No Priors Ep. 58 | The argument for humanoid robots with Brett Adcock from Figure

1. 0:00 – 3:09

   Brett’s background

   1. SGSarah Guo0:00

      (instrumental music plays) Hi, listeners. Welcome to another episode of No Priors. Today, we're here with Brett Adcock, the founder and CEO of Figure AI, which is developing and delivering humanoid general purpose robots that can do unsafe and undesirable jobs. They recently announced a monster round of funding, 675 million, from Microsoft, OpenAI, Nvidia, Intel, and Jeff Bezos. Brett, thanks so much for doing this.

   2. BABrett Adcock0:31

      Yes, sir.

   3. SGSarah Guo0:32

      You have this wild company, uh, doing humanoid robots. You just raised, uh, almost $700 million. Can you talk a little bit about how you get from, uh, a farm to software to, uh, vertical takeoff and landing to ro- humanoids?

   4. BABrett Adcock0:48

      Pretty normal path. (laughs)

   5. SGSarah Guo0:49

      Yeah. That's what I did too.

   6. BABrett Adcock0:50

      Yeah. (laughs) Yeah. So my story started in, I grew up in Illinois on a third generation farm. And, um, it ended up basically at a pretty early age started coding and getting to software and building things. Um, and that basically has been now about 20 years of building companies. Um, a little over 10 in software and a little under 10 in hardware. Um, at one point I started a software company and sold it, and then I started a company called, um, Archer Aviation. We build, uh, electric vertical takeoff and landing aircraft. And then, uh, about 21 months ago, I started Figure.

   7. SGSarah Guo1:29

      Can we pause there for a second? Because most people aren't like, "Oh, I'll just start an aircraft company." Like, how do you go from a software business, feels less exotic to me, to that?

   8. BABrett Adcock1:38

      I grew up around a lot of hardware, and so I, um, I looked at hardware as like I really wanted for a long time to build hardware, build like, um, kind of E- areas of deep tech.

   9. SGSarah Guo1:49

      Mm-hmm.

   10. BABrett Adcock1:49

       Only way to really do that was like self fund your own venture and get it really moving. So after I sold, I sold Vettery in 2017, and right away I knew I wanted to build electric aircraft. And I actually went back down to-

   11. SGSarah Guo2:02

       Do you fly planes or something? And-

   12. BABrett Adcock2:03

       I'm super passionate about A- like A, fixing traffic problems. We, we have like, you know, half the world live in cities and traffic's just getting worse and worse. There's just been no... There's been no solution there. And two is, um, big believer in sus- sustainable transport. Uh, I think, you know, all transport besides rockets will move electric, uh, hopefully in our lifetime. So what we do at Archer is we build vertical takeoff and landing aircrafts. So aircraft are kinda like helicopters but fully electric. Uh, they can take off among helicopter landing pads, like inside of a city, and they can take you from here back to San Francisco in under 20 minutes.

   13. SGSarah Guo2:35

       I'm sold.

   14. BABrett Adcock2:36

       That'd be a dream (laughs) -

   15. SGSarah Guo2:37

       It is.

   16. BABrett Adcock2:37

       ... instead of driving for two hours. So, it was really hard business. I, I basically started the company out of University of Florida. Um, I did, I started, uh, in engineering at University of Florida and, um, basically built a lab there for the first two years and built aircraft. And then moved the company out to California about three or four years ago.

   17. SGSarah Guo2:56

       That's now a public company. It's on the clearance path. How do you go from that to humanoid robots?

   18. BABrett Adcock3:02

       Uh, it was a less of a leap from software into Archer than it was from Archer to Figure. (laughs)

   19. SGSarah Guo3:06

       Yeah. But what's the thesis for it?

   20. BABrett Adcock3:08

       The thesis here

2. 3:09 – 5:51

   Figure AI Thesis

   1. BABrett Adcock3:09

      is that if you assume the technologies are, are possible to build a humanoid robot, and just for listeners, humanoid means human form, so legs, arms, hands, um, and you can do basically human-like work with a humanoid robot, um, A, it's gonna be the biggest business in the world by probably order of magnitude. Uh, half of GDP is human labor. Like it order of magnitude bigger than like all of transportation market. It's just an enormous industry. Two, is we think we can have like significant economical value if we can basically have robots doing real work every single day. I think it'll be a age of abundance as it relates to goods and services prices. And three is I, I think we'll have a... Over enough time, I think we'll have an impact on the AGI timeline here. It distills down to perhaps the most important business I think I could be working on in my life. So I left 21 months ago to start Figure, and what we're trying to do here is commercialize humanoid robots. We're trying to get them in the market, uh, in a significant way, build a fleet of robots, and then build an AI data engine to train those robots on how to do useful, useful work.

   2. SGSarah Guo4:09

      Why does being humanoid matter? Right? I think there are plenty of people who work in robotics that say like, "Uh, like, bipedal is really unstable." There are lots of reasons that these humanoid, um, forms are not optimal for doing-

   3. BABrett Adcock4:21

      Mm-hmm.

   4. SGSarah Guo4:21

      ... lots of different work. We can have 10 arms and be stronger and whatever else.

   5. BABrett Adcock4:25

      I think there's like really two schools of thought of how to go about this problem. We can either rebuild like thousands or millions of special types of robots that do special use cases all over the world, or we can build a humanoid robot. And the reason we believe humanoid robots are the right solution is that the whole world was built for humans. It was built around humans, meaning, um, the way we look biologically had a significant impact on the way our environment looks. If we were 10 feet taller and we had like, you know, six arms, a lot stronger, the world would look a lot different. When people ask like, "Is the humanoid the right form factor?" it's the wrong question. Just the wrong question to be answered. Uh, the humans are, the humans like the ideal form, it's the wrong way to look at it. We're like a weird biological species. We're a weird form.

