Latest / INSPIRE STORY / # 0266_ The First Interview with a Robot — The Future of AI, Humanoid Machines & Humanity

Transcript

# 0266_ The First Interview with a Robot — The Future of AI, Humanoid Machines & Humanity · INSPIRE STORY

0:05
Speaker 1: Brett Adcock, welcome to the show.
0:07
Speaker 2: Thanks for having me on.
0:10
Speaker 1: I've been looking forward to this for a long time.
0:12
Speaker 1: The robotics guy.
0:13
Speaker 2: Yeah.
0:14
Speaker 1: Let me give you an intro here real quick before
0:16
Speaker 1: we get started. Brett Adcock, a serial entrepreneur and founder
0:21
Speaker 1: and CEO of Figure AI, building general purpose humanoid robots
0:26
Speaker 1: for labor automation. Founded Vetteri, an AI-driven talent marketplace, which
0:31
Speaker 1: was acquired for approximately $ 100 million. Co-founder of Archer Aviation,
0:36
Speaker 1: developing electric vertical takeoff and landing EVTOL aircraft. Found Cover,
0:43
Speaker 1: an AI security company using NASA jet propulsion laboratory technology
0:49
Speaker 1: to detect concealed weapons in K-12 schools. That's amazing. In
0:54
Speaker 1: late 2025, you launched Hark, a new AI lab self-funded
0:58
Speaker 1: with 100 million to build what you call human-centric AI.
1:03
Speaker 1: You've raised billions in venture capital and Time named you
1:06
Speaker 1: one of the 100 most influential people in AI in 2024,
1:10
Speaker 1: married and a father of three children. And before we
1:14
Speaker 1: get too far into it, we always start off with
1:17
Speaker 1: a gift.
1:39
Speaker 2: Thank you.
1:39
Speaker 1: He didn't give you any tips on that, did he?
1:41
Speaker 2: All right, I got to hit one.
1:52
Speaker 1: What do you think?
1:52
Speaker 2: They're great.
1:53
Speaker 1: You can leave this guy here if you want to.
1:55
Speaker 2: This guy's staying.
1:56
Speaker 1: That is the coolest thing I've ever seen as far
2:01
Speaker 1: as giving somebody a gift on this show. That was awesome.
2:04
Speaker 3: Yeah.
2:04
Speaker 2: That was awesome. I got you another gift.
2:07
Speaker 1: Oh.
2:08
Speaker 2: Something you keep here, put on the shelves.
2:12
Speaker 1: Thank you. Yeah. No way.
2:19
Speaker 2: Yeah, little robot.
2:20
Speaker 1: That is awesome. Thank you.
2:23
Speaker 2: Yeah, no problem.
2:25
Speaker 1: Very cool. Well, Brett, we've got a lot to talk
2:32
Speaker 1: about here. So, man. How many companies are you running now?
2:41
Speaker 2: Man, I'm not sleeping. I got too many.
2:43
Speaker 1: I'll bet.
2:44
Speaker 2: Too many. Just like kids and work and just like, yeah.
2:48
Speaker 1: That thing is amazing.
2:49
Speaker 2: Never sleeping anymore.
2:50
Speaker 1: I'll bet. I'll bet. Yeah.
2:52
Speaker 2: What'd you think of the robot?
2:53
Speaker 1: I think it's incredible. I can't wait to talk more
2:56
Speaker 1: about it. So a couple of things, just one more
2:59
Speaker 1: thing to knock out here before we get into it.
3:03
Speaker 1: I got a Patreon account. It's a subscription account. And
3:06
Speaker 1: it's quite the community. And they're honestly the reason that
3:09
Speaker 1: I get to sit here with you today. So they
3:11
Speaker 1: get the opportunity to ask every single guest a question.
3:14
Speaker 1: This is from Steven Casey. In today's marketplace, we find
3:17
Speaker 1: that AI platforms can sometimes invent answers rather than admitting
3:21
Speaker 1: to a lack of information. Combining this in the physical
3:25
Speaker 1: realm of robotic action seems to multiply the downside effects exponentially.
3:31
Speaker 1: What safeguards are in place that we can put our
3:34
Speaker 1: trust in to prevent the potential for downstream harm to
3:38
Speaker 1: humans as a result of bad programming or computing errors. Yeah.
3:43
Speaker 2: Yeah, we don't want the Terminator popping out here when we.
3:45
Speaker 1: Definitely not.
3:46
Speaker 2: Let this work, right? I mean, I think, like, we
3:50
Speaker 2: were chatting about this outside. Like, you know, I think
3:53
Speaker 2: one thing to say, like, four years ago when I, like,
3:56
Speaker 2: you know, started the company, there was, like, no path
3:59
Speaker 2: for humanoid robots. Like... to make it into people's homes
4:03
Speaker 2: in the next 10 years. There's no good story. You
4:07
Speaker 2: had big hydraulic humanoids out there. They were all hand-coated
4:11
Speaker 2: to do certain tasks. What you really need is a
4:14
Speaker 2: cheaper electric humanoid that you basically can use neural nets,
4:20
Speaker 2: use basically an AI-first strategy with. There was just none
4:22
Speaker 2: of that existed. I think we're thankful now, looking back,
4:27
Speaker 2: that we... It feels like we somehow pulled 10 years
4:29
Speaker 2: of the future forward. We have like electric humanoids at
4:32
Speaker 2: like a reasonably price that can do like a useful
4:35
Speaker 2: human work with neural nets. And it's just like, I
4:37
Speaker 2: think it's just an incredible, it's an incredible place to
4:42
Speaker 2: be in getting those questions, which is like, how do
4:44
Speaker 2: we make this work now at scale in a safe way? Because,
4:47
Speaker 2: you know, that's the spot we want to be in,
4:49
Speaker 2: not like trying to make this work for 20 years.
4:51
Speaker 1: Yeah.
4:52
Speaker 2: So I think it's a very, I mean, this is
4:55
Speaker 2: a very, very tough problem. We have to get the
4:58
Speaker 2: product cheap enough. We have to make enough of them.
5:00
Speaker 2: We have to make it like, but the performance work
5:02
Speaker 2: in like very complicated things, like walk around a house
5:06
Speaker 2: and like do dishes, like laundry, like very complex things.
5:11
Speaker 2: Like small kids can't do this. Like it takes adults
5:13
Speaker 2: to kind of do like this level of work. And
5:15
Speaker 2: we need that all done in a mechanical system that
5:16
Speaker 2: doesn't have any humans around for maybe most of this
5:19
Speaker 2: that does it autonomously and not makes any mistakes. And
5:22
Speaker 2: then like, like, like your fan mentioned, like we have
5:24
Speaker 2: to do it safely over time. It's just, man, it's
5:26
Speaker 2: just an incredibly complex problem. I think for us, we
5:30
Speaker 2: have a safety strategy, both intrinsically, we want the robot
5:35
Speaker 2: hardware and the robots around humans to just be safe
5:38
Speaker 2: all times. And separately, there's a bunch of semantic safety
5:41
Speaker 2: and other things that we need, that we have either
5:44
Speaker 2: put in place or are putting in place now to
5:46
Speaker 2: make the robot just work safe in the environment. Like
5:49
Speaker 2: if you have a candle at home, you don't want
5:51
Speaker 2: the robot to accidentally knock it over. That's like an
5:54
Speaker 2: intelligence thing. in a lot of ways. Or there's a
5:58
Speaker 2: boiling pot of water, making sure we're very safe around it.
6:01
Speaker 2: And then there's the intrinsic safety of making sure this
6:04
Speaker 2: mechanical thing in your house is safe around everybody around it.
6:08
Speaker 2: I think the direct answer is there's still a lot
6:10
Speaker 2: of wood chop of getting this thing to a point
6:13
Speaker 2: where it's like, we trust it to be autonomous next
6:18
Speaker 2: to my kids all day long in my house.
6:19
Speaker 1: You have one in your house?
6:21
Speaker 2: We've had many robots now throughout my house in testing
6:24
Speaker 2: for the last year or so. And I've had them
6:27
Speaker 2: kind of near my kids in some aspects, but we're
6:29
Speaker 2: always monitoring it.
6:31
Speaker 1: What do your kids think?
6:34
Speaker 2: Man, it's just kind of normal for them now. Do
6:36
Speaker 2: they try to talk to them? Yeah, talk to it. Yeah,
6:40
Speaker 2: they want to go jump on it and touch it.
6:44
Speaker 2: And do kids things, you know what I mean? They
6:46
Speaker 2: want to go touch it and talk to it and
6:48
Speaker 2: be around it. And, uh, we're still not at that
6:50
Speaker 2: stage yet where I feel comfortable enough to, like, let
6:52
Speaker 2: loose and say, hey, you know, here's a robot. My
6:54
Speaker 2: kids are there and I feel okay. And we're not
6:56
Speaker 2: there yet. I think we will be in the next
6:58
Speaker 2: several years.
6:59
Speaker 1: What's the longest they've been around any one particular robot?
7:03
Speaker 2: Um, we've had a robot in my house for, like,
7:08
Speaker 2: maybe a couple months, uh, doing work kind of on
7:11
Speaker 2: and off, you know, daily, sometimes every other day. And, um...
7:16
Speaker 2: You know, the kids are kind of at school or
7:17
Speaker 2: sometimes at home. So, they weren't always around whenever the
7:20
Speaker 2: robots were running, but a lot of times. And, you know,
7:24
Speaker 2: that was just like our home robot.
7:25
Speaker 1: Do they get attached to them?
7:27
Speaker 2: They named it. Emotionally attached? They all had different names
7:29
Speaker 2: for the robot. And yeah, they love it. And it's
7:32
Speaker 2: actually a question where I was in an office of like,
7:34
Speaker 2: you have a robot in the home and it's like,
7:36
Speaker 2: it's got some like character to it, a little wear
7:38
Speaker 2: and tear. Do you want to keep that robot or
7:39
Speaker 2: do you want a new one?
7:40
Speaker 1: That's what I'm wondering. What's the emotional attachment? I think
7:44
Speaker 1: kids are the perfect test case.
7:46
Speaker 2: My kids wanted it. They wanted it there. We're not
7:49
Speaker 2: getting rid of this guy? Yeah, it's got a little
7:51
Speaker 2: banged up a little bit here and there and it
7:52
Speaker 2: has a tear here and they just loved it.
7:55
Speaker 1: That's wild, man. That is wild.
7:58
Speaker 2: Honestly, in our lifetime, we will be fortunate enough for
8:00
Speaker 2: every human to, I think, have a humanoid. Almost like
8:03
Speaker 2: a phone and car.
8:04
Speaker 3: Wow.
8:05
Speaker 1: Yeah, we were talking, just some of the stuff that
8:08
Speaker 1: you just mentioned. I mean, the complexity of the problem
8:12
Speaker 1: that you're solving here. I mean, all these little problems
8:15
Speaker 1: that I didn't even like knocking over a boiling pot
8:17
Speaker 1: of water. I never would have like, just thinking about
8:21
Speaker 1: something that it happens every day. And then you think
8:24
Speaker 1: of all the things that happen every day in just
8:26
Speaker 1: a regular household.
8:28
Speaker 2: It's like problem city, man. It's like a fun house
8:30
Speaker 2: of problems. There's just problems everywhere. It's like hardware problems.
8:34
Speaker 2: AI problems, problem scaling and commercializing and getting the system reliable,
8:39
Speaker 2: manufacturing problems. We have a problem fun house if you
8:43
Speaker 2: want to come by campus here and check it out.
8:45
Speaker 1: I'll bet you do. I'll bet you do. Well, some
8:49
Speaker 1: people say AI is in an economic bubble. And as
8:55
Speaker 1: of this recording, PolyMarket says there's going to be an
8:57
Speaker 1: 18% chance that the AI bubble will burst by December 31st, 2026.
9:04
Speaker 1: What do you think about that? Is AI in a bubble?
9:07
Speaker 2: Absolutely not. I think you'll see some of the most
9:13
Speaker 2: transformative events in technology happen over the next 36 months
9:17
Speaker 2: we've ever seen in our... Ever.
9:20
Speaker 1: Ever. I don't feel like we're in a bubble here.
9:25
Speaker 2: I feel like we're barely scratching the surface. I'm watching
9:29
Speaker 2: AI in a human body do human work. Like early,
9:33
Speaker 2: it's early. We don't have, you know, we don't, at
9:37
Speaker 2: some point here this year, we'll have thousands of robots.
9:39
Speaker 2: We have like, you know, we have hundreds now. Like,
9:42
Speaker 2: we need like millions of robots to make an impact.
9:44
Speaker 2: That's just going to take some time and it's going
9:45
Speaker 2: to be crazy cool. So we're just, we're at the
9:48
Speaker 2: start line of that happening, which is like, how do
9:50
Speaker 2: we get AI out into the physical world at scale?
9:53
Speaker 2: That'll for sure work and it'll go really far in
9:55
Speaker 2: our lifetimes. And then separately, we have AI now that
9:59
Speaker 2: can use computers like humans and can think. I've shown
10:02
Speaker 2: you a little bit of that here before the show.
10:05
Speaker 2: And that will manifest in a point where both in
10:09
Speaker 2: the physical and digital world, you basically have these little
10:12
Speaker 2: mini humans that can do human-like work and they can
10:15
Speaker 2: think and use computers and use machines.
10:17
Speaker 1: And....
10:19
Speaker 2: I mean, that's going to lead to such a productivity.
10:23
Speaker 2: Like we measure like GDP per capita, like per human.
10:26
Speaker 2: But if you're able to make like as many synthetic humans,
10:29
Speaker 2: like millions, billions, tens of billions of synthetic humans, in
10:32
Speaker 2: the case of the digital world, maybe trillions, that'll lead
10:36
Speaker 2: to the, I mean, I think the greatest increase in
10:40
Speaker 2: productivity we've ever seen in our lifetime and ultimately like
10:43
Speaker 2: reduce goods and service prices to unprecedented levels. Like a
10:47
Speaker 2: true age of abundance.
10:48
Speaker 1: Wow.
10:49
Speaker 3: Wow.
10:50
Speaker 1: What do you, I mean, I'm just curious, what do
10:51
Speaker 1: you think, what will humans be doing?
10:54
Speaker 2: I mean, I hope I don't have to, like, I
10:56
Speaker 2: woke up today, I was like, I'm in the dishwasher,
10:58
Speaker 2: getting my kids breakfast, like, just like busy work that
11:01
Speaker 2: I think my kids are sitting there. I'm like doing work.
11:03
Speaker 2: You know what I mean? Like, I wish I was
11:04
Speaker 2: just like, yeah, I just, uh, wish I wasn't doing
11:07
Speaker 2: that stuff. And then I'm like, uh, all throughout my day,
11:09
Speaker 2: I'm like trying to call, you know, call the car
11:12
Speaker 2: service and then trying to get on my flight and,
11:14
Speaker 2: you know, coming here and it's like ordering lunch. Like,
11:17
Speaker 2: all the stuff I'm doing all day. And I don't
11:20
Speaker 2: want to do any of that. I want to be
11:21
Speaker 2: like fully free.
11:23
Speaker 1: I get it.
11:23
Speaker 2: All that burden.
11:24
Speaker 1: No, I totally, I love it.
11:25
Speaker 2: And I just want to be like clear headed. And
11:26
Speaker 2: I want like my AI to run a little Brett
11:29
Speaker 2: Adcock operating system and run my life. And all these
11:32
Speaker 2: things I have in my head about what to order
11:33
Speaker 2: and pay his tax bill and like do this meeting.
11:35
Speaker 2: And I have to go back and do an engineering standup.
11:37
Speaker 2: I want all that stuff to be in my like
11:41
Speaker 2: operating system and like a human in a box.
11:45
Speaker 1: So you're basically saying The way this is going to
11:48
Speaker 1: turn out is your brain, I'm going to butcher this,
11:51
Speaker 1: you're basically exporting your brain and all the tasks that
11:54
Speaker 1: are going on in your brain. You're disseminating it to robots.
12:00
Speaker 1: I'm going to.
12:00
Speaker 2: Delegate all this out to robots.
12:02
Speaker 1: That's amazing.
12:04
Speaker 2: We'll do that in 24 months. We'll have.
