What are their expressions? Are they happy?
What do they look like from a rest standpoint? Are they well rested physically speaking? This whole industry runs on the physical well being of individuals. They talk to me, they lecture about safety all day long, safety this, safety that. And you look at everybody physically, is everybody healthy?
Are those parents going to believe me that this is the best place for their child?
Adam Sadelik
It's a hard sell.
Aaron
It's a really hard sell man. And it doesn't, it doesn't have to be that way. And it can't be that way. If there is a future in my opinion, we've got to fix this. And that's why I'm so passionate about the subject is because we. Part of the solution is.
Reducing, reducing man hours, reducing labor to produce more, to, to do more with less, one and again to raise wages. There's just, there's no other way to do it.
Adam Sadelik
Yep.
Aaron
It doesn't matter otherwise.
Adam Sadelik
I'm excited about this for the same reasons. For the opportunity to have at least some role in, in improving that. Yeah, that's how I got into this. Is that like looking at, okay, you can look at, you know, autonomous cars and that's a really compelling problem. But is that really the biggest problem that we have in the world? Like is that the most fundamental problem? And so if you look at something like the Maslow's pyramid of needs, like food and shelter.
Aaron
I love the Maslows is at the bottom.
Adam Sadelik
Right. It's true. I would argue that the basement of the pyramid, or the foundation of it is Earth moving. Because if you don't have Earth moving, that's working well, you can forget about food, you can forget about shelter. Right. You can have those things at small scale, kind of, you know, ancient times, but at the level that everybody expects now. It's impossible to do any of those things. Anything in that pyramid is impossible.
Aaron
Sure.
Adam Sadelik
Without Earth moving. And it was that realization that really gave me the energy and the ambition to do this once.
Aaron
And once you see it, you can unsee it. Yeah. Once you recognize how fragile society is and just how fundamental earthmoving is to the whole world. Like, again, I. There's a lot of days where I wish I was naive to that fact. Like everybody else. I wish I just had my head in the sand and I could just go about my life and just not care about the global economy and consumerism and energy and transportation, etc. But then. And you've seen it too. Once you start to see how the world actually works now, you know, man. And. And you start to see how fragile it is. Like it just has to break over here.
Adam Sadelik
Yeah.
Aaron
To just unwind the whole system.
Adam Sadelik
Yeah.
Aaron
And that can't happen. It can't. It can't work.
You. You're five years in. You're starting to grow.
More so than before.
Adam Sadelik
Well, we're on the exponential trend now.
Aaron
You're on the exponential trend.
Adam Sadelik
Yeah.
Aaron
Yeah. How's. How's that been. Because I know recently you had the announcement, you raised money, and then that goes everywhere. And I'm sure people are, you know, looking at you this and this and that. How's that been after. Now that, like I think you were saying, what was it before? You were in, like.
Not quiet mode.
Adam Sadelik
Stealth mode.
Aaron
Stealth mode. Yes. Yeah, yeah, yeah. So before you said. And I think when I first started talking to you, you're in stealth mode now.
That's not the case.
Adam Sadelik
We're coming out of stealth.
Aaron
Yeah.
Adam Sadelik
There's still a lot of things we cannot talk about because we are affecting these giant commodity suppliers.
Aaron
Sure.
Adam Sadelik
And if you know the production numbers, you can kind of work out the financials, and that's very tricky. But we have been more vocal about that this now exists and people are buying it, and we are running in production across these big ecosystems that we always wanted to automate.
Aaron
But how is that how's the conversation been now that it's changed a little bit?
Adam Sadelik
Well, very positive. I think also the timing is fortuitous is that these days there is a lot of optimism about automation in general. Like if you rewind, you know, six, seven years, like it was very rare still to see actually working robot around you, whether a car or some other robot.
Aaron
Yeah.
Adam Sadelik
And now that has changed pretty quickly. And so that takes with it the whole industry. And so if we go back to the dichotomy of the low supply, high demand, it as a package now it starts to make self consistent sense where you this is actually possible. These methods are powerful not just as a research paper but as an actual product that that can be counted on. And there's a long track record that we have in terms of safety and productivity that goes with this. And so you know, the last.