   6. SGSarah Guo5:07

      But that's the world that exists.

   7. BABrett Adcock5:09

      We built it so we can interact with it. If you go to Mars today, you're gonna wanna go grab coffee, you're gonna wanna walk around, you're gonna wanna like live in a habitat. You're gonna wanna do things. And you're gonna build a human world around it. And then so if you wanna automate work, you wanna build a general interface to that. You wanna build like the equivalent of like the keyboard and mouse to the internet in the physical world. That equivalent is a human form. You can do everything a human can. And the world was optimized specifically for us. It wasn't optimized for us to have two more arms or nine feet tall. It was optimized just for the average human.

   8. SGSarah Guo5:44

      Mm-hmm.

   9. BABrett Adcock5:44

      The average like non-expert human. So it's really clear to us here that the, the right

3. 5:51 – 7:36

   The argument for humanoid robots

   1. BABrett Adcock5:51

      form factor is the humanoid. You can build one hardware system that can do everything.

   2. SGSarah Guo5:55

      Mm-hmm.

   3. BABrett Adcock5:55

      Meaning you can spend all this money and time on this robot that can be amortized over like, you know, millions of different tasks and applications. And then conversely, let's say you don't believe that.... you gotta go out in the world and look at every single use ca- like, every single task and build a special robot for it. That special robot needs a company, needs a brand. It needs a team, it needs a culture. It needs to go raise capital. It needs ... the team needs probably an order of, like, 15 different skillsets that one single person can't do, like in a full stack software role. It needs firmware, and embedded software, an operating system, and rotor, and stator, and electromagnetics, and battery systems, and BMS systems, and power distribution systems, and thermals, and ... the list goes on. Controls, and AI, and everything else. And so they need to raise a lot of capital to sustain it. And then you gotta go build millions of those. How is that even tractable? (laughs) How, how, uh, you know, how do you, how do you do that across the whole world in all these, like, special use case ... And then you're gonna make a ro-

   4. SGSarah Guo6:53

      Well, it is, it is how the, the robotics industry works today. Specialized robots, right?

   5. BABrett Adcock6:59

      Yeah.

   6. SGSarah Guo6:59

      I'm not saying that's efficient, but-

   7. BABrett Adcock7:01

      Yeah.

   8. SGSarah Guo7:02

      ... uh, I think that's where the question comes from.

   9. BABrett Adcock7:04

      I don't know if we should be reasoning by that analogy.

   10. SGSarah Guo7:06

       Okay.

   11. BABrett Adcock7:07

       I think that's the wrong way to look at it.

   12. SGSarah Guo7:07

       One, one thing you and I have talked about before is like-

   13. BABrett Adcock7:09

       (laughs)

   14. SGSarah Guo7:09

       ... is this the right decade to build the company? Like, what made you say, like, "We have it now, we can do it"?

   15. BABrett Adcock7:15

       We need to prove it, and I think we've started to prove it on like a m- MVP level. Like, you can see the robot doing pretty useful stuff now, and it's been 21 months. Took the first year to h- hire enough employees to come here (laughs) to do stuff, so we're kind of been around for kind of a year, maybe a year and a half of like real useful, like, run rate in terms of like, um, what we've been able to accomplish.

   16. SGSarah Guo7:35

       Can I ask

4. 7:36 – 12:38

   Figure AI public demos

   1. SGSarah Guo7:36

      you to, um, just describe the two public demos so far and, like, why they're important?

   2. BABrett Adcock7:41

      Yeah. We've been doing, um, kind of like two divergent set of demonstrations to the world. The first is we, we do plan as a business to start launching into more kind of industrial solutions, like more like the corporate labor market, uh, you know, m- manufacturing, supply chain logistics, those type of areas. We think that'll allow us to, to build the AI data engine quicker because we're shipping robots faster and will ... it'll help us build manufacturing volumes quicker, which will help cost. Those are, like, the reasons why we're doing it. There's another market that we're extremely excited about, which is in the home. And the home is a really messy place. It's very unstructured. Everything's different. There's, like, a higher variance of failures where we're like, you know, if, if we drop like the number one dad cup at home or the number one mom cup, like, no gr- not great. If we drop a bin in, like, a warehouse, like, who cares? You know what I mean? It's a little, little bit different scenario. Also, safety is impacted. There's pricing compression as we move into the, uh, consumer world. There's just a bunch of stuff that happens. Um, so, so we, we've done a few demos so far. First is we've done, like, bin moving, very traditional industrial solutions roles where we're taking bins from pallets into conveyor systems. We're doing that fully autonomous end-to-end on our robots now, all bipedal. Um, and the second is we're doing, um, kind of full consumer level manipulation and, like, speech-to-speech reasoning. So we're able to talk with the robot. It's able to understand what we're saying. It's able to visualize over the scene. Uh, it's able to do useful works like go and grab things like an apple or certain plates or make a coffee in a Keurig. And it's able to do all of that end-to-end, not only autonomously, but only from neural nets. So it's taking in, uh, speech. It's taking in video feeds in the cameras. We're processing that on a, on a model and then in, doing inference, and then we're outputting trajectories across the robot. I would say, like, you know, as an entrepreneur, the one thing that we're always afraid of is hitting some, like, technical wall.

   3. SGSarah Guo9:26

      Mm-hmm.

