WEBVTT - Teaching AI to Build Stuff in the Physical World 0:00:15.356 --> 0:00:15.796 Pushkin. 0:00:20.436 --> 0:00:26.436 AI works amazingly well. It works terrifyingly well even for 0:00:26.836 --> 0:00:32.036 virtual things, for words, for pictures, for videos. This is 0:00:32.036 --> 0:00:36.476 true in large part because of the Internet. The Internet 0:00:36.516 --> 0:00:41.876 provides this wildly abundant, readily available source of words, pictures, 0:00:41.916 --> 0:00:47.876 and videos to train AI models. But there is no analogous, 0:00:48.356 --> 0:00:52.356 wildly abundant, readily available data set for the physical world. 0:00:53.116 --> 0:00:57.956 There is no gargantuan Internet like repository of data that 0:00:58.116 --> 0:01:02.836 describes how things move and bend and break in real 0:01:02.956 --> 0:01:06.876 physical space. And as a result, we do not yet 0:01:07.036 --> 0:01:11.996 have robust AI for the physical But people are working 0:01:12.036 --> 0:01:16.116 on it, and if they succeed, they'll change the way 0:01:16.196 --> 0:01:19.596 the world works, not just the world as it appears 0:01:19.596 --> 0:01:23.516 on our screens, but the actual physical world, the world 0:01:23.516 --> 0:01:32.876 where if you drop something on your foot it hurts. 0:01:33.276 --> 0:01:35.636 I'm Jacob Goldstein and this is What's Your Problem, the 0:01:35.716 --> 0:01:37.956 show where I talk to people who are trying to 0:01:38.036 --> 0:01:43.036 make technological progress. My guest today is Edward Mayer. He's 0:01:43.076 --> 0:01:47.436 the co founder and CEO of Machina Labs. Edward's problem 0:01:47.516 --> 0:01:51.596 is this, how can you use AI to turn robots 0:01:51.996 --> 0:01:58.316 from dumb, inflexible machines into skilled versatile craftsmen. Before he 0:01:58.396 --> 0:02:02.396 started Machina Labs, Edward worked in the rocket ship business, 0:02:02.676 --> 0:02:06.516 first at SpaceX and then at a company called Relativity Space. 0:02:07.076 --> 0:02:10.636 And in the rocket business, the word solw firsthand the 0:02:10.756 --> 0:02:14.676 problems of traditional manufacturing. It's the kind of problem he's 0:02:14.716 --> 0:02:17.956 now trying to solve with AI and robots. It's a 0:02:17.956 --> 0:02:24.916 problem called the rigid factory problem. So I've heard you 0:02:24.996 --> 0:02:27.676 use this phrase that's interesting to me, and it's the 0:02:27.796 --> 0:02:32.756 rigid factory problem. What's the rigid factory problem? 0:02:32.996 --> 0:02:36.236 That main problem with the factories today is that rigidity, 0:02:36.316 --> 0:02:39.156 meaning that if you have to build a physical product, 0:02:40.276 --> 0:02:42.916 you pretty much have to build a factory that's designed 0:02:42.956 --> 0:02:45.556 for it and built for it. There's a lot of 0:02:45.596 --> 0:02:49.436 components that goes into the factory, from machinery all the 0:02:49.436 --> 0:02:52.396 way to the tooling that is required to build products 0:02:52.436 --> 0:02:55.916 that are specifically designed for the geometry for the material 0:02:56.036 --> 0:02:58.196 that you're trying to use. The moment you want to 0:02:58.276 --> 0:03:00.836 change that, you have to change your factory, which is 0:03:00.876 --> 0:03:03.516 a huge investment. You know, I always give an example 0:03:03.596 --> 0:03:06.516 from when I was at SpaceX. You know, you think 0:03:06.556 --> 0:03:10.476 of SpaceX as a very innovative and it is you 0:03:10.476 --> 0:03:12.516 know on the edge of a hardware space in terms 0:03:12.516 --> 0:03:17.076 of innovation in the past twenty four years twenty three 0:03:17.156 --> 0:03:20.876 four years that they have existed, they have two rocket families. 0:03:21.236 --> 0:03:24.356 There's Starship and there's Falcon, right, because at the moment 0:03:24.356 --> 0:03:27.236 you decide on diameter of for example, Falcon nine or 0:03:27.276 --> 0:03:29.916 the Falcon family in general, the diameter of that core, 0:03:29.956 --> 0:03:31.756 it's very hard to change it. A lot of tooling 0:03:31.836 --> 0:03:35.516 and machinery specifically built for that diameter. And that's why 0:03:35.516 --> 0:03:37.236 for Starship they had to start from scratch. 