The recent times have been very powerful in scaling the company because now we have customers coming to us.
Aaron
Sure.
Adam Sadelik
Asking can we buy this? Because we have seen it next door, we have seen it used elsewhere.
We want to adopt this.
Aaron
Is there still though a hesitancy to that? Because like we were talking about, this is a tough world to sell into.
Adam Sadelik
Yeah. I mean of course this is not like you know, let's install it today.
Aaron
Sure.
Adam Sadelik
And we wouldn't do that even if they wanted to. There's a process that we go through and so we align on the objectives. We understand their side, we map into their world. Only then we start to actually set up the environment for this.
Aaron
Are there some applications that you wouldn't sell into right now?
Adam Sadelik
Yeah, like we don't do, we don't do a lot of things actually. It's, it's running a startup. It's more about, you know, saying a lot of no than saying yes.
Aaron
Yeah.
Adam Sadelik
And so we focus on ground engagement. We don't do the trucks for example, we don't do weapons.
We work on things that are aligned with the, with the ultimate terraforming vision. So we always work backwards from that. Like what are the capabilities and the partners that we want to have for an enduring company that can terraform this planet and maybe other planets down the road.
Aaron
What's been, what's been surprising to you as you've dug in more here, as you've gotten to know the industry?
Adam Sadelik
Well, the, we touched on this earlier but really this, this fact that these are like extreme, these are non negotiable ecosystems that need to survive and thrive. And while that's a fact, it's also a fact that they haven't really driven that much recently, and they haven't really changed much. And so that the more I learn about those things, the more surprising it is that you have this, like, thing that everybody depends on moving dirt. And it sounds pretty unglamorous, but it, like, really is the foundation for everything.
Aaron
Well, and. And see, that's where, you know, I listen to these.
Interviews with guys talking about asteroid mining or, you know, this or that, the sexy stuff, the space travel, etc. And maybe that is the future. I don't know.
But I just, in a way, I chuckle a little bit. Or even AI data centers. Like, I can't help but think, all right, you want all these data centers, that's great, but you've got to build them. How are you going to build them? 1, and then you've got to power them. How are you going to power them? I. And until you do those two things, all the other stuff is irrelevant. And it is. It is funny too, how, like, in a place like San Francisco, there's probably not a whole lot of conversations about how important Earth moving is to AI.
But without Earth moving, there is no AI, there is no technology, there is no space exploration. There's none of that.
Adam Sadelik
It is. It's foundational.
Aaron
It is the foundation. And it's completely ignored in all these circles, which is. Is. Is frustrating in one sense, but then is also the opportunity in the other sense. It's like that's. That's the whole game we're playing at the same time.
Adam Sadelik
And that's one of the reasons why I have so much respect for you, because you give those industries a voice.
Aaron
Yeah.
Adam Sadelik
That might otherwise not be heard.
Aaron
Well. And it's like, I've. Man, if we were sitting around waiting for someone else to do it, it's like the industries themselves are not doing a very good job and they know it. Like, they're not. And. And that's what I'm trying to tell them is like, listen, if you guys are.
We can do our best, but we're dropping the bucket in the grand scheme of things. We've all got to tell this story or else we. We don't have a chance. And that's unacceptable to me.
Adam Sadelik
So you asked me, like, what surprised me or what's changing most dynamically. And that piece actually is one of those things where we now have a deep network of partnerships in this space. Like, we're not doing this alone. So whether it's OEMs or distributors or dealers, I feel like the world is coming together and it's inflecting in a way that is very good for customer success.
Aaron
I think so. And I think that's the.
As I was explaining with the OEMs, I think that's the future. I think the strength of the OEMs is one in their manufacturing, you know, abilities.
Adam Sadelik
They're top notch.
Aaron
Yeah. And to go like you want to be a bulldozer manufacturer. Okay, good luck.
Adam Sadelik
I don't know anything about that.