   4. BABrett Adcock9:26

      (laughs) It's like, we go out and do this, and like, we, everything just slows. We hit this, like, you know, this upper bound, and we just can't push through. Um, we, we don't even know where the upper bound is right now, and, uh, that's what's really exciting for us and why we're really excited about the next tw- 24 months is we don't e- we, we don't even see it. We don't know where it's at, and we're still searching for that and moving as hard as we can to try to find it. Why is now the right time? Yeah, there's a few reasons. Uh, I don't think this was really possible five years ago. Looking back over the last, like, you know, even few decades, the, um, like, the power train or the system that we use is all electro-mechanical here, so it's batteries and motors. If you look at the amount of energy we have in the batteries or amount of, like, say, power torque density in the motors, those have improved significantly the last couple decades. We just wouldn't ... We didn't have the runtime 10, 20 years ago in the robot with, like, lithium ion batteries to make this really work.

   5. SGSarah Guo10:17

      Mm-hmm.

   6. BABrett Adcock10:17

      Uh, we didn't have the power out of the actuators, electric motors to make this happen. So like, a best analogy is like, you know, 10 years ago a Tesla went 100 miles, and now my Tesla goes 300 plus miles. It's because specific energy in the battery cell on a watt hour per kilogram has improved, you know, maybe a seven percent CAGR since, you know, last two decades.

   7. SGSarah Guo10:36

      That makes sense. Like, if it, you know, can't carry more than 20 miles or it can't carry-

   8. BABrett Adcock10:40

      It-

   9. SGSarah Guo10:41

      ... fast enough to go to the highway, it's just not a useful car.

   10. BABrett Adcock10:42

       Run like ... Yeah.

   11. SGSarah Guo10:43

       Yeah.

   12. BABrett Adcock10:43

       If it was really heavy and it runs like, you know, 10 minutes and it can't carry anything and the speeds of the motors are not very great, like, you just can't get anything useful.

   13. SGSarah Guo10:51

       Mm-hmm.

   14. BABrett Adcock10:52

       I think a second thing is locomotion controllers. Like, 10 years ago, humanoid robots, bipedal was a huge risk. Like, they were clumsy. They were slow. They were falling over. Um, the DARPA Robotics Challenge is a good example of that. Like, about, you know, about 10 years ago now, uh, you couldn't, like, look at it and envision that being, like, really useful, or at least in your home. Uh, that've completely changed here. We actually started walking our humanoid r- uh, from the t- ... From the time I filed the C-corp in the business and we, uh, walked the robot, it was under a year from when we started, which was a pretty impressive feat. Um, and I, I, I think we, we probably ... Or we're probably doing some of the, like, some of the best work in the world on bipedal locomotion here on a, from a controlling, controller standpoint. And the third is, like, basically AI systems, like the, the computation, the algorithms, um, were just not feasible to do, I would say, 10 years ago. Um, and I can say arguably the fourth would, which would be, uh, you know, we, we spent a lot of time networking with OpenAI as we, we announced. And, um, we believe the default user interface to the robot is speech. Um, you, you're gonna want to talk to the robot. We've ... Even in an industrial setting-

   15. SGSarah Guo11:58

       (laughs)

   16. BABrett Adcock11:58

       ... when you're unboxing the robot for the first time, we think the, the initialization process is, is speech.

   17. SGSarah Guo12:03

       Hmm.

   18. BABrett Adcock12:03

       There's no, um ... There's no, like, you know, opening up your phone-

   19. SGSarah Guo12:05

       Configuration, demonstration.

   20. BABrett Adcock12:07

       Yeah, you're talking to it.

   21. SGSarah Guo12:08

       Yeah. Hmm.

   22. BABrett Adcock12:09

       And I think, um-... and, I mean, humans, that's what we're doing today, right? We, we, we do gestures when we talk, um, either through, you know, written text or we're- we're speaking. Um, whether it's transcribed or not, we're- we're- we're- s- we're- s- we're, like, using language, and we think that's primarily the main user interface, we think, uh, by default, the robot's gonna be using. Uh, and, you know, five years ago, that wasn't possible, either.

   23. SGSarah Guo12:33

       You said you- you- the team wasn't sure where the bounds were. Like,

5. 12:38 – 15:20

   Mitigating risk factors

   1. SGSarah Guo12:38

      where are the walls you can imagine hitting, and what were you afraid of when you started?

   2. BABrett Adcock12:42

      I mean, this is just a hard problem. There's just, uh, if you're afraid of walls, this is, like, the nightmare scenario for anybody, um-

   3. SGSarah Guo12:49

      It's just a, it's just a fun house of walls.

   4. BABrett Adcock12:51

      Yeah. We're- we have this, like, one of our five corporate values is, like, aggressively optimistic. (laughs)

   5. SGSarah Guo12:56

      We looked at some slides together, um, where you were looking at one of the trades and some risk on a component, and you had these mitigating factors-

   6. BABrett Adcock13:05

      Yeah.

   7. SGSarah Guo13:05

      ... and the second mitigating factor was sprint harder. Like, we will figure it out.

   8. BABrett Adcock13:09

      Yes. Like, work harder.

   9. SGSarah Guo13:10

      Yeah.

   10. BABrett Adcock13:10

       Uh, and that's-

   11. SGSarah Guo13:11

       Love that.

   12. BABrett Adcock13:11

       ... that's the only solution to the- to the- to the risk at this point. Um, the- the- the risks are profound. Like, um, nobody's been able to, um, deliver a commercial humanoid into a market in human history. Um, we have to e- n- not only, like, like, the threshold is we have to do a certain amount of an equivalent of humanlike work performance, which is extremely hard. Humans are very productive. Uh, and we have to do that reliably and continuously o- over the course of months and years.

   13. SGSarah Guo13:45

       Mm-hmm.

   14. BABrett Adcock13:46

       Um, and then to add onto all that, our robot has, like, you know, over 30 degrees of freedom, like, you know, joints that can move on the robot, and, you know, the amount of a- like, the amount of the action space or the amount of orientations the robot could possibly be in is- is extremely high. Either you have to code your way of saying this is w- you know, I have to write C++, or I have to write a script for everything the robot should do.