0:03:37.636 --> 0:03:40.476 Start from scratch, meaning like not just design, but like 0:03:40.836 --> 0:03:43.476 the factory itself, like the Factorily had to build a 0:03:43.476 --> 0:03:46.436 whole new factory because they wanted to make a different 0:03:46.516 --> 0:03:47.396 sized rocket. 0:03:47.876 --> 0:03:51.076 Yes, different size, different material, All the tooling has to change, 0:03:51.156 --> 0:03:54.876 right almost almost, Yeah, you have to basically assume building 0:03:54.916 --> 0:03:57.516 from scratch, ground up factory. Why does it need to 0:03:57.556 --> 0:03:59.236 be there for us to build this new product? 0:04:00.316 --> 0:04:03.396 I heard you describe was this from your own experience 0:04:03.836 --> 0:04:08.876 the sort of era at SpaceX when the fact that 0:04:08.916 --> 0:04:12.316 you couldn't make the rocket wider led to all these 0:04:12.436 --> 0:04:16.036 kind of difficult things people were trying to do to 0:04:16.036 --> 0:04:18.116 be like, how can we do all these things under 0:04:18.156 --> 0:04:20.716 this fundamental constraint, Like, can you talk a little bit 0:04:20.756 --> 0:04:21.196 about that. 0:04:21.596 --> 0:04:25.636 Yeah, this is a lot of conversation happening in twenty twelve, 0:04:25.716 --> 0:04:30.996 twenty thirteen, twenty fourteen time when the diameter of the 0:04:30.996 --> 0:04:33.676 Falcon nine could not get any larger, and if you 0:04:33.716 --> 0:04:36.356 look at actually different Falcon versions, the height of that 0:04:36.436 --> 0:04:40.036 vehicle kept going higher, the diameter could not change. So 0:04:40.076 --> 0:04:42.796 it was about where what space can you find to 0:04:42.876 --> 0:04:49.116 put new features and new designs that exist within the vehicle. 0:04:49.196 --> 0:04:51.156 So there was a lot of stuff basically being crammed 0:04:51.196 --> 0:04:52.916 into the space that you have already got. 0:04:53.076 --> 0:04:55.556 So that's true for building rockets. I mean, what are 0:04:55.596 --> 0:04:59.836 some other just you know, different kinds of manufactured products 0:04:59.836 --> 0:05:02.036 where that kind of rigidity is a problem. 0:05:02.396 --> 0:05:06.436 Yeah, I think it is just common almost in all manufacturing. 0:05:06.476 --> 0:05:08.836 That's why this phenomenon. I think it's kind of funny. 0:05:08.836 --> 0:05:12.716 People take it for granted that a thing called economies 0:05:12.716 --> 0:05:16.236 of scale, uh huh, Like people take it for granted 0:05:16.276 --> 0:05:18.716 as if it's rule of nature. It's actually not. 0:05:19.236 --> 0:05:21.556 Just to be clear, it's basically, the more you build 0:05:21.556 --> 0:05:23.796 of a thing, the cheaper each one of those things gets. 0:05:23.796 --> 0:05:25.876 If you build one, it's really expensive. If you build 0:05:25.876 --> 0:05:27.236 a million each one's. 0:05:27.116 --> 0:05:29.716 A lot cheaper, yes, exactly, But then people don't think 0:05:29.756 --> 0:05:32.036 about it's like, oh, okay, intuitively makes sense, but why 0:05:32.196 --> 0:05:36.756 it's actually not that intuitive. It's actually a limitation of technology. Right. 0:05:38.836 --> 0:05:41.636 Why why economies of scale is a thing is because 0:05:41.636 --> 0:05:43.796 you have to make a huge amount of investment to 0:05:43.876 --> 0:05:46.156 make the first thing, and the moment you make the 0:05:46.196 --> 0:05:48.436 second thing and the third thing, then you can break 0:05:48.516 --> 0:05:51.876 even your investment onto more products that you're going to 0:05:52.116 --> 0:05:54.276 come out of it. But that's only true if the 0:05:54.316 --> 0:05:57.796 second product can be built for the first investment. You 0:05:57.876 --> 0:05:59.836 had to turn this concept and say, oh, this is 0:05:59.876 --> 0:06:02.076 a given. This is an axiom of the world that 0:06:02.156 --> 0:06:04.676 economies of scale is a thing, but in reality it 0:06:04.716 --> 0:06:08.196 is a technological challenge. Right. It means that you build 0:06:08.196 --> 0:06:10.676 a car, you ask what application you Once you build 0:06:10.676 --> 0:06:14.116 a factory for a car and all the toolings and 0:06:14.196 --> 0:06:17.276 dies that goes into stamping of the panels of that car, 0:06:17.716 --> 0:06:20.596 that's one hundred and fifty million dollar investment just for stamping. 