12:07
Speaker 2: All this stuff's so good that you won't go order
12:11
Speaker 2: food anymore, book stuff, do a lot of work behind
12:14
Speaker 2: a computer, physical stuff in the world of doing laundry
12:18
Speaker 2: and dishes.
12:19
Speaker 1: Just the bullshit legwork.
12:20
Speaker 2: Yeah. Does anybody want to do that? I don't want
12:22
Speaker 2: to do it.
12:23
Speaker 1: I don't.
12:24
Speaker 2: Yeah. So you clear all that from my life. I
12:27
Speaker 2: got to spend time with my kids, enjoy life, kind
12:29
Speaker 2: of be, I guess, clear-headed. Yeah. Um, two stuff I
12:32
Speaker 2: really love. Like I love working, but I don't like
12:34
Speaker 2: doing all this busy work. Yeah. Um, it's just like not,
12:37
Speaker 2: it's just like manual, like just like labor I'm doing
12:39
Speaker 2: behind computers or like in the physical world. And just like,
12:41
Speaker 2: I want to delegate that out to my AI to
12:43
Speaker 2: do and fully automate it out.
12:46
Speaker 1: That's a different, I don't know why I've never thought
12:49
Speaker 1: about that. I've never thought about it. Like I've always
12:51
Speaker 1: looked at it as fear. I've always been like, oh shit,
12:53
Speaker 1: they're going to take everything over.
12:55
Speaker 2: It's a compression algorithm. Like we're basically running a large
12:58
Speaker 2: scale compression, uh, So I think the way I look
13:06
Speaker 2: at it now is we basically have built synthetic human
13:10
Speaker 2: intelligence that can use computers and machines. So I'm going
13:13
Speaker 2: to delegate out all this busy work on both my
13:16
Speaker 2: digital life and physical life to robots, and they'll just
13:20
Speaker 2: do all of it. But it's good. I mean, we
13:23
Speaker 2: have AI systems now. in our lab at Hark that
13:26
Speaker 2: can use computers like a human can. It can talk
13:28
Speaker 2: to you. It can like, I just, I made a
13:31
Speaker 2: phone call to ours before we started and talked about
13:33
Speaker 2: my schedule and how to ask for things and ask
13:36
Speaker 2: it for things and have it do things.
13:37
Speaker 1: Yeah, you had a chicken salad ordered to deliver to
13:40
Speaker 1: your office.
13:41
Speaker 2: Yeah, exactly. Exactly. But no, like nothing besides a single, like, hey,
13:45
Speaker 2: make this order. And you can spin up computers to
13:47
Speaker 2: do that virtually. And then physically, like, I'll have all
13:51
Speaker 2: this work done by robotics. Both in you'll have it
13:54
Speaker 2: in the commercial workforce and the billions like manufacturing and
13:57
Speaker 2: healthcare and construction and every human at some point will
14:00
Speaker 2: have a humanoid just to do all that busy work
14:02
Speaker 2: for you. And not only that, but like something to
14:04
Speaker 2: come home to that you can talk to that will like,
14:06
Speaker 2: will know you.
14:09
Speaker 1: Wild. Yeah.
14:10
Speaker 2: It's like the, yeah, it's going to happen now, which
14:13
Speaker 2: is like really going to be fun.
14:14
Speaker 1: Yeah. Yeah. I'm excited to introduce to you the newest
14:20
Speaker 1: member of my family. We call him Stanley. We got
14:24
Speaker 1: Stanley this past Christmas and pretty quickly my focus became
14:28
Speaker 1: making sure he was safe while still giving him the
14:31
Speaker 1: freedom to actually be a dog. I went looking for
14:34
Speaker 1: the top-rated GPS fence, the number one, and that's how
14:39
Speaker 1: I found Spot On. Stanley wears their Nova collar. With
14:44
Speaker 1: Spot On, I set up a GPS fence for Stanley
14:47
Speaker 1: right on my phone. No physical fence, no leash, I
14:51
Speaker 1: just walk my property line or draw it on the map,
14:55
Speaker 1: and that's the boundary the collar recognizes. I can create
14:58
Speaker 1: multiple fences, save them, and adjust them whenever I need to.
15:03
Speaker 1: So whether we're at home or traveling, Stanley always knows
15:07
Speaker 1: where his boundaries are. Another thing that stood out to
15:10
Speaker 1: me is these collars are designed right here in the
15:12
Speaker 1: USA and assembled in New Hampshire by a team that's
15:15
Speaker 1: been working with high precision GPS technology for years. And
15:20
Speaker 1: you can tell. A lot of attention went into making
15:23
Speaker 1: this thing reliable. The Nova Collar uses a dual-band GPS
15:27
Speaker 1: system connected to more than 150 satellites, along with an
15:31
Speaker 1: antenna that's over five times larger than typical GPS fence collars.
15:36
Speaker 1: That keeps the boundary accurate, even around trees, terrain, and
15:41
Speaker 1: changing conditions. SpotOn's True Location technology has been independently tested
15:47
Speaker 1: and delivers 99.3% containment. which matters when you're trusting something
15:53
Speaker 1: with your dog's safety. It's incredibly durable, and on top
15:57
Speaker 1: of that, I can check his location in real time,
16:00
Speaker 1: send voice commands directly to the collar, and track his
16:04
Speaker 1: activity through the day. I use Spot On so Stanley
16:08
Speaker 1: gets the freedom to run and explore, and I get
16:11
Speaker 1: the peace of mind knowing he's safe. Let your dog
16:15
Speaker 1: roam with Spot On. Go to spotonfence.com slash SRS and
16:20
Speaker 1: use code SRS for $ 50 off the Nova collar. That's spotonfence.com
16:26
Speaker 1: slash SRS and use code SRS for $ 50 off. Well,
16:34
Speaker 1: I would like to do a little bit of a
16:35
Speaker 1: life story on you. Does that sound good to you?
16:38
Speaker 2: Yeah, let's do it.
16:39
Speaker 1: Where'd you grow up?
16:40
Speaker 2: Central Illinois.
16:41
Speaker 1: Central Illinois?
16:42
Speaker 2: Yeah, like a small town, like 700 people.
16:45
Speaker 1: 700 people?
16:46
Speaker 2: Yeah.
16:46
Speaker 3: Wow.
16:46
Speaker 1: We call them a Wequa, Illinois. That's even smaller than
16:50
Speaker 1: where I grew up.
16:51
Speaker 2: Yeah. Where'd you grow up?
16:51
Speaker 1: I grew up in small town, Chillicothe, Missouri.
16:54
Speaker 2: How small?
16:55
Speaker 1: About 8,000 people.
16:56
Speaker 2: At the time. Yeah.
16:59
Speaker 1: Yeah.
16:59
Speaker 2: We didn't have, uh, we didn't have anything. 700 people.
17:02
Speaker 1: Man. Um, what were you into?
17:05
Speaker 2: Um, yeah. So, uh, you know, kids, sports, computers, like, uh,
17:10
Speaker 2: I got into computers really early, uh, did a bunch
17:13
Speaker 2: of sports. Um, you know, we, I grew up on
17:16
Speaker 2: a farm, so it was corn and soybeans. My, my, uh,
17:19
Speaker 2: My family was third generation of this, so.
17:21
Speaker 1: No kidding.
17:22
Speaker 2: Yeah. Yeah, so.
17:24
Speaker 1: Third generation, three generations of farmers.
17:27
Speaker 2: Three generation agriculture farming.
17:29
Speaker 1: And then we switch over to AI.
17:31
Speaker 2: Yeah, yeah. And we're doing like humanoid robots now and
17:36
Speaker 2: AI systems. But yeah, no, I got really interested in
17:39
Speaker 2: computers like really young. Started a bunch of like startups
17:43
Speaker 2: in like, you know, in high school and college.
17:46
Speaker 1: What kind of startups?
17:48
Speaker 2: You know, at first, just like mostly things on the web,
17:52
Speaker 2: like selling things, did a bunch of like different types
17:56
Speaker 2: of like products I was selling on the internet for
18:00
Speaker 2: like throughout like high school and college. Small like drop shipping,
18:04
Speaker 2: retail electronics, like all kinds of things. Legion marketing and
18:09
Speaker 2: just fun stuff. It was like nothing serious, you know,
18:11
Speaker 2: just like playing around the internet, trying to make some money.
18:14
Speaker 2: I didn't grow up with money. So it was like
18:16
Speaker 2: internet was a way to like, you know, maybe make
18:19
Speaker 2: some money. It was really fun. I loved the ability
18:23
Speaker 2: to go out and create things and kind of control
18:26
Speaker 2: my destiny. So it was just something I attached to
18:28
Speaker 2: really early on.
18:29
Speaker 1: Right on, right on. Do you have brothers and sisters?
18:32
Speaker 2: I have a brother, yeah.
18:35
Speaker 1: Is he a farmer?
18:37
Speaker 2: Colby? No, he actually runs an AI defense company called Scout.
18:41
Speaker 2: No kidding? Yeah, basically building autonomy and AI models for
18:46
Speaker 2: defense in the military.
18:47
Speaker 1: So you guys both got into AI?
18:49
Speaker 2: We both got into AI. We live like a block
18:50
Speaker 2: away from each other today. Are you serious? Yeah, we
18:53
Speaker 2: grew up together really close. We went to the same college.
18:57
Speaker 2: We're like different ages, a couple years apart. And then
19:00
Speaker 2: we were in New York for about 15 years together.
19:02
Speaker 2: And then he just moved out to California. We live
19:04
Speaker 2: literally a block away. I see him almost every weekend.
19:07
Speaker 2: He has a startup like basically 10 minutes away from mine,
19:11
Speaker 2: you know, where I'm at now. And he's doing great.
19:15
Speaker 1: Man, I mean, what do your parents think when you're
19:20
Speaker 1: coming home with, with what you guys are involved in
19:23
Speaker 1: and what you're, what you're creating? Um, because you're such
19:27
Speaker 1: a wild, you know what I mean? From farmer to,
19:30
Speaker 1: to this.
19:31
Speaker 2: Honestly, like, uh, I think one thing my parents both
19:34
Speaker 2: drilled into, I think both of us like really early
19:36
Speaker 2: was like, um, you know, farming is like very entrepreneurial.
19:39
Speaker 2: Like my dad is like, you know, ran his own business. Like,
19:42
Speaker 2: you know, uh, you kind of have to go out
19:44
Speaker 2: there and put the work in or you're not going
19:45
Speaker 2: to get paid. Um, So early on, he's like, listen,
19:48
Speaker 2: if you want to control your destiny and if you
19:50
Speaker 2: you know if you want to make money and you
19:51
Speaker 2: know like be able to actually uh you know like
19:56
Speaker 2: do what you really want in life you need to
19:57
Speaker 2: like run your own business and that was like beating
20:00
Speaker 2: to our heads like growing up like uh you know
20:02
Speaker 2: at some point you need to you know you need
20:03
Speaker 2: to probably get out of here get out of farming
20:05
Speaker 2: it's not doing well and uh you need to start
20:07
Speaker 2: something on your own and uh so it just of
20:10
Speaker 2: just like by default i was like okay this is
20:11
Speaker 2: what i'm gonna i'm gonna go do since i was
20:13
Speaker 2: a kid.
20:14
Speaker 1: Yeah um but you got some proud parents man.
20:17
Speaker 2: Yeah parents are great wow yeah they they're like what
20:20
Speaker 2: the hell's going on here what are you what are
20:22
Speaker 2: you doing um but i've been doing pretty crazy stuff
20:24
Speaker 2: for a while now so i think it's like it's
20:25
Speaker 2: gotten a point where it's like uh you know even
20:27
Speaker 2: at archer we're building like 6 000 pound electric aircraft
20:30
Speaker 2: and uh before that doing you know internet startup stuff
20:33
Speaker 2: but um it's kind of been you know Working on
20:36
Speaker 2: crazier stuff now for a little over a decade.
20:38
Speaker 1: Were you building stuff as a kid too?
20:40
Speaker 2: Yeah. Constantly building stuff.
20:42
Speaker 1: What kind of stuff?
20:43
Speaker 2: Stuff on the farm, building stuff in software and internet.
20:47
Speaker 2: I just love building stuff all day. I'm very big
20:50
Speaker 2: into science and mathematics. I'm more of a visual learner too.
20:54
Speaker 2: I like building stuff and seeing it and touching things.
20:58
Speaker 2: Even honestly doing internet for like... I did work in
21:01
Speaker 2: the internet and software for 10 years. I just like
21:04
Speaker 2: always sat there every day, like wishing I was working
21:06
Speaker 2: on hardware. Stuff I could like touch with my hands.
21:10
Speaker 2: Stuff like when growing up was like, you know, I
21:12
Speaker 2: was like rebuilding computers or just like on the farm
21:14
Speaker 2: and building stuff. I always like envied things that you
21:17
Speaker 2: can go touch and build.
21:19
Speaker 1: Wow.
21:20
Speaker 2: Basically like atoms.
21:22
Speaker 1: Man. So where do you go? Where'd you go to school?
21:24
Speaker 2: I went to University of Florida.
21:26
Speaker 1: University of Florida.
21:27
Speaker 2: Yep.
21:28
Speaker 1: Where do you go from there?
21:29
Speaker 2: Um, So after school, I moved to New York and
21:34
Speaker 2: I started working on software startups. And during college, I
21:39
Speaker 2: was working on basically a bunch of like side, small,
21:42
Speaker 2: like internet things. And then kind of like shortly after college,
21:47
Speaker 2: I started a company called Vetteri. And the goal was
21:50
Speaker 2: to basically build like a, I got really kind of
21:54
Speaker 2: going through college is like, you got to look for
21:55
Speaker 2: a job. You had to go find something full time
21:58
Speaker 2: and got caught up in like the whole interviewing process
22:00
Speaker 2: of like looking for jobs. I just thought it was
22:02
Speaker 2: so broken, like applying for jobs and like never hearing back.
22:05
Speaker 2: And like, you have to go through headhunters. And then
22:07
Speaker 2: it basically became like somewhat of this like, you know,
22:11
Speaker 2: boys club of like trying to figure out where do
22:12
Speaker 2: you went to school? And then like certain people knew
22:14
Speaker 2: other folks of like how to get in. And it
22:16
Speaker 2: was just like, it wasn't very much a meritocracy. And
22:19
Speaker 2: I just thought the whole process was extremely broken. And
22:23
Speaker 2: so I started Vetteri. We were basically an AI recruiting marketplace.
22:26
Speaker 2: So the goal was like, if we can get all
22:27
Speaker 2: the world's talent and hiring on one platform, understand their
22:31
Speaker 2: needs really well, can we make matches at scale, like
22:34
Speaker 2: without like any humans involved? And like the head of
22:37
Speaker 2: the industry is like hundreds of billions of dollars a year.
22:40
Speaker 2: I won't even....
22:42
Speaker 1: I won't use it.
22:43
Speaker 2: No, like, I know.
22:44
Speaker 1: Because I just keep hearing everybody gets ripped off.
22:46
Speaker 2: Ripped off. It's so expensive. Like, pay, like, $ 50, 000 a hire. It's, like, insane.
22:50
Speaker 1: And then they'll coax the guy out that they just
22:52
Speaker 1: brought to you and have him go to another job.
22:55
Speaker 2: Yeah, they'll force you into this role so they get
22:56
Speaker 2: paid a commission.
22:57
Speaker 1: So Vetteri is a connector.
23:00
Speaker 2: Yeah, a connector. Well, funny enough, we ended up selling to, like,
23:03
Speaker 2: the world's largest recruiting company that does staffing. But, like,
23:05
Speaker 2: let's leave that for a minute. But we basically started
23:08
Speaker 2: in 2012.
23:08
Speaker 1: And....
23:11
Speaker 2: And the goal was like, how do we put like
23:13
Speaker 2: a lot of job seekers and a lot of employers
23:15
Speaker 2: on a platform, understand their preferences and match them at scale?
23:18
Speaker 2: Like just like how to use algorithms. At the time
23:20
Speaker 2: we were like, let's use AI, but it was basically
23:22
Speaker 2: like how do we use a lot of algorithms to
23:24
Speaker 2: figure out like what people want and then make matches.