   15. SGSarah Guo14:11

       It's not gonna happen. Yeah.

   16. BABrett Adcock14:12

       Everywhere in the world, you have to- you have to solve that through robot learning, um, is the short answer. Uh, so and we haven't seen that work extremely well in human history, either. I mean, we're- we're watching that unfold right now with self-driving cars. We have to take all of those challenges on hardware never having been done before, uh, reliability and safety and performance, and then we have to learn everything. So yeah. To get back to your question, it's like they're all there. They're all, like- like, they're all, like, in the shadows, all these problems. There's all these o- other shadow of, like, unknown unknowns that you hit at every single, you know, at every single time. The reason why we don't feel like we see an upper bound is 'cause now we kind of see the roadmap for the next 12 to 24 months, and we're kind of just optimizing. We're just making the robot, like, more reliable and faster. We're not trying to get the robot to do, like, an end-to-end application for the first time. We've already done, we've done that, and we've done that for our first clients' use cases. It just has to be more robust.

   17. SGSarah Guo15:11

       Zero to 100 or zero to, uh, you know, walking bipedal robot in a year since you incorporate, how do you assemble a team that's so multi-domain

6. 15:20 – 16:38

   Designing the org chart and finding the team

   1. SGSarah Guo15:20

      so quickly?

   2. BABrett Adcock15:20

      So hasn't been my first rodeo. Uh, having built the, like, you know, built the all- like, all my teams previously at Bettery and Archer. Um, when I started Figure, there was a few things that no matter, like, um, what company it was, I- I would do and I would do again. Uh, the first is we set the mission, vision, values. I then wrote a master plan, which is basically like a 10-year journey over what we're trying to do. I wrote a culture document, which also is online, about, like, what we stand for here. Like, we like to move fast. We, you know, we do this. We don't do this.

   3. SGSarah Guo15:53

      We're aggressively optimistic.

   4. BABrett Adcock15:55

      Yeah, exactly. We, like, we work hard, things like that, we have to say. Um, we work in the office every day, five to seven days a week. I spent a l- I spent basically the first year, like, hyper focused on building the team. How do we build, like, the world's best engineering team, and who are those people? And I built an org chart. You know, I'm at the- I'm at the top, and we basically built out the teams with- in detail of, like, who this- all these groups should look like, whether it's, you know, controls, AI, um, actuation, battery systems, kinematics, integration and tests, industrial design, all of this, and then the skill sets underneath there. So it could be motor. We have like a, you know, rotor, stator, transmission,

7. 16:38 – 20:41

   Deployment timeline

   1. BABrett Adcock16:38

      sensors, thermals, motor controls.

   2. SGSarah Guo16:41

      That all makes sense as, like, a picture of it-

   3. BABrett Adcock16:43

      Yeah.

   4. SGSarah Guo16:44

      ... but then, like, the reality of somebody who spends a lot of time recruiting for-

   5. BABrett Adcock16:47

      Okay.

   6. SGSarah Guo16:47

      ... early stage companies is like-

   7. BABrett Adcock16:48

      Yeah.

   8. SGSarah Guo16:48

      ... I can't just go, like, pick that guy-

   9. BABrett Adcock16:49

      Okay.

   10. SGSarah Guo16:49

       ... up from Boston dynamics 'cause I- I-

   11. BABrett Adcock16:51

       Okay.

   12. SGSarah Guo16:52

       ... decided he's the right guy.

   13. BABrett Adcock16:53

       Yeah. So I then went out and found everybody online that w- I thought was the best in the world, and then I did 300 phone calls over six months, and I cold emailed all of 'em.

   14. SGSarah Guo17:05

       So Jerry picks up, and he's like, "Sure, Brett."

   15. BABrett Adcock17:08

       They don't say sure on the first call. I wish they did that.

   16. SGSarah Guo17:10

       (laughs)

   17. BABrett Adcock17:10

       Um, so a few ph- calls later, yeah, or a few meetings later, they do. Yes. Uh, or a certain percentage will. Yes. This is no different than what I did at Bettery. I built, you know, the f- the first few hundred like this at Archer. Um, and yeah. The first, you know, 30 to 50 over the first year, I, um, I identified the role in the org chart, what skills were necessary. I went out and found the right ideal, uh, person. I cold emailed. I did phone calls. I closed them. I gave them offer letters. Um, I wrote their 30-60-90s. I brought 'em in. I worked on them with a shared vision of what to do, and I worked with them day to day next door to them. I literally s- I literally sit right there on the floor with everybody else, and I attend every engineering meeting. And I work with them on designing the product and making those trades locally on speed and timeline and what to do and help them sh- help people ship.

   18. SGSarah Guo18:03

       Entrepreneurs, you heard it here. Just, just follow the plan. (laughs)

   19. BABrett Adcock18:06

       Yeah, just follow the plan. It was just a lot of hard work, but we have, um... we have, uh, e- you know, you really want to build an incredible team. Then you want to build a really great product, and then you want to get that product to market that's really great, and you want people to really like it.... and then you want them to keep using it over time. Like, that's the secret. And so everything starts axiomatically with, like, where is that team and how do you go get them? So, um, it was a brute force effort-

   20. SGSarah Guo18:29

       Great.

   21. BABrett Adcock18:29

       ... to say the least.

   22. SGSarah Guo18:30

       My favorite, my favorite type of strategy.

   23. BABrett Adcock18:32

       Yeah (laughs) .

   24. SGSarah Guo18:32

       Aggressively optimistic, how far away are we from, like, you know, companies buying robots en masse, deploying them?