Aaron
Yeah. And you need a plant, you need a supply chain and then. Okay, cool. You can make as many bulldozers as you want. Now I think the next, I think arguably the most important thing with the OEMs and I believe why Caterpillar is. Caterpillar is then the dealer network and the dealer infrastructure and the amount of money it would cost to just create a support infrastructure that they have on a global scale.
Unbelievable. Unbelievable. The amount of capital required to even support. You want to go sell a thousand dozers. Okay. You've got to support all thousand. Good luck competing with that. And so I think that's their strength is in manufacturing.
Physical products, machinery and then supporting that machinery.
Technology is, is something I think others can do much better. But then if you tap into their ability to produce machinery, since we'll always need machinery and then ability to support that machinery, that's a great combination. You're doing what you do best, which is the technology, which is the automation. They're doing what they do best, which is selling machines, supporting machines. Now you've got something pretty cool.
Adam Sadelik
Yeah. I mean we are seeing some of these multiplicative effects in reality now where you know, you take the equipment, you upload more software into it, basically add few more sensors and now you have a machine that creates tremendous amount of value.
Aaron
Sure. Do you have. Before we talk about the red planet far away, do you have earth moving people that work at the business?
Adam Sadelik
Yeah, we do. So if you look at a team, kind of a cross section across the things that we need to deliver. So you have high tech people, you know, x Google X SpaceX software software, AI perception hardware as well.
Aaron
Okay.
Adam Sadelik
It's a great, great team. Amazing team. And that's the reason why we exist, why we're allowed to. To scale.
Aaron
Yeah.
Adam Sadelik
What the world allows us to scale. And then we have people who come from the earth moving industry. So whether it's, you know, former mining managers, Earth moving operators, operators of the heavy equipment itself.
Aaron
Okay.
Adam Sadelik
Mind planners, people who have built terraforming projects like US Army Corps of Engineers. They are responsible for a lot of the shoreline.
Aaron
They're one of the biggest terraformers in the world.
Adam Sadelik
That's right. Yeah, that's right. And so General DeLuca is on our advisory board and he was in charge of a lot of those projects.
Aaron
Interesting.
Adam Sadelik
And so we have been really fortunate to compose a company that has all of this insight, all of this talent.
Aaron
Because I think that's the only way you do it. You've got to have the technology talent, but you've also got to have the industry expertise or else it just doesn't work.
Adam Sadelik
Otherwise you're throwing over the wall some shiny object and it will never actually adopt.
Aaron
Yes. Yeah. Which is also why I think like the only way to make this work is a. Such somebody that's lived in that world even for a small amount of time, like you did with within Earthmoving before you went off onto this venture. You've gotta.
To use the term first principles. I feel like you have to understand the first principles.
Adam Sadelik
I think you have to feel the pain.
Aaron
Yes.
Adam Sadelik
Like basically every day we are designing for the pain that a lot of us used to feel when we are in that ecosystem.
Aaron
So you, I think in my first conversation with you, you were talking about terraforming, but not just Earth. You also want to terraform other places.
Adam Sadelik
Yeah. I mean, the mission keeps on going.
Aaron
Yeah, why not? Why not think big.
Adam Sadelik
So first of all, I think there's a lot of. Lot of work left on this planet.
Aaron
I think so.
Adam Sadelik
To fix it and make it better.
Aaron
I think so too.
Adam Sadelik
So we are heavily focused on that and I think that's going to consume us for a long time.
Aaron
I think that's a safe, A safe bet. Yeah.
Adam Sadelik
But as like a retirement project, I think you can imagine, basically, if humanity goes outside of this planet, just like Earth, moving is fundamental for what we do here, it will be just as foundational for those other worlds.
Aaron
Sure.
Adam Sadelik
Right.
Aaron
Which makes sense. Yeah.
Adam Sadelik
Right. It's hard and it's probably not going to be like humanoid robots with like wooden shovels moving dirt.
Aaron
Right.
Adam Sadelik
It's going to be specialized equipment that's already delivered value and we have experience with it. So I think it's safe to assume that like a lighter version of these vehicles will eventually go to space to do a variety of work.
Aaron
Sure.