23:26
Speaker 2: So you can just push up a button, connect the
23:28
Speaker 2: right folks and then make placements. And then we ended
23:31
Speaker 2: up charging most of our revenue came from subscriptions from
23:34
Speaker 2: big companies, like big banks or startups or tech companies
23:38
Speaker 2: basically looking for talent. We started just in tech. in
23:41
Speaker 2: the U in the U S. So, um, at one
23:44
Speaker 2: point we had about, I think about a little under
23:45
Speaker 2: 20 or so cities globally that we were operating in.
23:48
Speaker 1: Wow.
23:49
Speaker 2: But most of it was tech, uh, talent, tech spaces,
23:52
Speaker 2: you know, at that point.
23:53
Speaker 1: How long ago was this?
23:55
Speaker 2: Started in 2012 and then ended up selling the business
23:58
Speaker 2: in 2017 or 2018. So about five years, six years.
24:01
Speaker 1: Right on.
24:02
Speaker 3: Yeah.
24:02
Speaker 1: Then where do we go? Okay.
24:04
Speaker 2: So, um, that veteran was like a really tough, I
24:07
Speaker 2: like basically went like I didn't have much money, went
24:10
Speaker 2: fully all in with the business. I went into debt
24:13
Speaker 2: at one point in 2015. The business was having a
24:16
Speaker 2: tough time. And then we ended up selling, ended up
24:20
Speaker 2: doing really well. The business completely hockey-sticked in growth when
24:24
Speaker 2: we got all the things figured out. And just like.
24:26
Speaker 1: Veteri?
24:27
Speaker 2: Veteri was. And then ended up getting approached by the
24:30
Speaker 2: world's largest recruiting company. The same groups you and I
24:32
Speaker 2: were talking about. The same groups we were trying to
24:35
Speaker 2: take out of business. And they were like, oh, we
24:38
Speaker 2: want to acquire the company. And at the time we
24:42
Speaker 2: were like, I was like completely dead broke and put
24:44
Speaker 2: everything out of the business. It was like, I think
24:45
Speaker 2: it was almost seven years in. And, you know, we
24:51
Speaker 2: were excited about an acquisition a year before that, $ 10
24:53
Speaker 2: million from one of the big tech companies. And they
24:57
Speaker 2: came in at 110 million. And, yeah. And it was
25:01
Speaker 2: a good time for me. I felt like the business
25:03
Speaker 2: was doing well. I learned a lot and I was
25:04
Speaker 2: kind of ready for my next chapter. So I ended
25:06
Speaker 2: up selling that business to the Deco Group. It's like
25:09
Speaker 2: the world's largest recruiting company.
25:13
Speaker 1: But you didn't even have it for sale. They just
25:15
Speaker 1: approached you.
25:15
Speaker 2: Yeah, we didn't hire a bank or anything. Listen, at
25:18
Speaker 2: the time we were doing like, I don't know, 20, 30,000
25:22
Speaker 2: interview requests like a week. Like, so that was like
25:26
Speaker 2: no humans involved. Like think about how many humans it
25:28
Speaker 2: would take to do like 20 or 30,000 interview requests.
25:31
Speaker 2: It's like, you know what I mean? And then manage
25:32
Speaker 2: all that processes. So we were like, the growth was
25:35
Speaker 2: just unbelievable. And there was like, there's something better to
25:39
Speaker 2: like a human jamming you in roles, right? Like, it's
25:41
Speaker 2: just like, you need like a, and then to extent
25:43
Speaker 2: you can get, you know, all the world's like talent
25:47
Speaker 2: there and all the world's companies looking, you can really
25:49
Speaker 2: create an amazing environment where you can get people to
25:51
Speaker 2: like the right jobs. And right now it's not like that.
25:54
Speaker 2: It's like a really black box, both finding talent and
25:56
Speaker 2: looking for a job. It's just a terrible experience. So
25:59
Speaker 2: that clicked. The world's largest recruiting company came in and said,
26:03
Speaker 2: we got to buy this thing. Yeah, I'll bet they did. Yeah,
26:07
Speaker 2: I sold the business. And it was great. It was
26:10
Speaker 2: a good time for me. It was a point in
26:12
Speaker 2: my life where I really wanted to do something much bigger.
26:16
Speaker 2: And so I basically took about a year And, um,
26:20
Speaker 2: so it took about a year of a time. I
26:21
Speaker 2: got the term sheet to sell, uh, to when we
26:23
Speaker 2: actually sold and closed. It's a long process. You have
26:26
Speaker 2: to go through like tons of docs and then you
26:28
Speaker 2: announce the, announce the deal. Then you actually close the deal.
26:30
Speaker 2: And then it went into escrow. Then it finally hit
26:32
Speaker 2: my account. It's kind of like one of those processes. And, um,
26:36
Speaker 2: I want to go work on something really important, hard, and, uh,
26:39
Speaker 2: a couple of industries that I've, I've been interested in
26:41
Speaker 2: robotics and aviation and some areas of security for like
26:46
Speaker 2: basically since college. And, um, I basically spent a lot
26:50
Speaker 2: of time trying to figure out if I was either
26:51
Speaker 2: going to work on, at the time, school shootings, like,
26:54
Speaker 2: basically 10x' d. And I was like, man, there's got to
26:56
Speaker 2: be something to do here. And we can, you know,
26:58
Speaker 2: and then secondly, I really wanted to work on, like,
27:01
Speaker 2: flying cars. Like, having watched a lot of sci-fi as
27:04
Speaker 2: a kid is like, man, like, I really want to go.
27:07
Speaker 2: There's a near-term problem of, like, we got to go
27:10
Speaker 2: help with security in schools, K through 12, mostly in
27:13
Speaker 2: the US. And then how do we, I want to
27:15
Speaker 2: work on flying cars. And I ended up making a
27:17
Speaker 2: decision to work on flying cars at the time. So
27:21
Speaker 2: in 2018, shortly after the sale of Vetteri, I started
27:24
Speaker 2: Archer Aviation. And basically, the story here is you can
27:31
Speaker 2: build an electric aircraft that can take off like a helicopter.
27:35
Speaker 2: If you take off like a helicopter, you don't need
27:37
Speaker 2: to place the airports outside of cities. You can place
27:39
Speaker 2: them inside of cities. Think about like a helicopter can
27:42
Speaker 2: take off from a building or a helipad. If you
27:48
Speaker 2: can take off vertically, you can basically nestle the aircraft
27:50
Speaker 2: inside of cities. Half the world lives in cities today.
27:55
Speaker 2: By the middle of the century, it'd be like 70%
27:58
Speaker 2: of the world, and you just can't get around. It's
28:01
Speaker 2: just gridlocked everywhere in major cities. It just sucks to
28:04
Speaker 2: go 20, 30 miles. It takes an hour in most cases. Basically,
28:08
Speaker 2: you can design an aircraft that can take off vertically
28:10
Speaker 2: and then fly like an airplane, so you can get
28:12
Speaker 2: a lot of distance. And you can basically then re-architect
28:15
Speaker 2: the whole aircraft to be fully electric. The reason you
28:17
Speaker 2: want to do that is for cost and safety. You
28:20
Speaker 2: basically can make it like a lot less expensive. You
28:23
Speaker 2: can put a lot less parts in the aircraft that
28:25
Speaker 2: are also good for safety. So basically you can build
28:27
Speaker 2: like an electric flying car that you can move around.
28:30
Speaker 2: So instead of like calling an Uber or driving that
28:32
Speaker 2: might take you an hour in LA or SF or
28:34
Speaker 2: New York, you basically can fly there in 10 minutes.
28:38
Speaker 2: And if we can pool everybody together like in a
28:40
Speaker 2: kind of like an Uber pool style... you can do
28:43
Speaker 2: it for as cheap as an Uber. But the problem
28:47
Speaker 2: was like, I didn't know anything about how to build
28:49
Speaker 2: electric aircraft.
28:50
Speaker 1: I mean, where do you, you know, I'm just, I
28:53
Speaker 1: know you just sold your business for $ 110 million, but
28:56
Speaker 1: where do you get the confidence to, where do you
29:01
Speaker 1: get the confidence to go, I'm going to build vertical
29:04
Speaker 1: takeoff and landing flying cars now?
29:07
Speaker 2: I mean, listen, I didn't wake up to this world
29:08
Speaker 2: like learning how to build software. So like I learned
29:10
Speaker 2: how to, how to do that and run engineering and run,
29:13
Speaker 2: run the company. And, uh, there was like a lot
29:16
Speaker 2: of through trial and error. Um, and, uh, I just
29:21
Speaker 2: felt like, like, well, I just felt I could learn it. Uh,
29:23
Speaker 2: I started, uh, uh, uh, in industrial system engineering at
29:27
Speaker 2: university of Florida. So, um, and then, you know, ran
29:31
Speaker 2: engineering and ran the company at battery. Um, so I
29:34
Speaker 2: basically just hit the books. I tried to learn as
29:36
Speaker 2: much as possible about three subject areas. First was like electrification,
29:39
Speaker 2: which like, you know, the time, electric vehicles were really
29:41
Speaker 2: doing well, and even drones. Vertical takeoff and landing, vertical lift,
29:47
Speaker 2: which is traditional rotorcraft or helicopter. And the third is
29:50
Speaker 2: winged aircraft, like airplanes. You really need wings. So you
29:54
Speaker 2: basically have to learn about those three subjects. So I
29:57
Speaker 2: basically bought my basement downstairs at home, where there's every
30:00
Speaker 2: possible book on these subjects you could imagine, and started
30:03
Speaker 2: reading as much as I possibly could. And this was
30:05
Speaker 2: during the year transition. As I was transitioning out of
30:07
Speaker 2: veterinary into archery, I was reading every possible thing. And
30:10
Speaker 2: then I found a small community of folks that were
30:12
Speaker 2: hosting on-site either half-week or week-long courses for this. And
30:19
Speaker 2: so I would go to these, sometimes they were sponsored
30:20
Speaker 2: by NASA or by colleges or whatever it would be,
30:22
Speaker 2: on basically rotorcraft or electric propulsion. or winged aircraft aerodynamics.
30:28
Speaker 2: And I would basically like try to like, like learn
30:31
Speaker 2: as much as possible. It got to the point where
30:33
Speaker 2: I was like completely obsessed with this algorithm I was
30:35
Speaker 2: building on electric aircraft sizing. Like how, how do you actually,
30:39
Speaker 2: like how would you actually build an electric aircraft? So
30:42
Speaker 2: electric aircraft, which interesting is like in rotorcraft, like, you
30:47
Speaker 2: basically want to create the most efficient lifting device possible.
30:50
Speaker 2: You need as much of like as the rotor disc area,
30:54
Speaker 2: like in terms of surface area as you possibly can.
30:56
Speaker 2: That's why the helicopter rotors are so large. We want
30:58
Speaker 2: that to be really large. That'll reduce power and get
31:01
Speaker 2: you off the ground. In electric aircraft, the problem you're
31:04
Speaker 2: starting with is you have like 1 30th of the
31:06
Speaker 2: energy as you do in kerosene in a battery pack.
31:10
Speaker 2: So you're just like, you're off the bat of 1
31:12
Speaker 2: 30th less range or 1 30th less energy. And so...
31:17
Speaker 2: power becomes the dominating factor of how to basically build
31:21
Speaker 2: electric aircraft. How do you get power down as much
31:23
Speaker 2: as possible? You really want a lot of disk area.
31:26
Speaker 2: A lot of disk area is, one, it could be
31:30
Speaker 2: good for power, but it's also bad because you have
31:32
Speaker 2: no redundancy in the system. You have one rotor blade.
31:35
Speaker 2: If it doesn't go well, you go down. With electrification,
31:39
Speaker 2: you can basically build much smaller rotors. and be able
31:44
Speaker 2: to fully electric. The reason you can't do that with
31:46
Speaker 2: traditional turbofans or engines is it gets too inefficient at
31:51
Speaker 2: these sizes. You can't build 12 propellers on a helicopter.
31:57
Speaker 2: The efficiency just drops to nothing. With electrification, you can.
32:01
Speaker 2: You can size the electric motors to small sizes and
32:03
Speaker 2: they're still 90% efficient. Small electric motor on the table
32:06
Speaker 2: or a big one the size of your chair, Same efficiency.
32:10
Speaker 2: When you do that, you create a lot of redundancy
32:12
Speaker 2: across the system. So you can build an aircraft with
32:13
Speaker 2: 12 electric motors.
32:15
Speaker 1: So the rotors are underneath.
32:19
Speaker 2: Are they underneath? The problem here is you can design
32:22
Speaker 2: it however you want. You could put a bunch of
32:24
Speaker 2: rotors along the wings. You can put them laterally across
32:26
Speaker 2: the fuselage. You can make one big one. You can
32:29
Speaker 2: make 30 small ones. So how do you design it?
32:32
Speaker 2: That's the problem I hit in 2018 was how do
32:35
Speaker 2: you actually do this? And so it basically was like
32:38
Speaker 2: a crazy man trying to design this algorithm to like,
32:41
Speaker 2: what is the ideal aircraft design? And then how do
32:45
Speaker 2: I go build it? So it was actually at a,
32:47
Speaker 2: I was at a Hyatt Regency hotel in Atlanta in 2018.
32:51
Speaker 2: It was an electric propulsion week-long design course and like
32:57
Speaker 2: aerodynamics course for winged aircraft. And I met a guy
33:01
Speaker 2: there that, that was basically in the engineering department at
33:05
Speaker 2: University of Florida. He was doing his PhD in aerospace
33:07
Speaker 2: and asked him what he was doing there. He said,
33:10
Speaker 2: going from University of Florida, I was like, oh, I
33:11
Speaker 2: went to school there as well. He's like, I'm like,
33:15
Speaker 2: what are you doing here? He said, oh, I want
33:16
Speaker 2: to go do a career in eVTOL aircraft. This is
33:21
Speaker 2: called electric vertical takeoff and landing. A helicopter is a VTOL.
33:26
Speaker 2: You put a little E in front, it's over electric.
33:28
Speaker 2: He asked me what I'm doing here. I was like,
33:29
Speaker 2: I'm starting a company to do this and I need
33:31
Speaker 2: to figure out how to go build these things. And
33:33
Speaker 2: he's like, well, listen, my professor runs a small drone lab.
33:36
Speaker 2: He's got a full building. He's got 12 PhDs. Why
33:39
Speaker 2: don't you come down and meet him and see if
33:41
Speaker 2: you can start building aircraft with him? So I flew
33:43
Speaker 2: down that weekend to go meet his professor that runs
33:47
Speaker 2: all of basically mechanical engineering and aerospace. And long story
33:51
Speaker 2: short is I ended up taking over his lab. And
33:56
Speaker 2: me and him and his team started building aircraft in
33:59
Speaker 2: 2018 and 2019 down at University of Florida. And I
34:00
Speaker 2: temporarily moved down there with my my daughter at the
34:03
Speaker 2: time and my wife living in Gainesville, Florida. Like, um, and, um,
34:09
Speaker 2: and it was great. Like we ended up building, I
34:11
Speaker 2: ended up funding a lab, uh, right off of Archer road,
34:14
Speaker 2: a new lab. Cause we needed more space. We ended
34:15
Speaker 2: up calling the business Archer aviation is the main road
34:17
Speaker 2: down of university of Florida. And I spent the next
34:20
Speaker 2: like year, year and a half basically like, um, modeling
34:25
Speaker 2: and building electric EV tall aircraft.
34:28
Speaker 1: Holy shit.
34:28
Speaker 2: Yeah. And it was, um, It was a great time
34:31
Speaker 2: in my life. And the problem is there was no
34:36
Speaker 2: intersection of folks that knew electric, knew rotorcraft, or knew airplanes.
34:40
Speaker 2: There was no Venn diagram of overlap.
34:42
Speaker 1: Gotcha.
34:42
Speaker 2: So there was nobody in the world that understood how
34:44
Speaker 2: did this all stuff that works. So I had to
34:46
Speaker 2: go from scratch, like learn it from first principles. And
34:51
Speaker 2: then ended up moving the company out to California, basically
34:55
Speaker 2: a few years into the business. And then, you know,
34:58
Speaker 2: Things took off from there. We built bigger aircraft. I
35:01
Speaker 2: took the company public within three years of starting it.
35:03
Speaker 2: We're like a $ 6 billion publicly traded company today. And yeah,
35:08
Speaker 2: designed basically four or five generations of aircraft at Archer.