   25. BABrett Adcock18:40

       So, companies are, you know, paying us now for robots, um, as we're, like, delivering them this year. We hope over the next 12 to 18 months, our robots are in our clients' test, like, real facilities, doing real work in real work cells, getting paid for it. And I feel decently confident-

   26. SGSarah Guo18:57

       For our audience, what's a work cell?

   27. BABrett Adcock18:59

       Um, just, you know, like, I, I'm at BMW and I'm supposed to be, you know, moving a bin or a box. This is your work cell, and we're moving bins over here to a conveyor system. That's, that would, uh, define a work cell. Like, I need to be in this location. I need to be doing this job. Uh, it's basically a job or a task. So I feel pretty confident in the next 12 to 18 months we'll start doing those. Um, and then, then we gotta make it reliable, uh, extremely reliable, extremely safe as we, like, you know, branch out to hundreds and then thousands on, you know... in, inside of, say, a factory floor. Um, and then we gotta manage a fleet of it, and then we have to do AI training at scale. Um, and then updates that to, to, to that fleet at scale. Um, and then we have to manufacture at scale.

   28. SGSarah Guo19:39

       No problem.

   29. BABrett Adcock19:40

       Yeah. I think, like... I think if we can solve the robot humanoid doing useful stuff, the other things are ex- extremely doable. It's extremely doable to take a robot that is reliable and that can do the performance and make a lot of them.

   30. SGSarah Guo19:53

       A lot of these other problems have been solved before.
8. 20:41 – 23:04

   Build vs buy and vertical integration

   1. SGSarah Guo20:41

      you are as vertically integrated as I've ever seen. You have your own actuators, you wrote your own operating system. Like, why do that when the project itself is so huge to begin with?

   2. BABrett Adcock20:52

      The trade of, like, build versus buy, like, should we do this ourselves or should we, um, go buy them, is, um... it's something we do at every single, like, like, component level.

   3. SGSarah Guo21:02

      So it's not a philosophy of, like, we're gonna build everything-

   4. BABrett Adcock21:04

      Oh, no.

   5. SGSarah Guo21:04

      ... from scratch to begin with?

   6. BABrett Adcock21:05

      No, no. I think our philosophy is we would rather, um... or yeah, our philosophy is that we would rather not do-

   7. SGSarah Guo21:10

      Mm-hmm.

   8. BABrett Adcock21:10

      ... do it, and we'd rather buy. But you ended up building- Yeah, we ended up building-

   9. SGSarah Guo21:13

      ... a huge amount.

   10. BABrett Adcock21:13

       ... everything.

   11. SGSarah Guo21:14

       Yeah (laughs) .

   12. BABrett Adcock21:14

       Yeah (laughs) . Uh, we started buying by... like, we bought a... yeah, the mix of-

   13. SGSarah Guo21:17

       Basically everything but the GPU.

   14. BABrett Adcock21:19

       It's... it feels like that, yeah.

   15. SGSarah Guo21:20

       Yeah.

   16. BABrett Adcock21:20

       Everything maybe besides, like, the battery cell and the GPU and the CPU, at this point-

   17. SGSarah Guo21:24

       Mm-hmm.

   18. BABrett Adcock21:24

       ... it feels like we've done-

   19. SGSarah Guo21:26

       Mm-hmm.

   20. BABrett Adcock21:26

       ... or we're doing. And to be clear, our default is to buy. Like, building is extremely difficult. Any part that we have to go build, we have to have, like... we have to put job listings out, we have to go hire people, we have to manage humans and make sure they're happy.

   21. SGSarah Guo21:37

       Mm-hmm.

   22. BABrett Adcock21:38

       We have to make sure their performance is good. We have to push, you know, push product. We have to check it. Uh, we have to integrate it in. H- it has to not break anything. We have to, uh, maintain it when it gets broken. Uh, we have a supply chain to go manage, like... um, it's like... it's a mess (laughs) .

   23. SGSarah Guo21:53

       So eyes wide open-

   24. BABrett Adcock21:54

       Um-

   25. SGSarah Guo21:54

       ... like, how do you end up?

   26. BABrett Adcock21:55

       There is no mature supply chain for what we're doing, and there is no other option.

   27. SGSarah Guo21:59

       Mm-hmm.

   28. BABrett Adcock22:00

       We have this philosophy that we have at the... we do, like, a nine o'clock standup every day on the, um... like, every morning, like rain or shine. And during, like, bring-up processes where we're bringing up, like, new robots and stuff, it's a mess. Like, the robots, like, never really work well on the first time. Everything breaks because we're, like, getting all the systems to start working together. There are things on software and hardware that haven't communicated before. There's just nothing, um, available that we could go buy that would satisfy our needs.

   29. SGSarah Guo22:25

       Mm-hmm.

   30. BABrett Adcock22:26

       So we've been forced to go build. And, like, design and then, you know, in a lot of cases, we even manufacture.
9. 23:04 – 28:37

   Product management at Figure

   1. BABrett Adcock23:04

      like, an iterative design approach. We really don't believe on l- spending a lot of time, like, just, just doing research and analyzing. We spend a lot of time on just testing, bui- building and testing-

   2. SGSarah Guo23:13

      Mm-hmm.