Adam Sadelik
The ambitious version of this is on Mars, you start mining the polar caps with, with ice and process it. And so like it becomes kind of like you mentioned. I think if you squint your eyes in the future there will be mining type operations on Mars that look a lot like, like an iron ore site in Australia for example because it has similar, similar composition. The gravity will be the variable.
Aaron
Yeah.
Adam Sadelik
But you know the robots will do roughly the same things.
Aaron
Yeah. There will be the, the flies won't be as bad. Yeah.
Western Australia flies, man. That's. That's next level out there.
Adam Sadelik
Yes. And they'll be probably electric and so on. So that's another thing that we are seeing alternative fuels.
Aaron
So. And that's a whole other ball game is the only way.
Mining equipment can go to any kind of alternative fuel sustainably is through autonomy is how I've had it explained to me, especially batteries. Now my opinions on batteries and mining I don't see it all that commercially viable anytime soon. But the way I've had it explained is you need autonomous machines because they have to know they have to optimize themselves based on energy consumption and then based on charging. And you can only mathematically do it if the machine is optimized and controlled by software by a computer. Which actually makes sense.
Adam Sadelik
Yeah. I mean it's similar to. Yeah, I guess we do the same thing with a diesel powered machine and.
Aaron
You'Re optimizing based on energy.
Adam Sadelik
Yeah. And so I think with the battery management becomes more acute to do it really well.
Aaron
Yes. Yeah. Because it's, it the, the, the limiting factor there is charging and so you have to optimize for when and what you're charging over a 24 hour cycle and then a 48 hour cycle and 72 hour cycle and you have to be, you have to be so far into the future because if you're thinking about right now it can then hurt you. Next week is how I've again had it explained to me but it goes over my head pretty quick.
Adam Sadelik
Yeah. Yeah. But the robots can also swap their own batteries. That's an advantage of you know having an excavator or shovel on the side is that it can do a lot more things than just digging. And so for example it can swap batteries for its partners or other vehicles. Just lift the battery, swap it for a charge one instantly. So I think that's going to be the future of at least in construction. I think this will happen pretty quickly because there's more pressure to have quiet low emission sites.
Aaron
Yeah.
I think in some applications it makes sense. In indoor type applications, demolition. I think it's great thing. Underground mining, I think it's well worth exploring in very urban scenarios. If you're running next to a school, by all means.
Big picture though, I'm not a believer Honestly, after seeing it, and I've seen it, I've seen the most advanced battery job site in the world. It's amazing. Is it economical, though? Not even close. It is. So it's early days, sickeningly subsidized.
Adam Sadelik
But in mining, what we have seen work well is that you have actually a rope shovel or more static vehicles connected permanently to a power plant.
Aaron
I'm all for that. Yeah. That makes.
Adam Sadelik
That works. Yeah.
Aaron
Because. But that's driven by economics. It's a lower cost of operating. So when you're getting your lower, Lower operating costs, by all means, I don't, I don't care if it's electricity or diesel. I just care that it's economical and sustainable.
Adam Sadelik
It's economical and it's. And it's relatively straightforward because it doesn't move around that much.
Aaron
Correct.
Adam Sadelik
It just spins.
Aaron
Yes, yeah. And it just spins. So. Yeah, you don't have to be positioning the cable all that much. Like the drag line out there. All drag lines are electric. All rope shovels are electric.
Adam Sadelik
That's right.
Aaron
They have.
Adam Sadelik
So they're already in the future.
Aaron
100 years. Yeah, yeah, yeah, yeah, yeah. Or even the electric drive trucks. Electric drive dozers. I know the electric drive dozers they're having. There's some growing pains with those. But the electric driver drive trucks, that's a great technology. Electric drive loaders. That's a great technology. They're trying electric with shovels. I don't, I'm. I think the jury's out on that. But they're deploying more around the world, so it must be working. But then when it comes to other stuff, I'm like, I don't know, man.
Yeah.
Adam Sadelik
You know, one thing at a time.
Aaron
One thing at a time. Yeah, yeah, yeah.
Now.
Is the scaling phase to things. And so now is it. Just hang on and try to get as many of these machines as you can out into the world.