35:12
Speaker 2: And it was hard. It really set me up well
35:17
Speaker 2: for doing figure and cover and the rest of the
35:19
Speaker 2: stuff we can talk about later. But it was hard.
35:24
Speaker 2: Even going public was probably one of the hardest experiences Really?
35:29
Speaker 1: Why is that?
35:31
Speaker 2: We went public through a SPAC process. So, you know,
35:34
Speaker 2: SPACs at the time, like four or five years ago,
35:36
Speaker 2: were like all the rage.
35:37
Speaker 1: Okay.
35:39
Speaker 2: And it was a special purpose acquisition company. So, it
35:41
Speaker 2: was basically companies that were like going public through a merger,
35:43
Speaker 2: like a reverse merger. And it was hard because in 2018, 2019,
35:49
Speaker 2: coming off of software, I had never done hardware before. So, A,
35:52
Speaker 2: it was like hard to raise capital. And B, there
35:56
Speaker 2: was nobody funding like, deep tech, electric vertical takeoff and landing, like, companies. Like,
36:03
Speaker 2: you know what I mean? The big venture capital groups
36:05
Speaker 2: were not funding SpaceX or Tesla or Rivian. Like, none
36:09
Speaker 2: of these were getting funded by traditional investors. They weren't
36:11
Speaker 2: raising money from the named investors we all know about now.
36:15
Speaker 1: Oh, shit. Are they always behind like that?
36:17
Speaker 2: The mandate for most of these VCs in the Bay
36:19
Speaker 2: Area or Silicon Valley and stuff are not to do hardware.
36:22
Speaker 1: Gotcha.
36:23
Speaker 2: They don't really, and if they do hardware, they don't
36:25
Speaker 2: do deep tech. They don't do, like, rockets and, autonomous
36:28
Speaker 2: vehicles and I don't think there's a single top VC
36:34
Speaker 2: in the US that's invested in a humanoid company.
36:36
Speaker 1: No shit.
36:38
Speaker 2: As of last six months ago now, nothing. They just
36:41
Speaker 2: don't do this stuff. I just made $ 110 million or
36:42
Speaker 2: just sold the company for $ 110 million. I made a
36:44
Speaker 2: lot of money personally. I ended up going all in
36:47
Speaker 2: on Archer. through the IPO, like through going public, I
36:59
Speaker 2: put like all the money I basically had. I bought
37:02
Speaker 2: a house and the rest of the money went all
37:03
Speaker 2: into it. And it was a stressful period. So we
37:07
Speaker 2: went public through a SPAC. And the reason it was
37:09
Speaker 2: tough is we ended up getting to a point where
37:11
Speaker 2: we just couldn't raise enough money privately. Like it was
37:14
Speaker 2: either like raise, you know, $ 100 million privately at like,
37:18
Speaker 2: you know, some valuation, three, four or $ 500 million, or
37:21
Speaker 2: it was like go public and raise like a billion dollars.
37:24
Speaker 1: Wow.
37:25
Speaker 2: And we ended up going public and raising a billion dollars.
37:27
Speaker 1: Wow. You've got a huge appetite for risk, huh?
37:30
Speaker 2: And we got sued during it.
37:32
Speaker 1: Oh, really?
37:33
Speaker 2: Yeah. Like we got sued by basically like Boeing and
37:36
Speaker 2: a big startup that was founded by Larry Page, Google founder.
37:40
Speaker 2: That's got to be intimidating. Yeah. I woke up to
37:44
Speaker 2: like a front page of New York Times article about- Oh, shit. Yeah.
37:48
Speaker 2: It was crazy. I mean, the backstory is I- So
37:52
Speaker 2: Larry Page started a company in the Bay Area about
37:56
Speaker 2: 10 years ago called Kitty Hawk. And they did a
37:59
Speaker 2: great work over 10 years in electric VTOL aircraft. And
38:04
Speaker 2: I ended up taking basically the core 10 to 15
38:10
Speaker 2: folks that were there all came over to Archer within
38:13
Speaker 2: the first two years.
38:14
Speaker 1: Wow.
38:15
Speaker 2: And they retaliated by just trying to harass us while
38:17
Speaker 2: we were going public. And so, yeah, it was just
38:20
Speaker 2: a crazy story getting public. Ended up getting public, you know,
38:24
Speaker 2: a billion dollars on the balance sheet and we just
38:26
Speaker 2: started building like aircraft and started building the service, like
38:30
Speaker 2: thinking about the app and how you're going to check
38:32
Speaker 2: in and how you're going to build places like real
38:35
Speaker 2: estate to fly into. And yeah, and then like the
38:39
Speaker 2: engineering work we had to do around there of designing,
38:42
Speaker 2: you know, it's basically a flying robot. You have like
38:45
Speaker 2: battery systems, electric motors, sensors, embedded software and control systems.
38:51
Speaker 2: And basically like the robot you saw this morning, like very,
38:55
Speaker 2: you know, that's like, it's a flying, it's a 6,000
38:57
Speaker 2: pound four passenger piloted robot. And it has 24 degrees
39:02
Speaker 2: of freedom on the system. Like wing flaps, we tilt
39:08
Speaker 2: the front, the leading edge six motors, 90 degrees for
39:12
Speaker 2: basically take off vertically and then go into forward flight.
39:15
Speaker 2: and then all the propellers, fan blades on have variable
39:18
Speaker 2: pitch propellers. So it's like a highly overactuated system that
39:22
Speaker 2: needs like a really good software. Like no human can
39:24
Speaker 2: like fly it basically without a really good control software.
39:29
Speaker 1: What altitude is it? flying?
39:30
Speaker 2: About a few thousand feet. So about 2,000 to 3,000
39:33
Speaker 2: feet above ground level.
39:34
Speaker 1: And that's what it would normally be?
39:35
Speaker 2: Yeah, like traditional helicopters fly at these levels.
39:37
Speaker 1: I mean, what... I think about... I think a lot
39:41
Speaker 1: about Tesla and all the EV vehicles that are coming out. And,
39:47
Speaker 1: you know, it's... the government just seems so far behind
39:51
Speaker 1: on AI, you know, and you just brought up gridlock
39:54
Speaker 1: and all the cities. I've always wondered why aren't, when
39:59
Speaker 1: are we going to go full EV? I know there's
40:02
Speaker 1: a lot of pushback about that, you know, for, for, for,
40:04
Speaker 1: from an overreach standpoint. But if you, if you just
40:07
Speaker 1: think about the traffic in the cities and if you
40:09
Speaker 1: have the AI, you know, processing all this, that even,
40:13
Speaker 1: even without air vehicles, I feel like a lot of
40:15
Speaker 1: that would go away because the, the, AI will route
40:19
Speaker 1: you the quickest and take all the traffic patterns into
40:22
Speaker 1: account and it would just flow a lot easier. But
40:28
Speaker 1: there's all this government regulation.
40:29
Speaker 2: I think it's also hard because if you look at
40:32
Speaker 2: the number of installed cars in the world, like billion
40:34
Speaker 2: and a half or so installed cars, we make like
40:36
Speaker 2: 80 million or so cars a year in the world.
40:39
Speaker 2: It takes you on the order of 20 years to
40:43
Speaker 2: replace all the cars. So if all the cars were
40:44
Speaker 2: electric and autonomous today, Autonomous cars have, like, autonomous hardware
40:48
Speaker 2: in them. It's not like you can just go out
40:49
Speaker 2: and retrofit all the cars in the world right now. Like,
40:52
Speaker 2: it's a hard problem.
40:53
Speaker 1: Well, I mean, if you look at Tesla, for example,
40:56
Speaker 1: I mean, it can self-drive, right? It can come get you.
41:01
Speaker 1: But when you're driving, if you take your eyes off
41:03
Speaker 1: the road, it wakes you up. You have to come back.
41:05
Speaker 1: I mean, it seems it's inviting more error into the
41:09
Speaker 1: road by doing that, in my opinion.
41:12
Speaker 2: Yeah.
41:13
Speaker 1: It's almost like someone.
41:14
Speaker 2: Could be more dangerous. Am I wrong? We're just in
41:16
Speaker 2: this transitory state right now where in like five years,
41:20
Speaker 2: like everything will be like fully autonomous and trusted and
41:23
Speaker 2: fine and you won't have to do that. And we're
41:25
Speaker 2: just in this transitory. We're in this chapter in the
41:27
Speaker 2: book for the technology roadmap here where like we're living
41:30
Speaker 2: through it and it's like a little messy and it's
41:32
Speaker 2: not quite like straightforward. And we don't quite know where
41:35
Speaker 2: it's headed next, but where it's headed is at some
41:37
Speaker 2: point in like five, whatever years where, you know, When
41:41
Speaker 2: our kids grow up, they're never gonna have to think
41:42
Speaker 2: about this. It's just gonna be autonomous from the start.
41:44
Speaker 2: It's gonna be by default native. And it'll be trusted
41:48
Speaker 2: and easy and safe. I think we're just living through
41:50
Speaker 2: this period right now, which is a weird thing. But
41:52
Speaker 2: if you close your eyes long enough, you'll have this
41:55
Speaker 2: autonomy and electrification everywhere.
41:58
Speaker 1: How long do you think it'll be?
42:00
Speaker 2: I mean, so I live in the Bay Area. You
42:02
Speaker 2: can take Waymos now. I can take Waymo's everywhere. It's unbelievable.
42:05
Speaker 2: They're already all over over there? They're everywhere.
42:07
Speaker 3: Yeah.
42:07
Speaker 2: They're in my, I'm in South Bay. Um, but they
42:09
Speaker 2: were in the city for a while and now they're in,
42:11
Speaker 2: you know, um, the other Palo Alto, Menlo Park, like
42:14
Speaker 2: San Jose, like all this, all over the place. They're
42:16
Speaker 2: really great. Uh, I take it. It's like, uh, my
42:19
Speaker 2: wife and I would go to dinner and stuff on
42:20
Speaker 2: the weekends. We take Waymo. It's like, it's so fun.
42:22
Speaker 2: It's like, it sounds so like, like it sounds so basic.
42:26
Speaker 2: Like you already take a Waymo. It's fine.
42:27
Speaker 1: It's just, it's, it's awesome, man.
42:29
Speaker 2: Like it's, it's great. You have like, um, The car
42:33
Speaker 2: drives so human-like, and it's such a great experience not
42:35
Speaker 2: having a human there, to be frank. I love it.
42:39
Speaker 2: I order so many Ubers and stuff and come and
42:42
Speaker 2: the car smells or it's dirty or whatever else, and
42:45
Speaker 2: it's just easy. It's really cool. Technology is in the
42:51
Speaker 2: early chapters, but it's all here. We're going to have
42:54
Speaker 2: autonomy at scale everywhere. It's just going to take some
42:57
Speaker 2: time to roll that out. It's the time it takes to.
43:00
Speaker 2: Get the technology mature enough where they can run enough cities,
43:02
Speaker 2: enough places, and then it's the time it takes to
43:04
Speaker 2: get the install base of autonomous hardware and software in
43:07
Speaker 2: all these places. That's going to take some time too.
43:10
Speaker 2: We just can't snap our fingers. We just don't have
43:11
Speaker 2: enough install base of autonomous vehicles in the world. I
43:14
Speaker 2: think like Tesla's got like 10 million cars in the
43:17
Speaker 2: road and like maybe there's thousands or so of like Waymos.
43:21
Speaker 2: You have like over a billion cars on the planet.
43:24
Speaker 2: So you need to like, you know, make a large
43:28
Speaker 2: fraction of that all autonomous. So you're looking at like a,
43:31
Speaker 2: this isn't going to happen in a year or two.
43:34
Speaker 2: It's going to take some time.
43:36
Speaker 1: When are we going to see your vehicles?
43:38
Speaker 2: The aircraft? We have them now. We fly every week
43:42
Speaker 2: in California. The challenging part with Archer is that we
43:47
Speaker 2: are governed by the federal airspace. So to fly passengers
43:50
Speaker 2: and charge money, we have to have like basically a
43:53
Speaker 2: type certification from the FAA. That process moves the speed
43:58
Speaker 2: of like the post office. And the FAA is not
44:00
Speaker 2: incentivized to put anything in the air unless they know
44:02
Speaker 2: for sure it's going to be really safe. The safety
44:05
Speaker 2: standard for us that we want to certify to is
44:08
Speaker 2: one times 10 to the minus nine in terms of
44:10
Speaker 2: hours of reliability before a catastrophic event. So that's one
44:15
Speaker 2: in a billion hours we can have a failure.
44:17
Speaker 1: One in a billion hours.
44:17
Speaker 2: Yeah. You can't be able to... That is like... That's
44:22
Speaker 2: the standard when we fly. It's like one of the
44:24
Speaker 2: safest form of transportation we take. And it's because of
44:26
Speaker 2: those standards governed by the FAA. Which is great. I'm like,
44:30
Speaker 2: you know, we're like, that's the bar you need to
44:33
Speaker 2: be at. And that's the bar you need to hit,
44:34
Speaker 2: especially taking passengers over cities with aircraft overhead. You need
44:38
Speaker 2: to be at those levels. So that's like the, that's
44:42
Speaker 2: the long pole in the tent for us. And that's
44:44
Speaker 2: wherever you go. If you go to, you know, Europe,
44:47
Speaker 2: it's YASA or, you know, CAA in China, wherever you're
44:51
Speaker 2: going to go, there's like, there's federal mandates to get
44:53
Speaker 2: basically an aircraft to take passengers. So we're in the
44:56
Speaker 2: middle of the FAA certification now. We hope to be
44:59
Speaker 2: certified in the coming, you know, as soon as possible, basically.
45:04
Speaker 2: But it's like, it's not something you can like, there's
45:05
Speaker 2: not like a date on the calendar of like, you'll
45:07
Speaker 2: be certified here. You have to work through a very like,
45:10
Speaker 2: very long and slow process with the FAA to get
45:12
Speaker 2: through this. And then we're also dual tracking that against
45:16
Speaker 2: a couple of different entities globally now to make sure
45:19
Speaker 2: we can get certified and get in there. But it'll happen, man.
45:23
Speaker 2: The aircraft, it just, again, we're in like this chapter of,
45:27
Speaker 2: We're like flying cars, electric aircraft, or just like it's early.
45:30
Speaker 2: It's earlier than like AVs or EVs, autonomous vehicles, electric vehicles.
45:36
Speaker 2: But it'll happen. It'll happen in our lifetime. We'll be
45:38
Speaker 2: taking these things around.
45:39
Speaker 1: I mean, what do you envision? Let's fast forward 20,
45:43
Speaker 1: 30 years. What do you envision? What does it look like?
45:46
Speaker 1: Do we have roads? Do those get ripped up? What
45:50
Speaker 1: does the sky look like? What does everyday life look like?
45:53
Speaker 2: Yeah. The really important thing to hear about the airspace
45:57
Speaker 2: is that it's three-dimensional and the roads are not. They're
46:00
Speaker 2: 2D and we've built cities now and houses and restaurants
46:04
Speaker 2: all around these places. You can't like, there's nowhere to go.
46:07
Speaker 2: There's like no more roads to build in these cities.
46:10
Speaker 2: So you have left with no choice if you, and
46:12
Speaker 2: then humanity are moving to cities. We have this like
46:16
Speaker 2: secular trend where we all want to live in cities
46:17
Speaker 2: right now. It's like half the world lives in cities.
46:20
Speaker 2: It'll be like 70% by 2050. So we're like all
46:24
Speaker 2: moving to cities. The roads can't grow anymore. And we're
46:27
Speaker 2: constantly moving around, going to work or going to restaurants.
46:30
Speaker 2: And it's basically getting worse. The arteries are hardening here
46:38
Speaker 2: around this. It's getting worse and worse. And it's some
46:41
Speaker 2: of the worst time. to spin on a road in traffic.
46:44
Speaker 2: It's so soul-sucking. It's just the worst time to lose.
46:50
Speaker 2: So the good news about the air is it's three-dimensional.
46:52
Speaker 2: You can stack basically an infinite amount of, say, roads
46:56
Speaker 2: in the air.
46:56
Speaker 1: Different altitudes.
46:57
Speaker 2: Yes, at different altitudes, and even laterally. So you can
47:03
Speaker 2: basically build little tunnels in the sky. and you can
47:06
Speaker 2: basically stack them, and you basically can put orders of
47:10
Speaker 2: magnitude more things in the air than you can in
47:13
Speaker 2: the road. It's the same for below ground with tunnels.