   3. BABrett Adcock23:14

      ... uh, here. And, um, that helps us really shake out all the problems. It helps us learn. It helps us recursively add it into a continuum of product that's coming down, uh, coming out. And, um... so first, that's our strategy. We, um... we want to be continuously updating the hardware and software forever. It'll... I don't think it'll ever be good enough for us. Um, so we have a whole process built around building a robot from a des- like a basically hardware and software design that we run here. We first set out with understanding who are the customers, like, what does the robot need to do? From there, we, uh, we basically set requirements, like, "Okay, we need the robot to lift this much pounds. It needs to run this long. It needs to charge here." The safety requirements are that it can't... battery can't burn down the building. There's a bunch of stuff we have to... um, the environment on IP rating has to be done on all the actuators. There's just a bunch of requirements that come from there. From there, we look at those requirements and we, we do engineering design. And we have basically-... like, three big phases. Uh, we have a conceptual and preliminary and critical design review that we do here throughout the year. The whole company's involved. So we, um, have these, like, design gates that we work through. Similar practice that I instituted, exactly similar, well, similar practice I instituted at Archer from a, from an engineering design perspective, or philosophy. And, um, yeah. We work, we work through in a very methodical way, like, uh, all the way through that serially.

   4. SGSarah Guo24:42

      And how, how does, um, integration and testing work in a way that's different from a software company, since you've also done that?

   5. BABrett Adcock24:48

      We tr-

   6. SGSarah Guo24:48

      I'd imagine really differently.

   7. BABrett Adcock24:49

      Yeah. We try to, we try to test and... We try to prototype and test as fast as we can to see if we were right.

   8. SGSarah Guo24:55

      Mm-hmm.

   9. BABrett Adcock24:55

      Same with software. It just happens on a longer timeline.

   10. SGSarah Guo24:57

       Okay.

   11. BABrett Adcock24:58

       Well, software, you'll come in one day, um, and you'll, and I'll say, "Okay, um, we talked to the client, we believe the cli- we've, we talked to the client, we believe we have all these things on the product backlog list we wanna do." You'll somehow have some heuristics where you'll score those, and you'll basically comb the backlog and you'll say, "I'm gonna go... We're, we're gonna add these, like, six things to the sprint." You'll do story points, and you'll basically, you'll, you'll assign those out and you'll basically manage that whole process. And then you'll launch it-

   12. SGSarah Guo25:22

       Mm-hmm.

   13. BABrett Adcock25:22

       ... and then you'll get feedback, right? You'll try to either A/B test things or you'll watch the analytics, and you'll say, "Did that work? Did that work?"

   14. SGSarah Guo25:28

       Mm-hmm.

   15. BABrett Adcock25:28

       Uh, you, you really wanna do that. You wanna have a, kind of a scientific method around it, say like, "Okay. Was that... Did that actually help, you know, fix this problem?" Uh, same here. We have the client, we have requirements that we set, like they need to do this. We are designing things, say, like, we are har- designing hardware from scratch, like, and, um... So we take our... If we're designing an actuator, we're gonna take our CAD system and we're gonna, from scratch, design it. We're gonna make, uh, assumptions on, and trade studies on, like, what the different trade-offs are of how we could do it up front, so we don't spend a lot of time designing something that just didn't work.

   16. SGSarah Guo26:01

       Mm-hmm.

   17. BABrett Adcock26:01

       Uh, so we're gonna be pretty methodical about it, like, much more methodical than you are a software, because the timelines are, you know, order of magnitude plus longer.

   18. SGSarah Guo26:07

       Yeah.

   19. BABrett Adcock26:08

       And then-

   20. SGSarah Guo26:08

       I think, to me, that's the key. Because there's a lot of, like, so- you know, you have design choices and you're like, "Uh." (laughs)

   21. BABrett Adcock26:13

       Yeah. No, yeah, you have to... Yeah, exactly. You, uh-

   22. SGSarah Guo26:16

       "I like this framework." (laughs)

   23. BABrett Adcock26:17

       Yeah. No, you can't... Yeah. It's like, um... You have to, you have to be very objective about those decisions. You have to say, "Okay, from actually a design perspective, there's, like, there's, like..." You know, at a very high level, "Do we wanna have hydraulic systems?" Um, you know, like, is it pneumatic? Is it electro-mechanical? Like, all these different ways of, like, say, powering the joints. And then from there, we can go to, "Okay, we want it to be like a, like a electric motor driven, or electro-mechanical driven actuator." Um, "Do you want it to have a, like a linear drive or rotary drive?" Um, we have a lot of rotary drives on our, um, on our, on our, um, on our, um, our robot. And then from there, "Okay, how is..." Like, if you look at the cross-section of the actuator, if you cut it in half, "How are we actually gonna design it?" Um, and, "What are the requirements? Like what does it need to actually do when we actually make it?"

   24. SGSarah Guo27:04

       Mm-hmm.

   25. BABrett Adcock27:04

       So, so all of those are, like, top-down, uh, driven. And, uh, we do that through trade studies and requirement setting, and, and then iterative design is the process of picking those choices, building it as fast as you can, and then going back and saying, "D- did it actually accomplish what I wanted it to do?"

   26. SGSarah Guo27:22

       Mm-hmm.

   27. BABrett Adcock27:22

       That timeline is, like, 10 times longer than software.

   28. SGSarah Guo27:25

       Right.

   29. BABrett Adcock27:25

       Maybe 100 times in some cases.

   30. SGSarah Guo27:27

       Yes.
10. 28:37 – 31:58

    Corporate partnerships

    1. BABrett Adcock28:37

       So we, um, we announced BMW a few months ago. They're, um, our first announced commercial customer. They are buying robots from us to ship into their manufacturing facilities. So the first facility we're, we're gonna launch into is Spartanburg, South Carolina. It's their largest facility in terms of f- like, vehicle production globally-

    2. SGSarah Guo28:56

       Mm-hmm.

    3. BABrett Adcock28:57

       ... um, which is great.

    4. SGSarah Guo28:57

       On order of, what, magnitude?

    5. BABrett Adcock28:59

       They make, like, uh, like, 1,400 cars a day.

    6. SGSarah Guo29:02

       Okay.