Adam Sadelik
Well, in a responsible way, that's an important qualifier. I think, like the standard Silicon Valley playbook would be just like, go, you know, disregard any issues. Just keep going.
Aaron
Sure.
Adam Sadelik
But I know for this ecosystem, that's not the right way.
Aaron
Yes.
Adam Sadelik
Like, you want to go fast as long as you are very solid on all the things that matter. And so that's what we are doing. Like, you go pedal to the metal, but you make sure there are certain things around safety that you never cross.
Aaron
Yeah, of course. Yeah. And.
It has to work consistently.
Adam Sadelik
Yeah, that's right. Otherwise I won't be sitting here.
Yeah, it has to work. And it has to.
Have a trend line that's clearly pointing to the future.
Aaron
The way I see you scaling too is you start again with if an operation has. And correct me if I'm wrong, but 10 dozers you'll automate one or two to dip your toes into.
To allow them to see the benefits. To allow the operation to understand autonomous equipment. If they haven't seen it before especially. And then you'll continue to add autonomous machines from there is ultimately the goal.
Adam Sadelik
Usually goes in phases.
Aaron
Sure.
Adam Sadelik
Of that sort. Yeah.
And so what's changing is the rate of speed at which we go through those phases with the customers. So like the initial deployments took us a year to do. Now we are doing the same amount of work each week. And so is that compression that allows.
Aaron
The scale because you're gathering more data and the processing of that data can move faster.
Adam Sadelik
Yeah. It's really the whole connection of things. It's not a single factor. The technology is getting better. We are getting better at deploying it. Working with the customers. The customers also.
Know what to expect because now we have case studies. We have their peers who are using it. So the whole thing is a lot more smooth than when you're bootstrapping from scratch.
Aaron
Sure. And I guess once you. The mining companies too are very global. So once you get in with a mining company and it works over here, they'll want it over there. Over there. Over there.
Adam Sadelik
That's right. And like we discussed earlier here, the value creation is so staggering that it goes really quickly.
Aaron
Well. And once it reaches in theory, there's probably a tipping point in some way as well. From a. Well now because the growth and productivity is so substantial, everybody at a certain point has to do it.
Adam Sadelik
That's right. Otherwise you get left behind.
Aaron
Otherwise you're left behind.
Huh. It's so interesting. And it's. I think it's just again, like autonomous trucks are awesome. But that's been a technology that I'm like, it's a great. And the right application. It will never be everywhere. I just don't see that. Or anytime soon, the next few decades. I don't see it going mainstream. I just don't.
And maybe I'm completely wrong at that. But the other stuff.
Again, because you don't need to reconfigure a whole operation. You don't need this huge outlay of capital to make it happen. You can just retrofit existing machines and just let them go in an existing application, which I think is brilliant.
Adam Sadelik
And it's injected in this ecosystem that's reasonably ready for it. If you take the site that's successful with ahs, the autonomous trucks, and you now bring all the other machines and turn them to autonomous. So now you have a zero entry site because the trucks already have nobody inside. Now all the other machines become autonomous. And so that is the inflection point where now the site transforms to a just fundamentally different level of production.
Aaron
And that's the gold standard.
Adam Sadelik
Yeah. That is the North Star.
Aaron
And I know the issues they oftentimes have at an AHS site are because of people.
Adam Sadelik
Yeah. Light truck comes into contact.
Aaron
Yeah. Or manned equipment, etc. Getting in the way of the computer. If people are messing with the computer, it starts to mess up the optimization.
Adam Sadelik
Exactly.
Aaron
Which is quite interesting.
Adam Sadelik
Yeah.
Aaron
Yeah. And that's, that's how I've had it explained to me. And even, I mean, some stuff that you just don't think about, like, like blasting, for example, is a huge inefficiency in mining because you have to clear the whole radius, which is oftentimes, if you're especially working in a pit application, you've got to clear the whole pit, which takes 30 minutes, 45 minutes to clear everybody. And then you pull the shot. And then now you've got to go down and clear the shot. And so you're an hour and you're shooting sometimes twice a day. So you're losing. You're shutting your entire operation down an hour, two hours every single day.