47:16
Speaker 2: So the future of travel in cities is below ground
47:18
Speaker 2: in tunnels and above ground in the sky. The boring company. Yeah, exactly.
47:23
Speaker 2: Just dig tunnels, and it's great. The only problem with tunnels,
47:29
Speaker 2: With the node system on the ground with, say, we
47:38
Speaker 2: call them vertiports, but basically real estate for flying cars,
47:43
Speaker 2: let's say you had 10 different vertiports inside of a city,
47:50
Speaker 2: places to take off and land from, you can travel
47:52
Speaker 2: between any one of those routes. So it opens up like,
47:56
Speaker 2: you know, basically exponentially more places to go to. I
47:58
Speaker 2: can go to like any node on the system at
48:00
Speaker 2: any time.
48:00
Speaker 1: Hold on. So you're saying in order to take off
48:04
Speaker 1: and land, you'll have to go to specific locations. You
48:06
Speaker 1: won't be able to do it from your home?
48:08
Speaker 2: Yeah, you're not going to take off and land from
48:10
Speaker 2: your home.
48:10
Speaker 1: Okay.
48:11
Speaker 2: Just because like acoustics in the neighborhood, it's gonna be
48:14
Speaker 2: too loud. You need like a decent amount of infrastructure
48:16
Speaker 2: for that, for charging and for passengers and cleaning and
48:19
Speaker 2: like checking in and stuff like that. They'll be like
48:22
Speaker 2: in your neighborhood. And you'll Waymo there or walk or
48:26
Speaker 2: take a bike. And then you'll get on these and
48:29
Speaker 2: they will go to any node on the network. You
48:31
Speaker 2: can't do that with tunnels. Tunnels have to go to
48:32
Speaker 2: A to B. You can't jump to another tunnel downstream.
48:37
Speaker 2: I want to jump to another tunnel 100 meters down.
48:39
Speaker 2: It doesn't happen in tunnels. You can do that with
48:41
Speaker 2: the sky. You can basically jump to any node on
48:43
Speaker 2: the network. It's exponentially more routes you can basically do
48:45
Speaker 2: with less real estate. And then you can basically stack
48:49
Speaker 2: orders of magnitude more traffic and humans in the sky.
48:52
Speaker 2: So my ambition is that you're going to be, for
48:55
Speaker 2: most trips that you're traveling over 20 minutes, all of
48:58
Speaker 2: that will move to the sky. And not only that,
49:01
Speaker 2: but you will have cities being transitioned to a point
49:06
Speaker 2: where you can live well outside of cities and get
49:09
Speaker 2: to cities really fast. The reason we live in cities
49:11
Speaker 2: is because we're working there and we have friends there. Yeah,
49:15
Speaker 2: it's like, I want to go to dinner with somebody.
49:16
Speaker 2: I want to see my buddies over here. And we
49:19
Speaker 2: want to go. to work over here or like go
49:21
Speaker 2: to the mall over here. It's like, everything's there. And
49:24
Speaker 2: that's what we want to be. We're social creatures. We
49:26
Speaker 2: want to be there next to other humans, or some
49:28
Speaker 2: of us are. And so, yeah, so anyways, but like,
49:33
Speaker 2: you know, now that you can fly this 150 miles
49:35
Speaker 2: an hour in the air with no traffic, point to point,
49:37
Speaker 2: like no stop signs, no construction, no things jumping out
49:40
Speaker 2: in front of you. You don't have to travel different distances.
49:43
Speaker 2: You're going straight from A to B in most cases.
49:46
Speaker 2: So you're removing 10 or 20% of basically the distance
49:49
Speaker 2: just by going point to point. And then you have
49:51
Speaker 2: nothing stopping you going 150 miles an hour most of
49:53
Speaker 2: the way there. You can live far outside of cities
49:56
Speaker 2: and get down to city center in under 30 minutes.
49:59
Speaker 1: So will these be personally owned or will this be
50:02
Speaker 1: like an Uber service?
50:03
Speaker 2: It'll be like an Uber service. To get costs down,
50:06
Speaker 2: you'll basically just pay per trip. You'll pull up an
50:09
Speaker 2: app and you'll go like, I want to go downtown
50:10
Speaker 2: to whatever. It's 40 bucks and I'll be there in
50:13
Speaker 2: under 30 minutes. And you'll say, great, I want to
50:16
Speaker 2: be there that time. You'll hit a button. It'll be
50:17
Speaker 2: on demand. You'll ride your bike over a walk. You'll
50:20
Speaker 2: get in one. It'll leave in seven minutes. And then
50:23
Speaker 2: you're basically flying right down to town.
50:25
Speaker 1: Holy shit. And you're saying this will be in every neighborhood.
50:29
Speaker 1: This will be very accessible to everybody. Yeah, yeah.
50:33
Speaker 2: You're designing a whole... Electrification allows you to reduce the
50:36
Speaker 2: cost and the safety burden of all this.
50:39
Speaker 1: Wow.
50:40
Speaker 2: We have like a normal helicopter could have like 100,
50:47
Speaker 2: 200 safety-critical components that any component gives out, the helicopter
50:50
Speaker 2: can go down. An electric aircraft has none. You can
50:54
Speaker 2: lose a motor, you can lose a battery pack on
50:57
Speaker 2: board and still fly safe without having this.
51:00
Speaker 3: And.
51:01
Speaker 2: And so just from a safety, from a part count,
51:04
Speaker 2: from a cost, from an acoustic signature, it's not gonna,
51:07
Speaker 2: like helicopters are loud and very noisy and it's just
51:12
Speaker 2: a much better technology for this.
51:14
Speaker 1: Have you been in one.
51:16
Speaker 2: Yet? I haven't flown inside of ours yet. I've also
51:20
Speaker 2: been inside of our aircraft. And we basically have professional
51:26
Speaker 2: pilots at the company.
51:28
Speaker 1: Test pilots?
51:29
Speaker 2: Yeah, test pilots. And they're career test pilots and they're unbelievable.
51:32
Speaker 1: Not bad.
51:33
Speaker 2: You know, a lot from the military, a lot from
51:35
Speaker 2: the big aerospace groups, and they're just professionals.
51:39
Speaker 1: What do they think?
51:40
Speaker 2: I love it, man. This is the future of aviation.
51:42
Speaker 3: Man.
51:43
Speaker 2: Everything's going electric and.
51:45
Speaker 1: It's so crazy.
51:46
Speaker 2: Yeah, it's crazy. It works. Like, it's crazy it works
51:49
Speaker 2: and crazy we're in the right time period to make
51:51
Speaker 2: this happen. Yeah. And, you know, the good thing about,
51:56
Speaker 2: you know, Archer now is we've, like, we've demonstrated... The
51:59
Speaker 2: hard part is making sure you're in the right decade.
52:03
Speaker 2: You don't want to go do this and you find
52:06
Speaker 2: out it's a 2040 event and you can't get it done.
52:08
Speaker 2: It's just a waste of time. The good news for
52:13
Speaker 2: us is we're in a sweet spot here where this
52:16
Speaker 2: is going to happen. Aircraft now work. We're certifying now
52:20
Speaker 2: with the government bodies like the FAA to make it happen.
52:22
Speaker 2: We have a good balance sheet with cash. Team's great.
52:26
Speaker 2: And so it's just like, you know, get certified and
52:29
Speaker 2: get this thing going.
52:30
Speaker 1: Damn, that is, you're really changing the world.
52:33
Speaker 2: Well, we're at the start of it, but, you know,
52:36
Speaker 2: hopeful to make this thing work.
52:38
Speaker 1: Where are we going next?
52:42
Speaker 2: Humanoids?
52:42
Speaker 1: Let's, yes. Let's do it.
52:45
Speaker 2: Yeah.
52:46
Speaker 1: How did this idea start?
52:49
Speaker 2: Yeah, so, you know, I spent like five or six
52:53
Speaker 2: years working on like a pretty crazy robotics project work
52:56
Speaker 2: at Archer. And, um... Like, the ultimate, like, meta problem
53:02
Speaker 2: in robotic space is, can you... can you build, like,
53:06
Speaker 2: a general purpose machine to do everything in the world like...
53:11
Speaker 2: much of what, say, humans can in the world? And
53:14
Speaker 2: I have this big belief that, you know, we have this, like,
53:18
Speaker 2: weird biological species. Like, we look, we're like, you know,
53:20
Speaker 2: we have these weird hands and arms and legs and
53:23
Speaker 2: certain height and, um, sensors. And then we ended up
53:27
Speaker 2: building this world around us so we can interact with it.
53:30
Speaker 2: I mean, if we get dropped into Mars today, we're
53:32
Speaker 2: going to build coffee cups that we can hold and
53:34
Speaker 2: stairs and doors, and we're going to build this stuff again.
53:36
Speaker 2: And it's like the human operating system. We're building things
53:40
Speaker 2: we can use and operate in that makes it easy
53:44
Speaker 2: for our lives. And we built it around the form
53:46
Speaker 2: factor that we are. Meaning if we look differently, the
53:49
Speaker 2: world will look different. Our espresso machine would be different looking.
53:52
Speaker 2: We might not even like espresso or caffeine in this case.
53:56
Speaker 2: So we built this whole world around us. The holy
53:59
Speaker 2: grail for robotics is can you basically build a general
54:03
Speaker 2: purpose machine that can do what humans can, which for
54:06
Speaker 2: me is like a humanoid robot. And a humanoid robot
54:09
Speaker 2: is just a robot that has like a human form.
54:12
Speaker 2: So it has legs so it can walk upstairs and
54:15
Speaker 2: walk over like, you know, uneven terrain or say things
54:18
Speaker 2: on the ground and bend down, which legs are important for,
54:23
Speaker 2: or reach up. uh, has like arms and hands so
54:26
Speaker 2: we can manipulate objects and do things like, you know,
54:28
Speaker 2: grab his stuff, open, open these gummies and, um, you know,
54:33
Speaker 2: fold laundry and do, do real work. Um, and we
54:37
Speaker 2: have the right sensors so we can like see the
54:38
Speaker 2: world and understand what to go do and use a,
54:41
Speaker 2: you know, our biological neural net to kind of figure
54:43
Speaker 2: out how to reason from. And, um, you know, having
54:47
Speaker 2: worked on like, you know, kind of like aircraft for,
54:51
Speaker 2: you know, now five or six years, I, I thought
54:53
Speaker 2: it was pretty possible to go build an electric humanoid robot.
54:58
Speaker 2: And electric's important for cost, and it's important for safety,
55:02
Speaker 2: and it's important because the performance will be much greater.
55:05
Speaker 2: And at the time, I mean, one of the best
55:07
Speaker 2: humanoid robots then was probably like the Boston Dynamics Atlas.
55:09
Speaker 2: It had like a hydraulic system. It was like really
55:13
Speaker 2: heavy and big and high torque and very leaky, like
55:16
Speaker 2: the oils everywhere. And also didn't run very, like maybe
55:19
Speaker 2: ran for 20 minutes on a single charge. So you
55:21
Speaker 2: need to kind of radically transform the hardware. And then
55:24
Speaker 2: you needed to figure out a way to build like
55:27
Speaker 2: an AI brain. The humanoid is so complex. It has like,
55:32
Speaker 2: let's call it like 40 degrees of freedom. And degrees
55:35
Speaker 2: of freedom is like a joint. So like an elbow
55:37
Speaker 2: is a degrees of freedom. Your shoulder's got three, a
55:40
Speaker 2: ball and socket has three, like a pitch on roll.
55:43
Speaker 2: And our robot has about, let's call it 40 degrees
55:44
Speaker 2: of freedom in it. Each degree of the freedom is
55:47
Speaker 2: a motor that can spin 360 degrees. So if you
55:50
Speaker 2: only look at how many positions the body could be
55:52
Speaker 2: in at any given time, like this is a position,
55:54
Speaker 2: this is a position, and keep moving, the amount of states. Mathematically,
55:59
Speaker 2: it's 360 degrees to the power of 40 actuators. So
56:03
Speaker 2: there are more states in the robot than atoms in
56:06
Speaker 2: the universe. There's more positions the body can be in.
56:10
Speaker 1: No shit.
56:11
Speaker 2: By far. It's a much greater number. I've done the
56:14
Speaker 2: math a few times. Very confident in this, even though
56:16
Speaker 2: it sounds ridiculous. So you just can't code your way
56:20
Speaker 2: out of this problem. How are you supposed to write code?
56:22
Speaker 2: How's a human supposed to write C + + or
56:25
Speaker 2: code to tell the robot at any given timestamp what
56:28
Speaker 2: to go do? If I want to grab this, I
56:29
Speaker 2: need to move my whole upper body and maybe lean over.
56:32
Speaker 2: And I'm moving my fingertips and my hand, my wrist
56:38
Speaker 2: and hand get in position to grab this. It's an
56:43
Speaker 2: intractable problem for code.
56:44
Speaker 1: I mean, you were saying earlier, I'm going to butcher this,
56:49
Speaker 1: but it's updating the foot 200 times a second.
56:54
Speaker 2: Yeah, our controller is running... For balance. Our whole controller,
56:57
Speaker 2: so we have a main computer, is processing what to
57:00
Speaker 2: tell all the joints to do 200, a little maybe
57:03
Speaker 2: more than 200 times a second to make sure we
57:05
Speaker 2: can just balance. And then we can do the task.
57:08
Speaker 2: It could be reaching over and grabbing this or balancing.
57:11
Speaker 2: If we run that too slow, we just don't have
57:12
Speaker 2: enough feedback. and we just fall over, just like, yeah,
57:16
Speaker 2: we have to fully balance. It's dynamic. So generally, if
57:20
Speaker 2: you power it off mid-run, it's going to just fall down.
57:22
Speaker 2: It's not like a four-legged dog or a quadruped robot
57:26
Speaker 2: where at any given point, it's usually statically stable. So
57:29
Speaker 2: it makes it very difficult because you have to be
57:32
Speaker 2: able to even move your hand. I'm moving my pelvis
57:35
Speaker 2: and my whole body. My torso's moving. My head's moving.
57:38
Speaker 2: All of it becomes very complicated now. It's not just
57:40
Speaker 2: like move my hand. It's like move my whole body
57:41
Speaker 2: to get my hand in the right spot. So every joint,
57:45
Speaker 2: all those 40 joints have basically position encoders. So we
57:49
Speaker 2: know exactly what position the motor is at, or even
57:52
Speaker 2: the case of the knee or this. And we have
57:55
Speaker 2: force sensing, torque sensing on board. We have the ability
57:59
Speaker 2: to detect all the forces that that knee is seeing.
58:02
Speaker 2: It could be really high when it's walking, or it
58:03
Speaker 2: could be like, you know, it could be like, it
58:06
Speaker 2: could be powered off and have no forces on the leg.
58:08
Speaker 2: All of that feedback is being sent in the main
58:12
Speaker 2: computer And then we're telling all the joints what to
58:15
Speaker 2: do over 200 times a second. Some of the other
58:19
Speaker 2: feedback is happening at like five or six kilohertz. So
58:21
Speaker 2: the force feedback is happening five, 6,000 times a second
58:25
Speaker 2: to the motor control on board. And we do the
58:28
Speaker 2: motor control. The brain for all the motors has done
58:30
Speaker 2: it locally at the motor level because it needs to
58:32
Speaker 2: happen so fast. And that's being fed back to a
58:34
Speaker 2: main computer that runs a control software that tells the
58:38
Speaker 2: rest of the whole body what to go do at
58:39
Speaker 2: every timestamp to keep balance.