    7. BABrett Adcock29:02

       Um, and then that, that plant is higher, like, in terms of car productivity than any other plant globally. Um, happens to be in the US, which is great. We can hop on a quick plane, go over there. And we hope over the next 12 to 18 months, we're, we have robots in that facility doing real work.

    8. SGSarah Guo29:18

       Mm-hmm.

    9. BABrett Adcock29:18

       We've already picked out the first five use cases. We've already also then, um, chronologically ranked them in terms of what we're gonna start on first. Um, and we're already doing the first one. We're, like, actively working on, um, doing that fully and end to end reliably right now.

    10. SGSarah Guo29:33

        What can you, what can you say about the OpenAI partnership? It obviously makes sense that if you wanna communicate with robots with, uh, speech and you want them to have world model and reasoning-

    11. BABrett Adcock29:43

        Yeah. Yeah, we're super excited to be working with OpenAI. They've been really great. They started out in robotics, and some of the team that's working on the project with us were, were from that period of time, which is really cool. The highest level, we're working with OpenAI on building new model- AI models now for our robots to ship into commercial use cases. And what OpenAI brings is, they, they, I mean, they have the best vision language model in the world. And they have the best team in the world to work on the implementation of that. And we're working with them on trying it to, to do language reasoning on the robot. So think of this, like, the-... you know, like two, two parts of the brain. We're just a brain robot. We have a brain, like a centralized brain that you can talk to. And you can say, "I need you to go fold the laundry." Uh, that brain will then build tasks. It'll say, like, "I need to go do this, this, and this. I need to go grab the hamper." Um, we then build a path to go, you know, go find the hamper. We go, uh, we go to grab it, like, you know, like, OpenAI doesn't know how to grab it. It doesn't know how to command the robot.

    12. SGSarah Guo30:45

        Mm-hmm.

    13. BABrett Adcock30:46

        So we come in where we've designed AI systems here at Figure that can command the motors and the hands and the rest of the system to be ab- be able to do that only with neural nets.

    14. SGSarah Guo30:57

        Mm-hmm.

    15. BABrett Adcock30:57

        So we're kind of like combining forces of, like, kind of the, you know, we're doing like a low-level reasoning, and then we're doing, like, the language reasoning at the highest level.

    16. SGSarah Guo31:04

        Yep.

    17. BABrett Adcock31:04

        And we're combining those two together, and it's the more we spend time in this area, the more we feel it's, it's needed to be able to do this at real scale. I don't think there's really a way to, to do this without a, a really intelligent VLM sitting at the top. We have now kicked off a process to be able to, um, start integrating those and building new models from scratch. We put out a video a few weeks ago of working with them where we were basically doing speech to speech reasoning. We could talk to the robot. It could talk back. It could ask what's in the scene. It could, like, um, it could understand, you know, through memory, like, what happened in the past and implications it was making, you know, going forward. It was just, it was, um ... And that was like, you know, 13 days after, so after the announcement. Uh, so, you know, over the next six to 12 months, we're really excited to be working on, uh, these custom models with them.

    18. SGSarah Guo31:49

        Speaking of robots with memory that you can talk to, uh, what do you expect people will do with robots in the home?

    19. BABrett Adcock31:55

        Well, I think-

    20. SGSarah Guo31:55

        What are you gonna do? (laughs)

    21. BABrett Adcock31:56

        (laughs) What am I gonna do? Um-

11. 31:58 – 33:38

    Humans at home

    1. BABrett Adcock31:58

    2. SGSarah Guo31:58

       Yeah.

    3. BABrett Adcock31:58

       Well, what we want our humanoids to do is to do, like, physical work, and we want them to be a generalizer- generalizable replacement for human labor. So what I'm gonna have them do is I'm gonna have them, over time, do my laundry, um, cook me dinner. I have, like, this ... Every day I get home, there's kids' toys-

    4. SGSarah Guo32:19

       Yeah.

    5. BABrett Adcock32:19

       ... everywhere. I need them to clean the kids' toys up every day. (laughs)

    6. SGSarah Guo32:22

       (laughs)

    7. BABrett Adcock32:22

       That, that'll be on the docket for task planning for the r- my robot every single day. I think over a long enough period of time, everybody will own a humanoid just to do work for them.

    8. SGSarah Guo32:31

       Mm-hmm.

    9. BABrett Adcock32:32

       And you will choose, like, like-

    10. SGSarah Guo32:34

        Gotta make them cheap?

    11. BABrett Adcock32:36

        L- labor will be optional, and you'll choose to do work or not. Um, yeah, we'll make them cheap. There's a lot of, um, preconceived notions that these will be really expensive. I, I do not think they'll be expensive. I think, um, you know, we're working on cost reduction and stuff now. Like, um, the robots we have now we've showed are, are expensive. Uh, they're not cheap. The work that we're doing now and into the future is working on trying to reduce cost, and cost is gonna come down to really affordable levels. It'll take time. A part of that, um, cost reduction curve happens whenever you get manufacturing volumes up to certain levels. When there's an experience curve, manufacturing volumes and costs follow. So there's a certain amount of cost reduction we can do from designing for manufacturing, and there's a certain amount of cost reduction we'll get with real scale.

    12. SGSarah Guo33:17

        Ford Model T school-

    13. BABrett Adcock33:19

        Yeah, exactly. (laughs)

    14. SGSarah Guo33:20

        ... of robotics. Yeah. (laughs)

    15. BABrett Adcock33:20

        Yeah, yeah. So, uh-

    16. SGSarah Guo33:22

        Right.

    17. BABrett Adcock33:22

        ... we just need to get robots shipped.

    18. SGSarah Guo33:23

        Okay.

    19. BABrett Adcock33:23

        And shipped in a big way. And if we can ship them ... We ... If we can ship millions into industrial solutions, uh, we can use that as our pathway for, uh, the data collection process and volumes on, uh, the manufacturing side.