Adam Sadelik
Yep. And occasionally somebody lefts behind. Gets left behind.
Aaron
Yes. But instead of that, my point is you can just have robot machines continuing to go within a much tighter radius.
Adam Sadelik
Yeah, that's right.
Aaron
Yeah. Because they don't have people.
Adam Sadelik
And that's exactly what we are seeing. Yeah. So like there's a long range of like. Basically every week we find a new big benefit.
Aaron
Sure.
Adam Sadelik
That I wasn't aware of before.
Aaron
That's amazing.
Adam Sadelik
Toast the holidays in a new way and raise a glass of rumchata, a delicious creamy blend of horchata with rum. Enjoy it over ice or in your coffee. Rumchata. Your holiday cocktails just got sweeter. Tap or click the banner for more. Drink responsibly. Caribbean rum with real dairy cream, natural and artificial flavors. Alcohol 13.75% by volume, 27.5 proof. Copyright 2025 Agave Logo Brands Pojoaque, Wisconsin. All rights reserved.
Aaron
What is most misunderstood about all this?
Adam Sadelik
The value creation. Basically the default stance is that this is about cost savings. This is a robot that's a few percentage Points better at some metric. And it's going to be this annoying thing to integrate and at the end will save, you know, $50 a day or something on this.
Aaron
Yeah.
Adam Sadelik
And so then what's missed is that. No, like if you actually run the fleet in this way, as you can see in some of our case studies, you create a lot more output from the ground. And as a result, you get the maximum level of safety, which is nobody exposed to these conditions. So you get the output, the money, the power that comes with that. Having that kind of ability to manage the site with full control rather than kind of delegative control through many layers of management.
Aaron
It's so interesting. Yeah. That I've been thinking about it, even with what I knew in that more traditional sense of, well, this is. Yeah, it's a small cost savings. You're just eliminating the individual. And as you say this, it's like, oh, this makes perfect sense. Sense. You're able to do so much more with less or the same amount.
Adam Sadelik
That's what it boils down to.
Aaron
Huh.
Pretty exciting.
Adam Sadelik
Yeah. I have a lot of respect for these ecosystems. Like they're so important.
Aaron
I think you have to have respect to do what you're doing. I don't think there's a way to sell into this world without a huge degree of respect. I think this whole world is built on respect. And if you come in and you think you know what you're doing or you think you know better or you know, you're Mr. Tech Guy. Oh, I'm going to show you guys how to do mining. It would just wouldn't work. You'd get laughed out of the place no matter how good your technology is.
Adam Sadelik
But it's also hard not to have respect. Like, I go to these meetings in a rental car that was made from steel that came from that site that I'm going to.
Aaron
Yeah, yeah, yeah.
Adam Sadelik
So like, I wouldn't literally be there unless they did the good work.
Aaron
Sure. Yeah. But, but that's. And that's how you build respect is by going out to these sites and spending time with the people. I think a lot of times though.
I go to some, especially I'll see some people that serve this world in technology or other places. And it's like even just based on what they're saying, they're like, I just can understand you haven't been out, like, get out to a job site, go out there and see how this world works. Like even, even executives in this world now too. It's like.
Even last night I Was talking to somebody. They just made this change. And now this guy is running this mining operation on a different continent. He's never been there.
It's like.
I mean, you can only do so much without really understanding the world in which you're playing in. And there's only one way to really understand it, and that's by going to see it and spending time in it and spending time with, most importantly, the people doing it every day. Like, you can talk to leadership, you can even talk to site management. But I get so much more color by talking to a truck driver, by talking to a foreman, you know, frontline leader, by sitting in an excavator buddy seat and just talking to the excavator operator.
Adam Sadelik
Yep, absolutely.
Aaron
You learn so much from those people.
Adam Sadelik
That's why I love your example. Going there at the shift change.
Aaron
Yeah.
Adam Sadelik
And like having the parents see the ecosystem, that's the most real touch point.