58:45
Speaker 1: Starting something new isn't just hard, it's terrifying. Before I
58:50
Speaker 1: launched The Sean Ryan Show, those what-ifs were loud. What
58:55
Speaker 1: if nobody listens? What if this fails? Walking away from
58:59
Speaker 1: what's familiar to build your own thing takes real faith,
59:03
Speaker 1: but it ended up being one of the best decisions
59:05
Speaker 1: I've ever made. Whether you're starting a podcast or launching
59:08
Speaker 1: a store, it helps to have a partner like Shopify
59:12
Speaker 1: on your side to help ease those worries with their
59:15
Speaker 1: expertise and tools. Shopify is the commerce platform behind millions
59:20
Speaker 1: of businesses around the world and 10% of all e-commerce
59:25
Speaker 1: in the U.S. From household names like FIGS to brands
59:29
Speaker 1: that are just getting started today, one of the biggest
59:32
Speaker 1: hurdles is efficiency, and Shopify helps you accelerate that. Whether
59:37
Speaker 1: you're uploading new products or trying to improve existing ones,
59:41
Speaker 1: the platform is packed with helpful AI tools that can
59:44
Speaker 1: write your product descriptions, page headlines, and even enhance your
59:49
Speaker 1: product photography for you. They also allow you to tackle
59:53
Speaker 1: all your important tasks in one place, from inventory to
59:57
Speaker 1: payments to analytics. There's no need to juggle multiple websites.
1:00:02
Speaker 1: Everything is in one spot to make your operations smoother.
1:00:06
Speaker 1: And you've definitely seen that iconic purple Shop Pay button
1:00:11
Speaker 1: is used by millions because Shopify has the best converting
1:00:16
Speaker 1: checkout on the planet. It helps boost your conversions, which
1:00:19
Speaker 1: means fewer abandoned carts and more sales for your business.
1:00:23
Speaker 1: It's time to turn those what-ifs into with Shopify today.
1:00:29
Speaker 1: Sign up for your $ 1 per month trial at shopify.com.srs.
1:00:35
Speaker 1: Go to shopify.com.srs. That's Shopify.com. dot com slash SRS.
1:01:19
Speaker 2: So getting back to your original thing is my bet,
1:01:22
Speaker 2: you know, we're like three and a half years old
1:01:24
Speaker 2: or something like that figure. I bet three and a
1:01:26
Speaker 2: half years ago in 2022 when I started the company
1:01:29
Speaker 2: was that this was possible now. And my view is
1:01:34
Speaker 2: like it was, I don't, over some 10 or 20 years,
1:01:36
Speaker 2: this will work. And so I basically started on this
1:01:42
Speaker 2: endeavor to go basically rebuild from the ground up humanoid
1:01:46
Speaker 2: robots and AI software to try to see if you
1:01:48
Speaker 2: could make this work At the time, there was no
1:01:50
Speaker 2: good precedent. There was no precedent for showing that there
1:01:54
Speaker 2: was no AI that had ever worked on a human
1:01:55
Speaker 2: or robot in history. And there was no electric humanoid
1:02:01
Speaker 2: hardware that was remotely okay to show it would work.
1:02:06
Speaker 2: And there was no hands. There was none of this stuff.
1:02:08
Speaker 1: Wow.
1:02:09
Speaker 2: So I actually had a lot of trouble even... I
1:02:12
Speaker 2: had a lot of trouble early on even getting people
1:02:14
Speaker 2: excited about this because they were like, what the hell
1:02:16
Speaker 2: are you doing? And so I ended up having to
1:02:21
Speaker 2: like basically self-fund a lot of it in the first
1:02:25
Speaker 2: year I self-funded all of it.
1:02:26
Speaker 1: No kidding.
1:02:27
Speaker 2: Yeah. And it was a lot. I mean, we got
1:02:29
Speaker 2: the business to a million a month of burn in
1:02:32
Speaker 2: month four. But it was like, I knew what to do.
1:02:35
Speaker 2: We built a 40 person team and like as fast
1:02:37
Speaker 2: as possible. And I knew how to spin up hardware
1:02:40
Speaker 2: and software. The key characteristics of robotics, like electric motors,
1:02:45
Speaker 2: battery systems, you know, control software, embedded systems and sensors.
1:02:50
Speaker 2: And then even with electric motors, we build actuators. They
1:02:52
Speaker 2: have a rotor and stator and a gearbox and sensors,
1:02:56
Speaker 2: electronics and wiring and connectors and multiple sensors inside of there.
1:03:01
Speaker 2: And then firmware lives on the microcontroller. It lives on
1:03:07
Speaker 2: the motor control side. And then we have thermal characteristics.
1:03:10
Speaker 2: It's hot. And then you got to make that work
1:03:12
Speaker 2: at very high speeds and high torques, meaning motors don't
1:03:15
Speaker 2: like working when they're not moving. Motors hate not moving.
1:03:18
Speaker 2: Motors want to run on Highway speeds. They love that.
1:03:23
Speaker 2: Whether it's a generator, an appliance in your home, or
1:03:27
Speaker 2: an electric car, they want to run at highway speeds.
1:03:30
Speaker 2: They're designed to run at full RPMs. That's when they're
1:03:32
Speaker 2: the most efficient. When motors are stuck and not moving,
1:03:35
Speaker 2: but holding power and holding forces, it's a really bad
1:03:39
Speaker 2: point of the torque speed curve. They're not built well
1:03:44
Speaker 2: for this. Humanoid is used all the time. When we're standing,
1:03:48
Speaker 2: we're not moving, but holding forces. When we're holding something
1:03:51
Speaker 2: out and giving me the gummies, it's not moving now,
1:03:54
Speaker 2: but holding forces. It's just a really hard engineering problem.
1:03:57
Speaker 2: Just that one little aspect, which is hardware umbrella, just motors.
1:04:02
Speaker 2: We have whole teams in those areas, just in that
1:04:04
Speaker 2: one little area here doing rotor design, electromagnetics design, sator design,
1:04:10
Speaker 2: gearbox design, sensor design, motor control design, like all of
1:04:15
Speaker 2: this inside of Teams. It was an enormous lift to
1:04:18
Speaker 2: just get the team members there to do it. So
1:04:20
Speaker 2: we spent up a team to go operate that really quick.
1:04:24
Speaker 2: And then, you know, now like looking back, I think
1:04:25
Speaker 2: we raised, you know, two billion or so like now.
1:04:30
Speaker 2: It's a much different story. We have, you know, we
1:04:34
Speaker 2: ended up building figure one, which is our first generation
1:04:35
Speaker 2: robot and had that walking in under 12 months.
1:04:39
Speaker 1: So from when we- From inception to- Wow.
1:04:42
Speaker 2: Yeah, from when we basically incorporated the company in 2022,
1:04:46
Speaker 2: our goal was like, can we get a robot walking
1:04:49
Speaker 2: by itself in these dynamics in under 12 months? And
1:04:52
Speaker 2: we did it. We did it with like two days
1:04:55
Speaker 2: left in the year. At the time, I think it
1:04:58
Speaker 2: was the fastest time in history for anybody to do this.
1:05:01
Speaker 2: And then from there, continue to build the capabilities. We
1:05:06
Speaker 2: built Generation 2, Figure 2, which is that guy right there.
1:05:08
Speaker 2: It's our second generation robot. And I think one thing
1:05:14
Speaker 2: we did before we kind of moved even while designing
1:05:16
Speaker 2: Gen 2 is, I think it was probably like 2023
1:05:19
Speaker 2: at the time, we did a demonstration where we basically
1:05:23
Speaker 2: wanted to put this K-cup inside this Keurig and run it.
1:05:27
Speaker 2: And it was just on a pretty simple, like nothing crazy,
1:05:30
Speaker 2: but we had a Keurig machine, coffee cup, and a K-cup.
1:05:33
Speaker 2: And we had to go grab the K-cup, open the Keurig,
1:05:36
Speaker 2: put it in, close it, run it, and then make coffee. And,
1:05:41
Speaker 2: you know, it sounds simple, but like for a humanoid
1:05:44
Speaker 2: robot to do that is extremely hard. And then we
1:05:47
Speaker 2: wanted to do all of that with just neural networks.
1:05:49
Speaker 1: It sounds simple, but I mean- Simple task for humans.
1:05:52
Speaker 1: Give it to a four-year-old.
1:05:53
Speaker 2: Yeah, exactly.
1:05:54
Speaker 1: I mean, the dexterity on the hands of that thing
1:05:57
Speaker 1: is just to hand you the bag of gummy bears.
1:05:59
Speaker 2: Yeah. Yeah. Yeah, then can you do it with neural
1:06:03
Speaker 2: nets on board?
1:06:04
Speaker 1: It's crazy.
1:06:04
Speaker 2: Can you not code your way out of it? Can
1:06:06
Speaker 2: you have a, can you take in camera pixels and
1:06:08
Speaker 2: then output trajectories for the motors through a neural network,
1:06:12
Speaker 2: no code. And we did that in 2023 on figure one.
1:06:17
Speaker 2: And it was like probably the most significant demonstration we've
1:06:21
Speaker 2: done in four years now, almost four years, where we
1:06:25
Speaker 2: were like, internally, we were like, you know, how do
1:06:28
Speaker 2: we get neural nets to run on a humanoid? I
1:06:30
Speaker 2: don't know. I think it's probably one of the first
1:06:32
Speaker 2: examples in the world to ever have shown that. And,
1:06:37
Speaker 2: you know, this was like, game on. This is like,
1:06:39
Speaker 2: we have, let's go build a really good human on hardware.
1:06:41
Speaker 2: Let's make it cheap and really reliable. Let's make sure
1:06:44
Speaker 2: it can do what humans can from a hardware perspective.
1:06:47
Speaker 2: Meaning you want to look at like a phone where
1:06:49
Speaker 2: you can just add new apps to it. Like the, uh,
1:06:52
Speaker 2: do laundry app and the hardware doesn't need to change
1:06:56
Speaker 2: in the same exact hardware. Like human is only like new,
1:06:58
Speaker 2: like new hardware, uh, to be able to go off
1:07:01
Speaker 2: and like learn how to do new skill now in
1:07:03
Speaker 2: the physical world. So you want to build the humanoid
1:07:05
Speaker 2: hardware so it can do everything basically a human can
1:07:08
Speaker 2: or as most of possible. And then you want to
1:07:10
Speaker 2: go all in on neural networks because you just can't
1:07:12
Speaker 2: code your way out of this problem. And that was
1:07:14
Speaker 2: the first moment in 2023 where we're like, hot damn,
1:07:17
Speaker 2: this is going to really work. This is going to
1:07:20
Speaker 2: be humanoid robots. Hardware gets good. And then you're basically
1:07:24
Speaker 2: going to be, this is going to be a data
1:07:25
Speaker 2: play to train neural networks to run on humanoid hardware
1:07:30
Speaker 2: and do what humans do. And then we launched you know,
1:07:36
Speaker 2: we launched figure two, we did a lot, basically more work.
1:07:39
Speaker 2: We started unveiling Helix, which is our neural network stack
1:07:41
Speaker 2: internally that we do here. And now we've designed figure three,
1:07:47
Speaker 2: which is our third generation robot you have here, which
1:07:49
Speaker 2: is like the best human hardware in the world by far.
1:07:53
Speaker 2: And we're now running robots that do like, I watched it,
1:07:56
Speaker 2: you know, the other day, unload dishes and fold laundry.
1:08:01
Speaker 2: We had figure twos at BMW last year that worked
1:08:04
Speaker 2: six months every single day.
1:08:08
Speaker 1: Six months every single day.
1:08:09
Speaker 2: Every single day it worked. A 10-hour shift every day
1:08:11
Speaker 2: for six months. And we had, it was just, it
1:08:16
Speaker 2: was like the first time for us getting robots out
1:08:18
Speaker 2: to the real world, doing real stuff. Like, you know,
1:08:20
Speaker 2: it's fun doing like, you know, demos at the office
1:08:23
Speaker 2: and showing it can really work. But the real like
1:08:26
Speaker 2: level boss is like, how do we get robots out?
1:08:28
Speaker 2: And do clients fire us? Do they love it? Does
1:08:30
Speaker 2: it work? And the goals we have for clients is
1:08:32
Speaker 2: hard because we have to do human work. So we
1:08:34
Speaker 2: get like human KPIs in terms of speed and performance.
1:08:36
Speaker 2: Like humans, like, you know, in the case of manufacturing,
1:08:40
Speaker 2: they don't like, you might mess up every once in
1:08:42
Speaker 2: a while, but you like refix it. So you're not
1:08:44
Speaker 2: messing up every single time. You're pretty fast. In most cases,
1:08:48
Speaker 2: the humans are there, not like, you know, quitting or
1:08:50
Speaker 2: not showing up to work, but sometimes that does happen.
1:08:52
Speaker 2: So it's like, it's hard. It's a hard bar to
1:08:54
Speaker 2: go hit. And we have to wake up every day
1:08:56
Speaker 2: and be able to do that. And so we had
1:08:57
Speaker 2: robots in the manufacturing line They have a, they basically
1:09:01
Speaker 2: have a basic body shop that basically builds like X3
1:09:04
Speaker 2: and X5s. And January of 2020, what was.
1:09:12
Speaker 1: It?
1:09:12
Speaker 2: 2025, we started building our first BMW X3s on the line.
1:09:19
Speaker 2: Are you serious? I bought the first four that did that.
1:09:22
Speaker 2: They're on my campus now. One's in my house. And
1:09:26
Speaker 2: they didn't build, obviously, the whole car. There's a ton
1:09:28
Speaker 2: of parts, but we did a part of the whole process.
1:09:32
Speaker 1: What's BMW's feedback?
1:09:36
Speaker 2: They're great. I mean, BMW is like, if you go
1:09:39
Speaker 2: into a car manufacturing company, they're like the best roboticists
1:09:42
Speaker 2: in the world. There's robots everywhere. There's these giant 12-foot
1:09:49
Speaker 2: KUKA manufacturing robot arms on the floor. They're bolted to
1:09:54
Speaker 2: the ground. They're massive. These things are carrying car chassis
1:09:58
Speaker 2: around like they're kids' toys. The cars are so big
1:10:02
Speaker 2: and so heavy, a human can't hold it and pass
1:10:04
Speaker 2: it around. So you basically have machines that are building
1:10:07
Speaker 2: the car and then moving the car. So the whole
1:10:10
Speaker 2: body shop line is only automated end-to-end. And it's like,
1:10:15
Speaker 2: not end to end, but like there's humans involved, but
1:10:16
Speaker 2: like the car is being like built by machines. And
1:10:20
Speaker 2: then there's machines everywhere. There's special end effectors on every machine.
1:10:24
Speaker 2: They're switching these things out basically in real time, like
1:10:27
Speaker 2: grabbing a big end effector. The end effector is something
1:10:30
Speaker 2: that is grabbing a part. These end effectors are the
1:10:34
Speaker 2: size of my chair. They're switching them out in seconds.
1:10:37
Speaker 2: One or two seconds. They're doing it really fast. And
1:10:40
Speaker 2: then they're basically building a car with this. There's robots everywhere.
1:10:43
Speaker 1: And....
1:10:45
Speaker 2: So like BMW, it's been a privilege to see how
1:10:49
Speaker 2: much automation has gone into automotive. It's unbelievable. These machines
1:10:53
Speaker 2: kind of make what we're doing sometimes look like little toys.
1:10:56
Speaker 1: No kidding.
1:10:57
Speaker 2: We're doing something very complicated, but the car manufacturing is
1:11:00
Speaker 2: just like no joke.
1:11:01
Speaker 1: So what specifically were the figures doing?
1:11:04
Speaker 2: Yeah, there's a body shop line that's basically building the
1:11:07
Speaker 2: rear header. It's like the back plate. So the body shop,
1:11:11
Speaker 2: they basically build the car by putting basically sheet metal together.
1:11:14
Speaker 2: welding them onto the chassis, and then they're basically building
1:11:17
Speaker 2: the car around that. Like you end up putting the
1:11:19
Speaker 2: seats in, bolting them down, putting the car doors in,
1:11:21
Speaker 2: wiring them up, the harnessing. And we were in the
1:11:24
Speaker 2: body shop line helping basically attach the rear head, like
1:11:27
Speaker 2: basically putting the rear header on this fixture. So today
1:11:31
Speaker 2: we basically, or last year when we were there, we
1:11:33
Speaker 2: basically take a piece of sheet metal and we basically
1:11:35
Speaker 2: put it on this fixture and we do that over
1:11:38
Speaker 2: and over again. And they do that, you know, 10
1:11:41
Speaker 2: hours full shift. And there's three different parts on. Those
1:11:45
Speaker 2: three parts go on. This thing rotates and this big
1:11:48
Speaker 2: giant kooka machine, like this robot arm, goes and spot
1:11:50
Speaker 2: welds it and switches it, switches out to another effector
1:11:54
Speaker 2: and then grabs it and puts it down the line. Wow.