    20. SGSarah Guo33:36

        How do you think about social acceptance

12. 33:38 – 35:41

    Social acceptance

    1. SGSarah Guo33:38

       of robots? What are you already seeing-

    2. BABrett Adcock33:40

       I knew-

    3. SGSarah Guo33:40

       You show people.

    4. BABrett Adcock33:41

       ... I knew you were gonna ask that. (laughs)

    5. SGSarah Guo33:42

       Yeah. I mean, I'm doing my best robot impression. I'm really excited, but ...

    6. BABrett Adcock33:44

       Yeah. Yeah, we also had this at Archer where it was like, you know, are people going to be okay with, like, things in the air over the city and things like aircraft taking off in your backyard? We had to think really long and hard about how we were going to do that and integrating that solution. I think this notion of social acceptance as it relates to safety and just robots in the world and everything else needs to be proven. Like, it needs to be shown and proven. Uh, I don't think somebody's gonna wake up one day and say, "Okay, I think I'm okay. It's a Friday, I think I'm okay with humanoids today." I think it's, I think you're gonna start gradually seeing robots in BMW, in these different industrial places. They're gonna be doing real work. They're gonna be building a safety record. Um, we're obviously not gonna be, like, having robots at BMW unsafe and then try to launch them in your home. It's just gonna take time to build that out, and, and to build that trust and to build the brand like. Um, so I, I, I think it'll be gradual, and I think that trust and that social acceptance has to be earned.

    7. SGSarah Guo34:39

       Mm-hmm.

    8. BABrett Adcock34:40

       So we think about it, everything we do, every time that everybody sees a robot, every time we put out a video, every time we're ... as we're launching into our com- commercial customers, like, we have to think about that from the very beginning. I think, you know, it's, it's even interesting, more interesting because, like, every sci-fi movie we've ever seen has, like, ended poorly in this area for humanoid robots. (laughs) You know? And so yeah, it's just like, it's never like, um ... Ne- never great. So, uh, there's a little bit of that stigma, which is almost like fantas- fanciful. Um-

    9. SGSarah Guo35:06

       Yeah.

    10. BABrett Adcock35:07

        Yeah.

    11. SGSarah Guo35:07

        There's always some contingent that's on the robot side. Uh, just, you know, for, for the record, if they win, I'm, I'm on that side.

    12. BABrett Adcock35:14

        Yeah, great.

    13. SGSarah Guo35:15

        Yeah. (laughs) Okay.

    14. BABrett Adcock35:15

        Yeah. I really want it to work out well.

    15. SGSarah Guo35:17

        Yeah.

    16. BABrett Adcock35:17

        And I think if it works out well, it'll be like ... It'll be really cool. It'll feel like, um, 50 years of the future got pulled forward today. It'll be just magical.

    17. SGSarah Guo35:26

        So given that, I have to ask you, like, AGI, does this make it come faster? Um, I think there are a, a bunch of people who think of the, the lack of, like, actuation for increasingly capable models as like the big

13. 35:41 – 38:00

    AGI vs the robots

    1. SGSarah Guo35:41

       actual safety barrier. But you're like, "Oh, here's the actuator," right? (laughs)

    2. BABrett Adcock35:44

       Yeah. Yeah, it's funny. It's like, who's gonna get to market first, the humanoids or AGI? Um, or are, you know, do we need humanoids to get there? My view here is that we need to get the humanoid robot figured out pre-AGI.

    3. SGSarah Guo35:57

       Mm-hmm.

    4. BABrett Adcock35:58

       And, uh, whe- whether we need the humanoid to get there, like, as it relates to the timeline for AGI or not, wh- what I think is a really quite dystopian future is if we have ... AGI's here, and AGI wants to do something. And it's gonna, like, com- it's gonna ask you or force you to go do it, whether it force you through money or it's just gonna, you're gonna have to go do the real work for it. And if you walk into, like, a warehouse today-

    5. SGSarah Guo36:22

       Like, we are the actuators for the, for the model.

    6. BABrett Adcock36:26

       If you walk into a warehouse today, or a manufacturing facility, everybody's getting told what to do from a computer to a, you know, to a barcode scanner or a phone.

    7. SGSarah Guo36:34

       Mm-hmm.

    8. BABrett Adcock36:35

       They're literally getting ... They're, they're scanning something. They're getting told what to do next from a warehouse management system. They're literally a cyborg.

    9. SGSarah Guo36:40

       Hmm, that's a little dark, yeah.

    10. BABrett Adcock36:42

        Yeah. And if we have, like, super artificial i- s- like super intelligence, um, what do you think that's gonna be like? My hope is that we can figure out the humanoid thing prior to that, and we can have humanoid robots doing all that work. It's, it's becoming more clear, I think at least to our team, that at least these large language models are having a really difficult time, like, reasoning around the physical world. Planning, actions, everything across the board. We, we kind of believe that, that over the next five or 10 years, we really hope to make a significant impact on the AGI timeline here at Figure. We think that information is extrem- like, coming out with a robot is extremely important to solving this last big piece into the AGI timeline. Jury's still out. (laughs) We will see. But we're, um, we're hopeful that we can, yeah, help here.

    11. SGSarah Guo37:36

        Awesome. No small plans. Small trades.

    12. BABrett Adcock37:40

        Yeah.

    13. SGSarah Guo37:42

        Find us on Twitter @nopriorspod. Subscribe to our YouTube channel if you want to see our faces. Follow the show on Apple Podcasts, Spotify, or wherever you listen. That way, you get a new episode every week. And sign up for emails or find transcripts for every episode at no-priors.com.

Episode duration: 38:00