Aaron
Yeah, it is. Well, and that's. I prioritize shift change. So every time I go out to a mining operation, that's the very first thing I say is what time shift change at 6, 6, 35, 37. And let's go out at shift change. I just, I like being there at shift change because again, it's. It humanizes the operation because after shift change, everybody gets in a big truck and it's just a bunch of big trucks moving around the landscape, which is cool. It's great. But it's, it's, it's. It's inhuman. Whereas when you see all the people making it happen out in the middle of nowhere, you're like, oh, so this is how the world works. Like these people.
Adam Sadelik
Mm.
Aaron
That's the only way it happens.
Cool. I'm excited. It's really cool what you're.
Adam Sadelik
Yeah, it's a. It's a great time to be alive.
Aaron
Yeah, I think so. I'm. I'm again in my. I'm. I've been thinking a lot about.
Like the. The summit for us, the area dirt world summit. That's now like my biggest speaking opportunity every year. And I've been approaching my speaking. I started speaking years ago, and when I started speaking, I would get up on stage. I'm in my 20s. I. I'm not naturally a speaker. And so I made the rookie mistake of trying to sound smart and be smart on stage and trying to be worthy of being on stage. Well, I'm up here. I better, I better say something valuable. Right.
And as I've done it over the years, I've realized one.
That is, that's not the case at all. I don't have to be smart. I can just report on what I'm seeing because I have a really unique view of the industry. Like, you do too. You get to see a lot of the industry a lot more than most people do. So I have this unique view, unique perspective. But two, it's always changing as well as you get more information. And it's cool. Like, being a human is quite cool as well, is because, like, and I've started to realize this for the first time in my Life. You know, one year at 30 is not like one year at 23. Because now you're stacking experience on top of experience.
Adam Sadelik
That's right.
Aaron
I feel like you're able to just develop more and faster in a lot of ways.
Adam Sadelik
You have, like, more followers, bigger team. Like, your span of control is bigger.
Aaron
Yeah. And, yeah, you're interacting with more people. It's just, there's, there's, there's more going on. You're able to just create more value in your life. And so all of this to say every year I, I sit down and I essentially, like, rework my talk. Now I start with a blank sheet of paper. I don't like building it off of what I've already done. A blank sheet of paper. All right. I'm going to do my best to then craft this as good of a message as possible. And this year, the message is really about the future. And the future being better than it is today because it's like.
The future is. Is on the shoulders of the people in this industry.
That's it. That's it. I think it's, it's on the shoulders of people creating or producing natural resources, extracting natural resources and building infrastructure and maintaining infrastructure. And if we're not here to make the future better.
What'S the point to everything we're doing? What's the point? What's the point of all the money and the big business and this and that? If we're just here to extract and then that's it. And just leave a mess behind? It's like, okay, maybe you're good to go, but what about your kids?
What about your kids? It doesn't matter how much money they have or what they go do. If this whole thing fails, they're done. And then, okay, cool, they're good. What about their kids? Like, that should be the whole point of what we're trying to do here. Everybody, we need everybody aligned on the fact that we need to make the future better. And the future is on our shoulders and we need to accept that responsibility and we need to be very good stewards of that responsibility.
And part of that is.
Producing more with less, building better, reducing man hours, increasing wages and being a lot more productive. Because we haven't been for 50, 50 straight years now in the United States.
Adam Sadelik
And to your point earlier, right, if you do that, you get to reinvest that and make it more sustainable, make it like a closed loop cycle.
Aaron
That's how you make it sustainable rather than runaway effect. Yes.
Adam Sadelik
And that's what we are seeing already. Like if you have access to automation, it turns out reclamation becomes a lot less painful. And so you can do it as a byproduct of the normal operations that are now turbocharged. And so we are seeing evidence for these things already.
Aaron
Yeah, yeah. Yes. Because it allows you.
If you're, if you're in a survival state, if your business is in a survival state, you can't afford to think about the future.
Adam Sadelik
That's right.
Aaron
And so all you're thinking about is let's just extract as much as possible. Reclamation. Screw it. We can't even afford to think about it right now. But yeah, if you can maximize that extraction and you can make more money now, you can start to think about other things.