0:06:21.276 --> 0:06:23.116 And this is a number for example from Tesla. Tesla 0:06:23.156 --> 0:06:25.276 spens one hundred and fifty million dollars in a stamping 0:06:25.316 --> 0:06:29.756 plant they have in Giga factory in Texas, right, and 0:06:29.836 --> 0:06:33.076 that can only make Model Y or Model three right 0:06:33.196 --> 0:06:35.956 the moment you have to change that. That means go 0:06:36.036 --> 0:06:40.156 through every eighty two hundred and thirty sheet metal panels 0:06:40.156 --> 0:06:42.436 that exist on that car and design a new tool 0:06:42.476 --> 0:06:44.756 for it. And each of these tool is going to 0:06:44.836 --> 0:06:47.156 be a few hundred thousand dollars to sometimes a million 0:06:47.196 --> 0:06:49.396 dollars or a million and a half million dollars. And 0:06:49.436 --> 0:06:51.756 you're talking about like eighty t one hundred and thirty 0:06:51.796 --> 0:06:52.676 tools per vehicle. 0:06:53.036 --> 0:06:55.956 And like all you're doing you're not reinventing the car there, 0:06:55.996 --> 0:06:58.596 you're just making a car that's a slightly different shape. 0:06:58.716 --> 0:07:01.756 Basically, yes, maybe you may get sedan, and you're not 0:07:02.076 --> 0:07:04.916 trying to do a slightly longer version, a slightly bigger version. 0:07:06.076 --> 0:07:08.476 And that's why economies a scale of the thing you saying, Okay, 0:07:08.516 --> 0:07:11.836 I made a fact. Now it only pays back if 0:07:11.876 --> 0:07:14.556 I make a million of this car, right, because I 0:07:14.596 --> 0:07:16.596 had to just drop one hundred and fifty million dollars 0:07:16.676 --> 0:07:20.116 on just a stamping plant. So yeah, it's all over manufacturing. 0:07:20.156 --> 0:07:22.836 We abstract this whole concept and gave it the name, 0:07:22.916 --> 0:07:24.036 says economies, of scale. 0:07:24.556 --> 0:07:30.556 Yeah, so you left SpaceX and you went to Relativity Space, right, 0:07:30.596 --> 0:07:33.356 a company that was also in the space business that 0:07:33.516 --> 0:07:34.876 was using three D printing. 0:07:34.916 --> 0:07:35.076 Right. 0:07:35.076 --> 0:07:37.396 That was the idea of the company, which seems like 0:07:37.596 --> 0:07:40.556 an approach to this problem that you're talking about. 0:07:40.596 --> 0:07:40.716 Right. 0:07:40.756 --> 0:07:43.996 An advantage of three D printing is that it is 0:07:44.076 --> 0:07:48.156 much more flexible and less rigid than traditional manufacturing. Right, 0:07:48.196 --> 0:07:49.476 So tell me about that. 0:07:50.036 --> 0:07:54.236 Yeah. So, yeah, we saw this challenge at SpaceX and 0:07:54.276 --> 0:07:56.356 I joined Relativity very early on. I was the fourth 0:07:56.396 --> 0:08:01.476 person on that team. And the goal over there was, Okay, 0:08:02.236 --> 0:08:05.316 let's just think about this fundamentally, can we build a 0:08:05.396 --> 0:08:09.396 rocket that all built with flexible technology and a time? 0:08:09.396 --> 0:08:14.476 Three D printing was that forefront of everybody's minds because 0:08:14.556 --> 0:08:18.236 people were already starting to build that. NASA SpaceX people 0:08:18.276 --> 0:08:21.036 were already starting to build engines out of three D printing, 0:08:21.436 --> 0:08:23.916 and the concept was like, well, that's great, it's very flexible. 0:08:24.036 --> 0:08:27.796 Three D printing has this promise of geometry agnostic, material agnostic. 