1:11:56
Speaker 2: So like these, these, these facilities are being like fed
1:12:00
Speaker 2: these parts into the machine. And we were like, we
1:12:04
Speaker 2: were a piece of that. And the goal was just like,
1:12:08
Speaker 2: can we run robots every day? You know, are we
1:12:10
Speaker 2: going to get our ass handed to us? Is it
1:12:13
Speaker 2: going to be easy? Is it going to be hard?
1:12:15
Speaker 2: And I think it was in the middle. I think
1:12:17
Speaker 2: we got the robot to a great spot where it
1:12:21
Speaker 2: ran every day. It was great. I think the biggest
1:12:22
Speaker 2: learning lesson we took away is we had... I really
1:12:26
Speaker 2: cared about, can we do that? And can we clone
1:12:28
Speaker 2: it times 1,000, times 10,000? Would we have any issues scaling?
1:12:32
Speaker 2: That was the part for me. Is it just like,
1:12:35
Speaker 2: does it completely shit the bed? And he's like, we
1:12:37
Speaker 2: need to rework our plan and go back to the office.
1:12:39
Speaker 2: Does it do it incredibly well and you can just
1:12:41
Speaker 2: copy-paste this thing to everywhere in the world? Like, how
1:12:43
Speaker 2: does it, how did it work? And the biggest learning
1:12:45
Speaker 2: lesson we got was that the robots, the robot that
1:12:47
Speaker 2: started the first day at the start of six months
1:12:50
Speaker 2: and the robot that ended the shift that day was
1:12:52
Speaker 2: the same. Like, even though we had multiple robots in
1:12:54
Speaker 2: operation every day, we had like the same robot that
1:12:58
Speaker 2: did the start and the finish.
1:12:59
Speaker 1: Wow.
1:13:00
Speaker 2: And it was cool. Like, you know, like it was
1:13:01
Speaker 2: same robot ended six months later. And this was the
1:13:03
Speaker 2: thing where, you know, the worry was that humanoid robots
1:13:06
Speaker 2: couldn't last a month, couldn't last a week in these
1:13:08
Speaker 2: kinds of environments. And wear and tear, it's running 40
1:13:13
Speaker 2: degrees of freedom motors every single day. Can they operate
1:13:16
Speaker 2: really well? I think from a hardware perspective, it did
1:13:19
Speaker 2: an A plus job. I think from a software perspective,
1:13:22
Speaker 2: we did, I would say, a B job, B plus.
1:13:27
Speaker 2: And that's mostly from my perspective, the architecture decisions I
1:13:30
Speaker 2: made to scale. About half the stack, we had traditional
1:13:36
Speaker 2: code and heuristics in. So like the controller to walk
1:13:39
Speaker 2: was done by a C + + controller. It was
1:13:42
Speaker 2: done by code. The walking you saw today we had
1:13:45
Speaker 2: back then was done in code.
1:13:46
Speaker 1: Okay.
1:13:47
Speaker 2: The rest of this, we had a bunch of other
1:13:48
Speaker 2: stuff in there done by neural nets. And like some
1:13:51
Speaker 2: of the perception stacks, some like, you know, some of
1:13:54
Speaker 2: the parts around everything else. And, you know, this was
1:13:58
Speaker 2: a year and a half or so ago when we
1:14:00
Speaker 2: were first launching. And I was like, man, the biggest
1:14:05
Speaker 2: problems we're having is the coding parts get stuck. The
1:14:08
Speaker 2: robot either doesn't see something right on the part and
1:14:13
Speaker 2: misses the object detector, doesn't really understand what's going on.
1:14:17
Speaker 2: The controller, when it gets out of bounds of what
1:14:20
Speaker 2: it's ever seen before, you have carpet in here now
1:14:22
Speaker 2: and it's really squishy. The robot's doing fine, which is great,
1:14:24
Speaker 2: but I think our old controller would not do well.
1:14:26
Speaker 2: You have really shaggy carpet here. It's harder for a
1:14:29
Speaker 2: robot to walk around. So that was like, we had
1:14:36
Speaker 2: a really difficult time seeing that, even though it did
1:14:39
Speaker 2: well every day, seeing that scale to like lots of robots.
1:14:42
Speaker 2: So we went back to the office, this is about
1:14:44
Speaker 2: a year and a half ago, and said, we need
1:14:45
Speaker 2: to basically refactor everything into a neural network. And one
1:14:49
Speaker 2: of the big, and I think we just announced Helix 2,
1:14:53
Speaker 2: two or three months ago now, I forget, like end
1:14:55
Speaker 2: of last year. And it's basically entirely down the stack,
1:14:59
Speaker 2: including the controllers and neural net now. There's like no
1:15:03
Speaker 2: code left really on the robot. some code in certain pieces,
1:15:06
Speaker 2: but mostly just almost all of the thing is like
1:15:08
Speaker 2: a neural net at this point. We removed the need
1:15:10
Speaker 2: for almost over 100,000 lines of code when we launched
1:15:14
Speaker 2: Helix 2. And so what you saw today was just
1:15:17
Speaker 2: like a robot that we can put now back in, say,
1:15:21
Speaker 2: the factory and these places that will run only on
1:15:23
Speaker 2: a neural net. And I think we're running these robots
1:15:26
Speaker 2: right now, we're getting ready for deployment to customers, and
1:15:28
Speaker 2: they're running incredibly well. We have robots running basically now
1:15:33
Speaker 2: in 24-7 shifts without stopping, without any faults for like
1:15:39
Speaker 2: days and days. We just went like over, we just
1:15:43
Speaker 2: had like record time this past week on the robot
1:15:46
Speaker 2: running until we saw like a fault, like almost, yeah,
1:15:49
Speaker 2: basically a whole week. And they basically like, they can
1:15:52
Speaker 2: run like four hours or so, five hours and when
1:15:57
Speaker 2: you need to charge, another robot knows that, steps in,
1:16:00
Speaker 2: steps behind the robot, say, and gets ready for work.
1:16:02
Speaker 2: The robot then backs off They're about swaps in spot,
1:16:05
Speaker 2: like in the next 10 seconds it's doing work again.
1:16:07
Speaker 2: So we can run that now in 24 seven shifts
1:16:09
Speaker 2: where they're talking to each other, all autonomous, no humans
1:16:12
Speaker 2: or you can go to bed or whatever. And they're
1:16:15
Speaker 2: running that shifts all day and all night. And we
1:16:17
Speaker 2: do it across multiple use cases now at the office
1:16:19
Speaker 2: in 24 seven shifts. And it's just like, we're running
1:16:21
Speaker 2: them hard.
1:16:22
Speaker 1: What kind of stuff are they doing at your office?
1:16:23
Speaker 2: We do a few things. We have a logistics use
1:16:25
Speaker 2: case that we run in 24 seven shifts constantly. We
1:16:29
Speaker 2: really like it. It's done with a neural net. It's
1:16:31
Speaker 2: moving packages around. And, uh, it's a really good use case.
1:16:34
Speaker 2: We like, uh, we like it. And we, um, we,
1:16:37
Speaker 2: we want to like, I want to run it for
1:16:38
Speaker 2: months and have a failures and we still, we still,
1:16:41
Speaker 2: we see failures right now. And it's most of it's
1:16:42
Speaker 2: in software. Robot gets to some spot where it feels unsafe,
1:16:46
Speaker 2: doesn't know what to do. And it'll stop for a
1:16:47
Speaker 2: little bit. And then the, you know, the robots not
1:16:49
Speaker 2: on the line for a couple of minutes, we call
1:16:50
Speaker 2: it a failure and we're not happy with it. Um,
1:16:53
Speaker 2: we have robots that are greeters and visitor boss that
1:16:56
Speaker 2: walk around the office all day, 24 seven. So you're
1:16:58
Speaker 2: over the office, you're getting lunch or walking around or
1:17:00
Speaker 2: interviewing with us. You see robots everywhere. And those run
1:17:03
Speaker 2: in 24-7 shifts, all day, all night, weekends, Christmas day,
1:17:06
Speaker 2: whatever we run them.
1:17:07
Speaker 1: Greeters?
1:17:08
Speaker 2: Yeah.
1:17:09
Speaker 1: How do they greet you?
1:17:10
Speaker 2: Come talk to you.
1:17:11
Speaker 1: They'll just come talk to you?
1:17:12
Speaker 2: Yeah, they'll come talk to you. And like, you can
1:17:14
Speaker 2: go talk to it and ask it for things. We
1:17:18
Speaker 2: really wanted to go like, you know, the end state
1:17:21
Speaker 2: for us is like, it's going to replace like somebody
1:17:23
Speaker 2: like meeting the candidates that are interviewing there, taking them
1:17:25
Speaker 2: to the conference room, getting them water or coffee, like
1:17:27
Speaker 2: all of that end to end. and whole experience. Holy shit. Yeah,
1:17:31
Speaker 2: right now they're walking, I mean, right now they're walking
1:17:33
Speaker 2: in the office, like at nighttime, they're walking office everywhere.
1:17:36
Speaker 2: And it's a good stress test for us because these
1:17:38
Speaker 2: are neural nets that are running for navigation or planning
1:17:44
Speaker 2: or manipulation or whatever it would look like. And it's hard.
1:17:48
Speaker 2: This is a new thing. It's not like these things
1:17:49
Speaker 2: have been around for decades and we like understand that
1:17:53
Speaker 2: they're really mature, they're not. So we really stress test
1:17:55
Speaker 2: them like crazy by running them all the time.
1:17:57
Speaker 1: What's a conversation you've had with a robot?
1:18:01
Speaker 2: We've been really working on deep memory because I think
1:18:04
Speaker 2: one thing I really don't like is these conversational AIs
1:18:07
Speaker 2: you talk to that don't know anything about you. It's
1:18:10
Speaker 2: not much to talk about. It's like, what's the weather?
1:18:12
Speaker 2: You ask things about Wikipedia or something. It's the way
1:18:14
Speaker 2: to work. It's kind of nonsense. It kind of feels
1:18:18
Speaker 2: really stupid to me. So we've been working a lot
1:18:21
Speaker 2: on deep memory.
1:18:21
Speaker 1: So it will actually get to know you. Oh, yeah.
1:18:27
Speaker 2: Yeah, it needs to know who you are. Like, who
1:18:28
Speaker 2: am I talking to? Is this Sean or Brett? And
1:18:31
Speaker 2: then based on Sean, like, do you have the permissions
1:18:32
Speaker 2: to tell the robot to go do something or not? Like,
1:18:36
Speaker 2: if you're visiting, no. You might be able to get
1:18:38
Speaker 2: coffee or water, but like, you want to have it, like,
1:18:39
Speaker 2: go do something new. Like, it won't do it.
1:18:42
Speaker 1: I've not even thought of that either.
1:18:44
Speaker 2: Yeah, yeah.
1:18:45
Speaker 1: What do you want to be able to command? Yeah, yeah.
1:18:47
Speaker 2: What are the permissioning systems and authentications of the robots?
1:18:50
Speaker 2: I mean, like robots in my house, my kids are
1:18:53
Speaker 2: going to be like, hey, give me ice cream every
1:18:56
Speaker 2: 10 minutes and we can't have the robot doing that, right?
1:18:59
Speaker 2: You know, as I get home from work and the
1:19:01
Speaker 2: kids are just like, you know, through pints of ice
1:19:03
Speaker 2: cream and the robots are just getting whatever they need,
1:19:05
Speaker 2: like it'd just be chaos. Yeah. Yeah.
1:19:08
Speaker 1: So what is it, voice recognition?
1:19:11
Speaker 2: Yeah, you have to do voice for something. Something's voice
1:19:13
Speaker 2: isn't enough. If you think about it in an extreme example,
1:19:17
Speaker 2: you want it to go order food or spend money
1:19:19
Speaker 2: or send a wire, voice recognition won't be enough. You'll
1:19:23
Speaker 2: have to do a higher level of authentication.
1:19:25
Speaker 1: How would you do that?
1:19:26
Speaker 2: Facial recognition.
1:19:27
Speaker 1: Okay.
1:19:27
Speaker 2: And then if you have a... Perhaps even fingerprint scanning
1:19:32
Speaker 2: is possible too, but facial is what you really want
1:19:34
Speaker 2: to do.
1:19:35
Speaker 1: Gotcha. Yeah.
1:19:36
Speaker 2: So those are not, all those systems are not, like,
1:19:38
Speaker 2: robust enough right now. We're working through them. And, like,
1:19:42
Speaker 2: the goal is, like, to get it to, like, super robust. But, like,
1:19:44
Speaker 2: you know, we want to have conversations with the robot.
1:19:45
Speaker 2: We want to ask it to go do things. Like,
1:19:47
Speaker 2: you really, you want the main modality to be speech
1:19:50
Speaker 2: with robots. You want to just, like, hey, man, go
1:19:53
Speaker 2: make me, like, go make me, like, go make me food. Or, like,
1:19:56
Speaker 2: when I'm gone today, do the laundry. After you, like,
1:19:59
Speaker 2: unload the dishwasher, like, you know, do laundry in, like,
1:20:03
Speaker 2: my kids' rooms today or something like that. or text it.
1:20:07
Speaker 2: So language is a super important UI. So we're spending
1:20:10
Speaker 2: a lot of time on speech.
1:20:11
Speaker 1: You can text it too.
1:20:13
Speaker 2: Yeah, every robot we have has 5G by default on board.
1:20:18
Speaker 2: We actually run 5G by default now. So every robot
1:20:23
Speaker 2: off the line has 5G enabled. We have a T-Mobile 5G.
1:20:28
Speaker 2: T-Mobile is an investor of ours and every robot has
1:20:31
Speaker 2: an eSIM card for T-Mobile 5G. So it comes with
1:20:34
Speaker 2: a line, we use 5G. for all the main network.
1:20:37
Speaker 2: So if you want to like, you know, if our, like,
1:20:40
Speaker 2: if our systems want to tell the robot what to
1:20:42
Speaker 2: do or can the robot do something, we do it
1:20:44
Speaker 2: through 5G. And so, yeah, you can, like, you can
1:20:47
Speaker 2: text it.
1:20:48
Speaker 1: So you can be at work and say, hey, I
1:20:51
Speaker 1: want to get the pizza out of the freezer, put
1:20:53
Speaker 1: it in the oven.
1:20:54
Speaker 2: Yeah.
1:20:55
Speaker 1: 425 degrees, 15 minutes.
1:20:57
Speaker 2: Yeah. I mean, we can't do that in right now,
1:21:00
Speaker 2: but like, that's the goal is like, we got to
1:21:01
Speaker 2: get there. Like, we got to get to a point
1:21:03
Speaker 2: where like, that is certainly possible. And, um, you want
1:21:06
Speaker 2: that to happen. You want to be like, uh, yeah,
1:21:08
Speaker 2: I'm at work. Uh, when the groceries come, make sure
1:21:10
Speaker 2: you put them inside and, uh, put them in the
1:21:12
Speaker 2: fridge and do, do all this. Or they would even
1:21:14
Speaker 2: know that.
1:21:15
Speaker 1: Go check the mail, have it on the counter when
1:21:16
Speaker 1: I do.
1:21:17
Speaker 2: For sure.
1:21:17
Speaker 1: Like all, feed the dog.
1:21:19
Speaker 3: Yeah.
1:21:19
Speaker 1: Everything.
1:21:20
Speaker 2: Like watch the dog, make sure the dog's okay. Yeah.
1:21:22
Speaker 1: Holy shit. So it's, it's, it's, it's a nanny, housekeeper, gardener.
1:21:30
Speaker 2: It's, uh, the Jetsons.
1:21:31
Speaker 1: All of it.
1:21:33
Speaker 2: Yeah. It's going to be all of it. I mean,
1:21:35
Speaker 2: you might want to garden, you might want to do that.
1:21:37
Speaker 2: All this physical labor we do today, I think will
1:21:39
Speaker 2: be optional in the future. So you'll like, you might
1:21:43
Speaker 2: like gardening. Um, you know, you might like mowing the lawn.
1:21:46
Speaker 2: You just like mow the lawn. If you don't want
1:21:47
Speaker 2: to mow the lawn, don't mow the lawn. Like all
1:21:49
Speaker 2: this will be a choice.
