Fireside Chat With Figure AI Founder Brett Adcock

Yeah, so my background is pretty straightforward. I've been working on building technology companies for 20 years now. So, what Figure does is we are looking to deploy autonomous humanoid robots. So, a humanoid is kind of like a human form. So, we have legs and arms and hands. And what we're trying to do is physical work similar to what humans do today. We have a big belief here at Figure that the whole world was built for humans. Really what we're doing here in a nutshell is we're trying to build an automation system, an autonomous automation system to be able to do human-like work. So, our robot is about 5'6. It's all electromechanical. And we have an AI-first strategy to introduce this into service to do physical labor. And over time, I have a strong belief that every human will own a humanoid, much like your car phone today where this robot will be able to do anything you want it to do physically. So, grab me a coffee, do my laundry.

I think the conventional wisdom that we feel from the media and the world is that we're going to put robots in and they're going to replace humans doing this work. When we spend the time going getting boots on the ground with our customers and walking into a really large warehouse or retail or manufacturing facility, it's literally the complete opposite. They are running out of humans to hire. They are turning over 100% per year. The jobs are sometimes very dangerous and really boring. They just can't find a way to automate the human mobility and dexterity that we see today with humans, and there's just no plan. They don't know what to do. And when you look at a larger company that we say is potentially one of our clients, and you look top-down through the factory floor, you'll see these bays of really high automation where they'll have ASRS conveyor systems or robotic arms and different things, and they'll be all over the place, all over the facilities and globally. And then between all that work, humans are bridging all that manually. They're moving bins and boxes or, you know, carts or whatever it is. And they have hundreds of thousands and in some cases a million folks that are just doing this globally, and the companies just don't know how to automate this work. And it's a labor crisis going on inside the organization of how do I stop refilling this role every 9 months? So what we see is I would say a fairly large market of relatively simple things that we can start doing, meaning we could be moving bins and boxes, pushing carts. So for us, that allows us to basically a really structured environment that we can get into. So we think at this point we'll be able to ship tens of millions of robots even in the US in a very narrow part of the world, like warehousing and manufacturing, I guess, that are just moving bins and boxes and helping basically keep the operations going. It's much larger than what I thought a year and a half ago in starting the company.

I would say, actually, mechanically, the biggest problem we've really faced here in the last year and a half is there's really not, like when we build a humanoid, we need a certain amount of functional things that need to be built in order to make it work. We have a lot of different types of software, like whether it's the operating system, the AI systems, the control algorithms that are written in C++, the firmware. We have a lot of areas of mechanical design, joints and kinematics and rotor and stator design, the motor design. We have a lot of sensors on the robot. There's over 200 PCBs on our next generation robot at this point. So there's just a lot of stuff: we have computes and GPU and battery systems. It's a lot of stuff. The biggest issue we face in the last year and a half is there's just nowhere, there's no good places to get really great supply chain across these areas. It's been... I've probably been off by like an order of magnitude in my expectation. Like we'd be able to buy electric motors and actuators and sensors and battery packs and maybe other types of electronics, but it's just not the case. If we wanted to keep mass down, keep performance up, make it really reliable, make it safe, make it high performance, we've had to design almost all of the robot. Like we've done all the actuators, all the battery systems, all the electronics, all the structures, like it's just everything we've designed. And even down to some of the sensors inside the actuator itself that are generally stuff that we thought we would never do even like 6 to 9 months ago, we're now designing. And it's basically been, we've been basically forced to. There's a lot of good news that will really emerge out of this where we can keep cost down, we can really control it. It's going to build a really great robot that's really reliable. But it's certainly been, to be honest, it's been extremely painful for us.