0:08:27.836 --> 0:08:29.396 You can just feed it a design and can build 0:08:29.436 --> 0:08:31.716 a product for you. And it worked very well with 0:08:31.796 --> 0:08:35.316 rocket engines. I think probably the future all rocket engines 0:08:36.516 --> 0:08:39.356 would be would be three D printed, And the concept was, 0:08:39.796 --> 0:08:43.996 can we take this and scale it to the whole vehicle, right, 0:08:44.316 --> 0:08:46.476 can we build the whole vehicle with a process like 0:08:46.476 --> 0:08:50.836 three D printing so that it is flexible Today, if 0:08:50.836 --> 0:08:53.236 you want to build a rocket with twelve foot diameter, 0:08:53.316 --> 0:08:55.636 we can do it. And then if our calculation changes 0:08:55.676 --> 0:08:57.356 and we wanted to go to another orbit or do 0:08:57.356 --> 0:08:59.916 a different type of emission, then we can change that 0:09:00.036 --> 0:09:02.996 diet twelve diameter to twenty diameter twenty fie. 0:09:03.076 --> 0:09:04.796 Don't have to build a new factory, don't have to 0:09:04.796 --> 0:09:07.556 build new machinery, just change three D printer. 0:09:07.716 --> 0:09:12.676 Yeah, exactly. So that was a concept behind relativity. That's 0:09:12.676 --> 0:09:15.316 a thesis behind relativity, and that was the goal there. 0:09:15.396 --> 0:09:17.396 The goal was, you know, three D print a whole 0:09:17.476 --> 0:09:18.796 rocket so they can be flexible. 0:09:19.916 --> 0:09:22.476 But it hasn't. It hasn't worked at least in the 0:09:22.556 --> 0:09:25.276 kind of maximalist version, right, Like, they just haven't been 0:09:25.316 --> 0:09:27.556 able to do it. They've they've sort of backed off 0:09:27.596 --> 0:09:29.996 of that that big dream, as I understand it. 0:09:30.156 --> 0:09:36.396 Yeah, yeah, So I think the challenge was that three 0:09:36.436 --> 0:09:39.276 D printing is just one process and it's necessarily not 0:09:39.396 --> 0:09:44.276 good for every type of part. You know, manufacturing is 0:09:44.356 --> 0:09:47.036 very versatile. You do different types of geometries, different types 0:09:47.076 --> 0:09:50.036 of material, and three D printing has a very small reach. 0:09:50.196 --> 0:09:53.556 There's certain type of parts like rocket engines, very good fit. 0:09:54.036 --> 0:09:58.316 You're building a tank, maybe not so right. So yeah 0:09:58.356 --> 0:10:00.196 it's good for certain type of parts, but there as a 0:10:00.236 --> 0:10:01.876 whole lot of other parts. Like I said, you know 0:10:01.916 --> 0:10:05.196 you're building a fuel tank, which is basically large sheet 0:10:05.276 --> 0:10:09.516 metal or thin walled structure, then maybe three D printing 0:10:09.556 --> 0:10:11.196 is not as good as a fit because it takes 0:10:11.236 --> 0:10:14.196 a long time, and also because it's thin, you have 0:10:14.236 --> 0:10:16.716 a lot of physical challenges in terms of controlling the 0:10:16.756 --> 0:10:22.916 geometry and the tolerances. So we realize soon that maybe 0:10:23.636 --> 0:10:27.356 other processes are also need to be automated the same 0:10:27.356 --> 0:10:29.916 way three D printing is. We need to have more 0:10:30.236 --> 0:10:34.276 flexible processes that are not just one process, more flexible platforms. 0:10:34.276 --> 0:10:36.756 They can do different types of processes, not just three 0:10:36.836 --> 0:10:41.956 D printing, to be able to cover a whole variety 0:10:41.996 --> 0:10:44.436 of products in a flexible manner, the same way the 0:10:44.436 --> 0:10:47.596 three D printing that's for certain type of products. And 0:10:47.636 --> 0:10:50.476 that was actually the thinking behind MARKETA Labs is that, okay, 0:10:50.636 --> 0:10:53.276 can we step back and say, what do we need 0:10:53.316 --> 0:10:56.956 to build? What is this flexible platform that can do 0:10:57.156 --> 0:10:59.556 three D printing if needed, or it can do sheet 0:10:59.596 --> 0:11:01.996 forming if it's needed, It can do machining if it's needed, 0:11:02.916 --> 0:11:06.476 but chooses the right operation, right flexible operation for the 0:11:06.516 --> 0:11:09.436 right part, but still very agile and doesn't require a 0:11:09.476 --> 0:11:11.236 lot of tooling and it's not inflexible. 