Adam Sadelik
So I'm really excited to see a bootstrap of a new cycle that's a virtuous loop. Because you create this value, then it goes back to the business and it makes it better if managed correctly.
Aaron
Sure.
Adam Sadelik
By the way, I'm looking forward to your talk later this week.
Aaron
It should be good. I'm looking forward to it too.
I've spent a lot of time on this. Yeah, I'm really excited.
Adam Sadelik
Can't wait.
Aaron
Yeah. Well, thanks for coming.
I have a lot to think about. This is just, it's, it's really interesting what you guys are doing. And again, it aligns so nicely with all the stuff that, that I talk about and it doesn't make the industry any less human. It makes it more sustainable and more human. As far as I'm concerned.
Adam Sadelik
It's.
Aaron
It's the only way. I see.
Adam Sadelik
That means a lot coming from you, Aaron.
Aaron
Yeah. Yeah. So, yeah, thanks for what you're doing. How do people learn more about what you, what you're doing?
Adam Sadelik
Go to our website.
Aaron
Go to your website.
Adam Sadelik
Read our social. We have been becoming more open about what we do. Yes, we'll do more of that.
Aaron
Yeah, I'm very excited. How did we even meet to begin with?
Adam Sadelik
I don't remember?
Aaron
Did I reach out to you?
Adam Sadelik
I actually don't remember. It's a long term relationship at this.
Aaron
Point to find out, I don't even remember. Yeah. Because you were. You were still in stealth mode.
Adam Sadelik
Yeah. Very much you. Yeah.
Aaron
And so there wasn't. But now I see stuff, your stuff all the time.
Adam Sadelik
Yeah.
Aaron
All right. Well, thanks for stopping by. I will be.
Watching very carefully. And maybe we can catch up again in a few years when your machines are all over the world.
Adam Sadelik
They are real.
Aaron
They are. They all.
Adam Sadelik
Yeah. But many more of them.
Aaron
Many. Yeah. Yeah. Right on. Thanks.
Adam Sadelik
All right. Thanks.
Aaron
It.
Guest: Adam Sadilek (Founder, AIM Intelligent Machines)
Host: Aaron Witt
Release Date: December 11, 2025
This episode of Dirt Talk features an in-depth conversation between host Aaron Witt and Adam Sadilek, founder of AIM Intelligent Machines. The central theme revolves around the automation of heavy construction and mining equipment via AI and machine learning, with a particular focus on the transformative potential in the earthmoving and infrastructure sectors. Adam shares: his unique journey from early AI research to Google X and eventually forming AIM; the technological and operational challenges of automating ground-engaging machines; and the broader implications for workforce, productivity, and the future of infrastructure development ("terraforming").
[04:07–08:19]
[11:48–13:28]
[14:03–17:44]
[18:49–22:54]
[23:20–27:43]
[28:01–34:45]
[41:00–46:14]
[47:32–56:45]
[74:24–79:43]
[85:00–88:47]
| Timestamp | Segment | |---------------|-------------| | 00:47–04:07 | Opening banter & background | | 04:07–08:19 | Adam’s early AI/Google X experience | | 11:48–13:28 | Becoming an accidental earthmoving contractor | | 14:03–17:44 | Complexity of automating ground-engaging machines | | 18:49–22:54 | Mining truck automation vs. dozer/excavator automation | | 23:20–27:43 | The “terraforming” vision & infrastructure examples | | 28:01–34:45 | AIM’s tech, learning process & operator integration | | 41:00–46:14 | Human impact, operator roles, labor shortages | | 47:32–56:45 | Productivity, safety, reinvestment, industry sustainability | | 74:24–79:43 | Company growth, partnerships, scaling strategy | | 85:00–88:47 | The future: electric machines, Mars, and more | | 104:10–106:27 | The industry's core mission & responsibility |
For listeners: This conversation is for anyone interested in the intersection of heavy civil work, AI and machine learning, workforce dynamics, and the enormous societal impact of infrastructure and resource extraction. The tone is candid, thoughtful, and refreshingly free of hype.