0:11:12.276 --> 0:11:14.916 So it's it's sort of zooming out more. It's saying 0:11:15.676 --> 0:11:18.276 three D printing is not going to do everything the 0:11:18.276 --> 0:11:21.476 way manufacturing works now. It's just too rigid, too hard 0:11:21.476 --> 0:11:24.876 to change things, to rely on on scale to make 0:11:24.916 --> 0:11:28.316 the economics work out. So like that's a very big, 0:11:28.596 --> 0:11:33.876 very abstract thought. To start a company, you got to 0:11:33.916 --> 0:11:36.196 make something or you got to make something that makes 0:11:36.236 --> 0:11:39.236 something like what do you what do you actually do? 0:11:39.836 --> 0:11:44.756 Yeah? So it was interesting, right, you know we actually 0:11:44.756 --> 0:11:47.956 the solution was in our past. Right if you look 0:11:47.996 --> 0:11:48.716 at like. 0:11:48.636 --> 0:11:50.636 The lesson in a movie, it's like the Wizard of 0:11:50.636 --> 0:11:52.596 Oz or something exactly. 0:11:53.316 --> 0:11:56.876 If you look at manufacturing, I mean up to Industrial Revolution, 0:11:56.996 --> 0:12:00.716 it was arts and crafts. Right, it was basically humans 0:12:01.276 --> 0:12:03.516 trying to figure out how to conquer nature, right, Like, 0:12:03.636 --> 0:12:06.356 how am I gonna use my hands? On my brains 0:12:06.356 --> 0:12:11.156 and very few primitive tools to deform a product or 0:12:11.396 --> 0:12:16.196 shape a product from raw material. Right and to this state, 0:12:16.396 --> 0:12:19.196 if you are in a very high mixed manufacturing still 0:12:19.236 --> 0:12:22.516 a lot of that creativity exists. There is a person 0:12:22.636 --> 0:12:25.396 at Space IX. His name is Big John. I don't 0:12:25.396 --> 0:12:27.956 think he's there anymore, but there was this guy. It 0:12:28.076 --> 0:12:32.236 was like, you know, a very skilled maker, a craftsman. 0:12:32.396 --> 0:12:35.196 You could figure out how to use simpler tools to 0:12:35.196 --> 0:12:37.756 build different things in a creative way. Maybe it's not 0:12:37.796 --> 0:12:41.396 a repeatable way like you know a stamping works or 0:12:41.436 --> 0:12:44.396 ejection molding works, but you can be flexible. You can 0:12:44.396 --> 0:12:46.036 do different types of things. You can be creative about 0:12:46.076 --> 0:12:47.956 it and do different type of things. So the inspiration 0:12:48.076 --> 0:12:52.036 came from how actually humans used to do manufacturing but 0:12:52.196 --> 0:12:56.476 realized in order to be flexible, you actually need two components. 0:12:56.516 --> 0:13:00.516 You need intelligence and you need set of simple tools 0:13:00.716 --> 0:13:03.156 with a lot of kinematic freedom. Now you can pick 0:13:03.196 --> 0:13:05.236 up those simple tools, and as long as you have 0:13:05.236 --> 0:13:07.916 the intelligence on how to use tools and what sequence 0:13:07.956 --> 0:13:11.156 and what kind of a process how to use those tools, 0:13:11.156 --> 0:13:13.796 you can actually do a whole variety of projects. 0:13:14.156 --> 0:13:16.636 And so when you say kinematic freedom, you basically mean 0:13:16.716 --> 0:13:19.956 like robot arms that can move in lots of different ways. 0:13:20.076 --> 0:13:22.876 Is that practically what kinematic freedom means in this context? 0:13:23.036 --> 0:13:26.436 Yes, basically can apply these tools in a lot with 0:13:26.516 --> 0:13:28.596 a lot of freedom to the material, right the same 0:13:28.596 --> 0:13:31.076 way humans humans do it, right, you know as a human, 0:13:31.676 --> 0:13:33.596 you know, if you think about it, you can pick 0:13:33.676 --> 0:13:36.516 up a welder and weld something, and then you drop 0:13:36.556 --> 0:13:38.156 the welder, and you pick up a drill and you 0:13:38.196 --> 0:13:40.116 put a hole in it, and you drop to drill 0:13:40.436 --> 0:13:43.676 and you pick up a you know, hammer and maybe 0:13:43.756 --> 0:13:46.276 hammer it into shape. So you actually have a few 0:13:46.276 --> 0:13:48.196 set of tools, but you have a lot of good 0:13:48.276 --> 0:13:52.396 kinematic freedom and most importantly, very creative mind too tells 0:13:52.436 --> 0:13:55.236 you how to apply these tools to the material, so 0:13:55.236 --> 0:13:58.516 they can actually get very complex set of products and 0:13:58.876 --> 0:13:59.636 a lot of diversity. 0:13:59.756 --> 0:14:03.076 So plainly, instead of big John, you want a robot, right, 0:14:03.156 --> 0:14:06.916 That's where that's the kinematic freedom. The tools are kind 0:14:06.916 --> 0:14:09.996 of like old tools, but optimized for the root. And 0:14:10.036 --> 0:14:12.996 then when you say intelligence, that's the one where it's 0:14:13.036 --> 0:14:15.476 like feels more frontier ish, like does that mean like 0:14:15.956 --> 0:14:19.236 clever engineers figuring out how to automate the robots doesn't 0:14:19.236 --> 0:14:20.836 mean AI? Does it mean both? 0:14:21.316 --> 0:14:24.076 Yeah? So I think yeah, you're basically getting to the 0:14:24.076 --> 0:14:27.196 crux of how do you scale it? Right? You need 0:14:27.196 --> 0:14:31.476 to have those three components, and how does the intelligent piece, 0:14:31.516 --> 0:14:34.076 which is the most important piece, comes into play in 0:14:34.116 --> 0:14:37.836 an automated fashion. So early days we started from basic 0:14:37.836 --> 0:14:40.556 intelligence of humans. But then we had a plan to 0:14:40.596 --> 0:14:43.836 capture data and train AI so that you can replace 0:14:44.836 --> 0:14:47.196 the thinking and the creativity that human had to put in. 0:14:47.636 --> 0:14:50.356 What's the first thing you decided to try and build. 0:14:50.356 --> 0:14:52.196 What's the first sort of problem you want to solve? 0:14:52.556 --> 0:14:54.796 Yeah? I think so. I left relativity in twenty eighteen, 0:14:55.156 --> 0:14:57.916 and the idea when I left relativity was there, right. 0:14:57.956 --> 0:15:00.476 I was like, okay, we need to build basically what 0:15:00.556 --> 0:15:03.596 I had in My mom called it robot craftsmen. Robocraftsman, 0:15:03.636 --> 0:15:05.316 we call it a time. How can you build a 0:15:05.436 --> 0:15:07.756 robot system? To your point, you can pick up different tools, 0:15:07.756 --> 0:15:09.476 has the same king mean, but also have to have 0:15:09.476 --> 0:15:13.516 the intelligence. The challenge is you know you said, in 0:15:13.596 --> 0:15:15.796 order to train these robots with AI, you need to 0:15:15.796 --> 0:15:18.076 have a lot of data. And this is not the 0:15:18.156 --> 0:15:19.916 data you can find on internet. 0:15:20.116 --> 0:15:23.476 Right, this is the AI robotics problem, it seems right, 0:15:23.556 --> 0:15:26.676 like unlike with large language models, like that's why we 0:15:26.756 --> 0:15:30.036 have large language models and not AI robots, right because 0:15:30.836 --> 0:15:33.476 because we have the data just sort of randomly sitting 0:15:33.476 --> 0:15:35.516 around on the Internet, and we don't have that physical 0:15:35.516 --> 0:15:37.036 world data for robots. 0:15:36.716 --> 0:15:41.036 Right exactly. So basically the problem narrowed down into Okay, 0:15:41.556 --> 0:15:44.476 how can I generate enough data? How can I create 0:15:44.476 --> 0:15:47.236 a business that has a sustaining way of generating data 0:15:47.636 --> 0:15:49.996 so I can actually build these models, I can build 0:15:49.996 --> 0:15:54.156 this intelligence for these robots. And the thinking was, Okay, 0:15:54.876 --> 0:15:58.516 I need to create a solution that can scale in 0:15:58.556 --> 0:16:03.716 the industry with limited amount of data and some heuristic. 0:16:04.876 --> 0:16:07.116 But then because it's scaling, we can generate a lot 0:16:07.116 --> 0:16:09.036 of data and it starts building AI mods. 0:16:09.436 --> 0:16:12.156 Right. You need a first thing that you can actually 0:16:12.236 --> 0:16:15.996 do before you really have AI, to generate the data 0:16:15.996 --> 0:16:16.876 that will get you to. 0:16:17.596 --> 0:16:22.236 AI exactly exactly. So we're thinking about, Okay, it needs 0:16:22.276 --> 0:16:25.676 to be a large enough market right where we can 0:16:25.716 --> 0:16:28.396 get mass adoption, and we need to solve a problem 0:16:28.476 --> 0:16:30.916 that's big enough it's ten times at least better than 0:16:30.916 --> 0:16:34.236 the current solution so it can actually get adoption, right. 0:16:34.756 --> 0:16:36.876 Meaning you can't just do something as well, you have 0:16:36.916 --> 0:16:38.276 to do it ten times better. 0:16:38.876 --> 0:16:42.156 Yeah, Because I think what we realize is that through 0:16:42.516 --> 0:16:45.116 the last two companies, if something is not ten times better, 0:16:45.236 --> 0:16:49.196 cannot overcome the inertia that exists in an industry for 0:16:49.236 --> 0:16:51.796 adoption because you know, if you're doing something for the 0:16:51.916 --> 0:16:54.116 same way, and in manufacturing, people have been doing things 0:16:54.476 --> 0:16:56.716 the old way for hundred of years, right. 0:16:56.676 --> 0:16:58.836 Yeah, and it's a risk, right if they're going to 0:16:58.876 --> 0:17:01.796 try working with you, they're immediately taking a risk. And 0:17:01.836 --> 0:17:03.636 if it's only going to be a little better, why 0:17:03.636 --> 0:17:04.836 should I take that risk? 0:17:05.276 --> 0:17:08.436 Exactly? So the idea was, Okay, we need to find 0:17:08.476 --> 0:17:11.356 it large enough market for our first application, and we 0:17:11.396 --> 0:17:13.436 need to have a solution that at least ten times better. 0:17:13.756 --> 0:17:15.436 So that landed us. We actually looked at a lot 0:17:15.436 --> 0:17:17.876 of things, from three D printing to forging to a 0:17:17.876 --> 0:17:20.036 lot of things, and then landed on sheet metal. So 0:17:20.076 --> 0:17:23.476 sheet metal is the largest metal processing sector out of 0:17:23.476 --> 0:17:26.316 all It's a two hundred and eighty billion dollar industry today, 0:17:27.236 --> 0:17:30.956 and forming complex sheet metal shapes is very tool intensive. 0:17:31.276 --> 0:17:34.956 So so what we started to do was, okay, can 0:17:35.036 --> 0:17:37.876 we make our robot craftsman's first operation to be forming 0:17:37.916 --> 0:17:40.756 sheet metal, basically forming sheet metal the same way a 0:17:40.796 --> 0:17:43.076 sheet shaper hammer is a sheet into shape. 0:17:43.276 --> 0:17:45.516 And when I think about sheet metal, I mean I 0:17:45.556 --> 0:17:47.276 don't know anything about sheet metal. I think of like 0:17:48.036 --> 0:17:50.556 I think of cars, I think of planes, right, I 0:17:50.596 --> 0:17:53.356 think of like you know, detroit, like stamping. 0:17:53.436 --> 0:17:53.676 Is that? 0:17:53.716 --> 0:17:55.876 Am I thinking about the right thing? Am I missing 0:17:55.996 --> 0:17:57.716 huge or huge sheet metal universe? 0:17:57.756 --> 0:17:57.836 Like? 0:17:57.876 --> 0:17:58.996 What's the sheet metal universe? 0:17:59.156 --> 0:18:01.836 Yes? So sheep metal almost is everywhere. I think is 0:18:01.836 --> 0:18:06.356 the most common metal part that you see on day 0:18:06.356 --> 0:18:08.876 to day, right, because most of the time we use 0:18:08.916 --> 0:18:11.756 metal to be a container for other things. So it's 0:18:11.836 --> 0:18:15.396 usually a thin metal structure that's formed in complex shape 0:18:15.676 --> 0:18:17.916 to hold something else. Now you know it can be 0:18:17.956 --> 0:18:23.556 from case of a computer. Uh, you know, to a car, right, 0:18:23.596 --> 0:18:25.276 you know you're sitting in a freeway you're in to 0:18:25.316 --> 0:18:28.156 see a sheet metal or to a to a airplane 0:18:28.196 --> 0:18:31.316 you're in a sheet metal can to a rocket body. 0:18:31.716 --> 0:18:31.836 Uh. 0:18:32.116 --> 0:18:33.876 For for a lot of rocket someone will composites with 0:18:33.876 --> 0:18:39.636 a lot of a machine metal. And to agricultural heavy 0:18:39.636 --> 0:18:45.476 equipment machinery you think of combines, tractors to even building equipments. 0:18:45.516 --> 0:18:50.396 You look at your h ract ducts are all sheet metal, right, 0:18:50.476 --> 0:18:53.316 because it just makes sense. It's we mostly use metal 0:18:53.316 --> 0:18:56.156 parts to contain other things, and we give it complex 0:18:56.196 --> 0:18:58.756 shapes and that's where she forming comes into play. So 0:18:58.796 --> 0:19:01.796 you pretty much see it everywhere. But the challenge is 0:19:01.796 --> 0:19:04.836 that in almost in all cases, you have to create tooling. 0:19:04.876 --> 0:19:06.316 It goes back to that first problem. He said, you 0:19:06.396 --> 0:19:09.476 have to create tooling for each of those geometries. And 0:19:09.476 --> 0:19:12.996 that's why you know a Ford needs to make sure 0:19:13.116 --> 0:19:15.796 they can sell a million of an FUN fifty before 0:19:15.796 --> 0:19:18.036 they can invest in a plant that makes a new 0:19:18.116 --> 0:19:19.676 version of FN fifty, right. 0:19:19.556 --> 0:19:24.356 Because you basically have to build a bespoke factory just 0:19:24.436 --> 0:19:28.236 to shape sheet metal in a new way exactly for 0:19:28.276 --> 0:19:30.836 a new geometry, for for a new design. Exactly where 0:19:30.876 --> 0:19:34.356 is that a particular problem? Like where is it? Where 0:19:34.396 --> 0:19:36.916 is that? Where does that acutely bind the fact that 0:19:37.156 --> 0:19:39.236 sheet metal is so hard to do if you're not 0:19:39.276 --> 0:19:39.756 working at. 0:19:39.716 --> 0:19:42.716 Scale so expensive? Yeah, so I think now you're coming 0:19:42.716 --> 0:19:45.876 to the even the third stage of how do you 0:19:45.916 --> 0:19:48.476 scale this technology? You need to first find you know, 0:19:48.476 --> 0:19:49.796 you said you need to be ten ex better we 0:19:49.836 --> 0:19:53.036 need Right, you're in an area that has a lot 0:19:53.076 --> 0:19:54.236 of pain with today's time. 0:19:54.316 --> 0:19:56.436 I was like, oh my god, thank god, you've walked 0:19:56.476 --> 0:19:58.236 through the door. We've been waiting for you. 0:19:58.636 --> 0:20:02.436 Yeah, So end up being very much defense in airspace. Right, 0:20:02.516 --> 0:20:06.756 So think of you know, think of our military for example, 0:20:07.196 --> 0:20:12.836 right today, they we have fifty sixty different weapon system 0:20:13.116 --> 0:20:15.956 or defense systems you can basically think of like aircrafts 0:20:17.236 --> 0:20:19.916 that they're maintaining. And some of these systems have been 0:20:19.916 --> 0:20:23.036 built from sixty seventy eighty years ago, like think of 0:20:23.116 --> 0:20:26.156 B fifty two C one thirty like World War. 0:20:25.996 --> 0:20:27.916 Two planes still flying kind. 0:20:27.756 --> 0:20:31.836 Of still flying, yes, exactly, and they have like you know, 0:20:31.996 --> 0:20:35.076 thirty of one, fifty of another, one hundred of another one. 0:20:35.396 --> 0:20:38.076 And these things get break down, right, and unlike a 0:20:38.116 --> 0:20:41.996 Ford factory, there is no factory for seventy different products 0:20:42.036 --> 0:20:43.156 that they're carrying. 0:20:42.876 --> 0:20:46.156 Right, and presumably the factory they built in nineteen forty 0:20:46.196 --> 0:20:48.076 one to build this plane doesn't. 0:20:47.796 --> 0:20:50.716 Exist any It doesn't exact. Even the vendor might completely 0:20:50.756 --> 0:20:54.356 have disappeared, right, that made that misspecific component. So they're 0:20:54.396 --> 0:20:57.796 constantly battling with this challenge of an aircraft goes down, 0:20:58.036 --> 0:20:59.876 how can I fix it? How can I find the part? 0:20:59.916 --> 0:21:02.276 And there are thousands of parts in each of these aircrafts, right, 0:21:02.316 --> 0:21:04.436