1:21:50
Speaker 1: Holy shit.
1:21:52
Speaker 2: Yeah.
1:21:52
Speaker 1: And you said it, it's gonna, it'll download apps for
1:21:56
Speaker 1: different tasks.
1:21:57
Speaker 2: You want to think about the software layer. Like you
1:21:59
Speaker 2: want to think about the, so for us, like, um,
1:22:01
Speaker 2: what's so powerful about a humanoid is you, you don't
1:22:04
Speaker 2: want to go out and change hardware. whenever we have
1:22:06
Speaker 2: a new like app on your phone, you just like
1:22:08
Speaker 2: download it and connect, do new things. Now, like he's
1:22:10
Speaker 2: got my bank account. Now I can do bank account stuff,
1:22:12
Speaker 2: or you got to download, you got a calculator, can
1:22:14
Speaker 2: do calculator stuff. Uh, you really want to treat the
1:22:16
Speaker 2: hardware like this, where you basically similar to phone, where you, um,
1:22:20
Speaker 2: you don't have to change the hardware for new capabilities. Uh,
1:22:23
Speaker 2: you want it to learn how to do, uh, like,
1:22:25
Speaker 2: you know, complex towel folding or, um, like a load,
1:22:31
Speaker 2: like a load in the dishwasher. Um, making coffee on
1:22:35
Speaker 2: a Keurig, like all this, like walking the dog. Like
1:22:37
Speaker 2: these are like, almost like the matrix where you get
1:22:40
Speaker 2: like plugged into a system that re-uploads like weights into
1:22:45
Speaker 2: the neural net weights into the robot where it can
1:22:47
Speaker 2: like learn new things. So that's what we do now.
1:22:49
Speaker 2: Like if the robot, like if we can't do package
1:22:52
Speaker 2: logistics well, we get data for package logistics. We train
1:22:54
Speaker 2: our Helix neural net for a week and then we
1:22:58
Speaker 2: load it to the robot and it can like then
1:22:59
Speaker 2: the same robot that was like folding towels like the
1:23:02
Speaker 2: week before, can now just sit there for 24-7 and
1:23:05
Speaker 2: do logistics work and package work. Wow. Nothing changes.
1:23:10
Speaker 1: Where is this going to go first? Consumers, businesses?
1:23:13
Speaker 2: You'll ship into businesses first. The engineering complexity that we
1:23:19
Speaker 2: have to ship is proportional to the variability that we
1:23:22
Speaker 2: see on site. So the variability at homes is extremely high.
1:23:27
Speaker 2: My home is chaos. Kids are just like, just dismantling
1:23:31
Speaker 2: the house it was like it's basically in real time
1:23:33
Speaker 2: uh and then there's just like food or they're eating
1:23:36
Speaker 2: snacks toys like it's just like it's just chaos and
1:23:39
Speaker 2: then like you know if we go to your house
1:23:41
Speaker 2: in my house we probably have like different appliances or
1:23:43
Speaker 2: different toasters and different microwaves all a little different everywhere
1:23:46
Speaker 2: we go so the home is just like this like
1:23:48
Speaker 2: um tons of entropy like uh tons of veritability a
1:23:52
Speaker 2: wide distribution of tasks it's like the it's like the
1:23:56
Speaker 2: ultimate like challenge for robotics in the home. It's like
1:23:59
Speaker 2: the hardest, most variable thing we got going. And in
1:24:04
Speaker 2: the workforce, it's like, you have this like work cell
1:24:07
Speaker 2: that you're doing. So like, if you're doing like manufacturing
1:24:09
Speaker 2: logistics or, you know, a lot of tasks, you have
1:24:12
Speaker 2: like this area you're doing work in and you can
1:24:14
Speaker 2: basically kind of write down on a piece of paper,
1:24:15
Speaker 2: like how to do every step. In the home, you
1:24:18
Speaker 2: can't do that. You can't write down a piece of paper.
1:24:20
Speaker 2: I can't write down a piece of paper, like how
1:24:21
Speaker 2: I can interact with your house. I mean, you've seen it.
1:24:22
Speaker 2: But like the next assembly line or the next like,
1:24:26
Speaker 2: It's like, I kind of know what to do. It's like,
1:24:27
Speaker 2: I get the package, I flip it down and I
1:24:29
Speaker 2: need to do it every three seconds. Like you kind
1:24:30
Speaker 2: of have like, you know, a good understanding of what
1:24:31
Speaker 2: to go do. So, um, it just makes it easier.
1:24:35
Speaker 2: It's like, uh, the analogy would be like highway driving
1:24:38
Speaker 2: for autonomous vehicles. That's just happened sooner because the veritability
1:24:43
Speaker 2: is lower than in a city.
1:24:46
Speaker 1: Gotcha.
1:24:46
Speaker 2: Um, so it'll happen first at scale. And then the
1:24:48
Speaker 2: industrial thing has a good, good thing where it's like,
1:24:50
Speaker 2: you kind of have your own work area. So the
1:24:52
Speaker 2: safety areas are not as high. The hardest thing in
1:24:56
Speaker 2: the home will be once you figure out how to
1:24:58
Speaker 2: get performance there, meaning it's capable of doing everything in
1:25:00
Speaker 2: the home, like say you can go into your home
1:25:01
Speaker 2: and do everything, the longest pull from there is gonna
1:25:04
Speaker 2: be safety. Where like me and you feel safe, like being,
1:25:08
Speaker 2: like having this here with our kids. And that is,
1:25:13
Speaker 2: that's gonna be the hardest challenge by far. And that's
1:25:17
Speaker 2: gonna take some time. It's a very, there's some trust
1:25:19
Speaker 2: that needs to build, there's a tracker that needs to
1:25:21
Speaker 2: be built, there's like system safety engineering that needs to
1:25:23
Speaker 2: be done extremely well. So that just, and then the home,
1:25:27
Speaker 2: like you can charge like 10X, you can charge like
1:25:31
Speaker 2: 10X more in the commercial market than you can the home.
1:25:34
Speaker 2: Home needs to be like 500 bucks a month, your
1:25:36
Speaker 2: car lease.
1:25:38
Speaker 1: Those will be, you think those will be around 500
1:25:41
Speaker 1: bucks a month?
1:25:41
Speaker 2: Yeah, I think it'll be like that level, like, you know,
1:25:43
Speaker 2: that like order of like more magnitude, yeah. Yeah, so
1:25:47
Speaker 2: I think, and then the commercial workforce, you can charge
1:25:50
Speaker 2: like 10 times more. So like, so it's just like
1:25:55
Speaker 2: the commercial and in the commercial market for humanoids, like,
1:25:58
Speaker 2: you know, I mean, half of GDP is human labor,
1:26:00
Speaker 2: you know, maybe a little under half. So it's like
1:26:04
Speaker 2: 3 billion humans in the workforce is like contributes to
1:26:06
Speaker 2: like 40 something percent of GDP.
1:26:10
Speaker 1: Wow.
1:26:10
Speaker 2: So like you're talking about the largest market in the
1:26:12
Speaker 2: world is sitting in the commercial workforce.
1:26:14
Speaker 1: Wow.
1:26:15
Speaker 2: So you have like, you have like that plus the
1:26:17
Speaker 2: variability is lower. Plus you can charge 10 times more.
1:26:19
Speaker 2: It's like the, like for investors are like, dude, why
1:26:23
Speaker 2: would you ever work? Why would you ever do homework?
1:26:26
Speaker 2: You know what I mean? Like, why would you like
1:26:27
Speaker 2: spend time over here when you can just go over
1:26:29
Speaker 2: here and build like a $ 20 trillion company? And my
1:26:33
Speaker 2: answer for that is just like, I just want to,
1:26:35
Speaker 2: I want robots in the home. So don't really care.
1:26:40
Speaker 2: You know, like we got to make that work. Yeah.
1:26:43
Speaker 1: I mean, you say in 10 years, every home will
1:26:47
Speaker 1: have a humanoid.
1:26:50
Speaker 2: We don't have every home in 10 years, but we
1:26:51
Speaker 2: will have.
1:26:52
Speaker 1: Pretty close.
1:26:53
Speaker 2: We will have in 10 years.
1:27:00
Speaker 2: You have like two long pulls. You have like a
1:27:02
Speaker 2: long pull with manufacturing enough volumes for this. And then
1:27:05
Speaker 2: you have a long pull where you can actually technically
1:27:07
Speaker 2: do the work fully end-to-end. My belief is that the
1:27:11
Speaker 2: hardest thing in the stack is not manufacturing. The hardest
1:27:16
Speaker 2: thing in the stack is... Sorry, the hardest hill right
1:27:19
Speaker 2: now is can you put a robot into your home
1:27:23
Speaker 2: today and do the five hours of work you need
1:27:27
Speaker 2: without ever seeing your home before? The first group to
1:27:31
Speaker 2: do that, I think will become like the largest company
1:27:34
Speaker 2: in the world. And you can do that with maybe
1:27:38
Speaker 2: a hundred robots.
1:27:40
Speaker 1: No shit.
1:27:41
Speaker 2: Yeah, I think you can solve a general purpose humanoid robot.
1:27:43
Speaker 2: I think you can solve general purpose robotics with maybe
1:27:46
Speaker 2: like hundreds or low thousands of robots. Maybe a hundred.
1:27:51
Speaker 1: How so?
1:27:53
Speaker 2: Um, at this point, the issue we have, so we
1:27:57
Speaker 2: can go into my home today and we can do
1:27:59
Speaker 2: little pockets of work. We can do like, I can
1:28:01
Speaker 2: unload the full dishwasher. I can, once like laundry's in
1:28:06
Speaker 2: the basket, I can take it, walk it and fill
1:28:08
Speaker 2: up the, uh, the washer and run it. And we
1:28:12
Speaker 2: can do pockets of work. We can like take the laundry,
1:28:15
Speaker 2: put it on my, like my bed and we can
1:28:17
Speaker 2: fold it all. And then, uh, So we're doing like
1:28:21
Speaker 2: little spots of it and it's pretty good. But there's
1:28:27
Speaker 2: a lot more spots to go fill for like long
1:28:29
Speaker 2: horizon work than just that. And we have to be
1:28:32
Speaker 2: like extremely robust to maybe different types of clothes or
1:28:36
Speaker 2: like different types of like, don't wash my jeans. like
1:28:40
Speaker 2: that type of thing. And all these different like variability
1:28:42
Speaker 2: that you might see. And we haven't been able to,
1:28:47
Speaker 2: as of today, that's like, that's the hill we got
1:28:48
Speaker 2: to go solve. That hill looks really hard.
1:28:51
Speaker 1: So how, I mean, how, let's say, let's fast forward
1:28:55
Speaker 1: 10 years. I'm getting one of these guys. I put
1:28:59
Speaker 1: them in the home. How does it, I mean, do
1:29:01
Speaker 1: I train it? Do I personally train it? Hey, when
1:29:03
Speaker 1: you're emptying the dishwasher, the cups go here, the plates
1:29:06
Speaker 1: go here, the silverwares go here, the forks go here.
1:29:10
Speaker 1: When you're doing the laundry, I want these ones washed cold.
1:29:13
Speaker 1: I want these ones washed hot. This is where they go.
1:29:17
Speaker 1: This is where the jeans drawer is. This is where
1:29:19
Speaker 1: I hang my shirt. Is that how it works?
1:29:22
Speaker 2: Yeah, you'll get a robot in a box. You open
1:29:25
Speaker 2: it up, robot get out. It'll start talking to you.
1:29:28
Speaker 2: It'll ask you.
1:29:30
Speaker 2: To show you the house. And, um, you'll, it'll, you'll,
1:29:34
Speaker 2: you'll say like, you know, it'll say like, you know,
1:29:36
Speaker 2: can you walk me through your home and it'll follow
1:29:38
Speaker 2: you around and you will tell it all that. Like
1:29:41
Speaker 2: you would, um, let's see, let's, let's see it.
1:29:43
Speaker 1: You'll see it.
1:29:43
Speaker 2: A friend saying for two weeks at your house that,
1:29:46
Speaker 2: you know, needed to like cook and use your stuff. Like,
1:29:49
Speaker 2: you know, like, you know, you wanted to wash clothes
1:29:51
Speaker 2: and stay in one of your rooms. Like you'd walk
1:29:52
Speaker 2: that person around and you'd be like, Hey man, this is, um,
1:29:55
Speaker 2: this is recycling here. This is where trash is at. Uh,
1:29:58
Speaker 2: like here's, here's where you get water. Like, The trash
1:30:01
Speaker 2: goes out every Monday. We do blankets on the couch,
1:30:07
Speaker 2: but we want them in the cabinet when they're done.
1:30:10
Speaker 2: You know what I mean? Or we want these folded
1:30:12
Speaker 2: and put it over here. All these things you have
1:30:14
Speaker 2: in your home, they're important. Just like you would walking
1:30:20
Speaker 2: a human around for the first time. That's what you'll do.
1:30:24
Speaker 2: And the robot will semantically understand. A, we'll remember all
1:30:29
Speaker 2: of this. And it will learn based on what you want,
1:30:34
Speaker 2: your preferences, like what to go do. Holy shit.
1:30:37
Speaker 1: It'll be that. So it's just like turning into me.
1:30:39
Speaker 2: This is like not, this is not 10 years. We'll
1:30:41
Speaker 2: do this. This is really soon. Like I think in
1:30:46
Speaker 2: the next, like, I mean, I'm hoping this year we
1:30:49
Speaker 2: can like drop a robot in your home and do
1:30:50
Speaker 2: a good amount of stuff. It's just, we'll see. I mean,
1:30:55
Speaker 2: this is like, this is like solving like the Holy
1:30:57
Speaker 2: Grail of robotics. This is like, Solving for a good
1:30:59
Speaker 2: general purpose humanoid robot. Maybe we don't solve it this year.
1:31:03
Speaker 2: Maybe we solve it next year. Maybe we don't solve
1:31:04
Speaker 2: it next year, but it's 2020. I don't know. We're close.
1:31:07
Speaker 2: We feel like we're in the red zone with like,
1:31:10
Speaker 2: we feel like we know the architecture. We have the hardware.
1:31:12
Speaker 2: We know how to get the data. We put the
1:31:13
Speaker 2: data in. The robot does it. We need to like
1:31:15
Speaker 2: now learn how to generalize. We need to move deeper
1:31:19
Speaker 2: into pre-training. We know the directions we need to go ahead.
1:31:23
Speaker 2: We think to solve this. And we're seeing a lot
1:31:25
Speaker 2: of both positive transfer and... a lot of just like,
1:31:30
Speaker 2: we're seeing internally, we think the right direction to make
1:31:33
Speaker 2: this work.
1:31:34
Speaker 1: When you were talking about trusting the robot with your kids,
1:31:38
Speaker 1: what are, I'm just curious, what are your concerns?
1:31:42
Speaker 2: Yeah, at Archer, I always.
1:31:43
Speaker 1: I haven't thought about this.
1:31:45
Speaker 2: I think at Archer, it was always like, I'll never
1:31:48
Speaker 2: feel safe. I never feel comfortable recommending people to fly
1:31:52
Speaker 2: in Archer and letting people fly in Archer until I
1:31:55
Speaker 2: would fly an Archer aircraft with my kids. That's the
1:31:59
Speaker 2: level of safety we need to get to. It's like
1:32:00
Speaker 2: a really high bar. That's what you want though, right?
1:32:03
Speaker 2: To take a aircraft like that around. So I think
1:32:07
Speaker 2: the same thing for figure here is we'll be safe when,
1:32:12
Speaker 2: or to me, it will be safe when I feel
1:32:14
Speaker 2: comfortable putting the robot around my kids. I have a
1:32:17
Speaker 2: one-year-old and four-year-old and seven, you know what I mean?
1:32:20
Speaker 2: I have young kids. They like want to jump on everything. And,
1:32:23
Speaker 2: you know, it's like, they're like, yeah, and the robot, like,
1:32:26
Speaker 2: you know, The robot needs to be extremely safe there.
1:32:28
Speaker 2: So that's another hurdle. It's like getting to general, like
1:32:32
Speaker 2: solving general purposeness, getting safety to work, and then making
1:32:35
Speaker 2: enough of them. Those are kind of like the equations
1:32:36
Speaker 2: from here. Listen, we have a good plan on what
1:32:39
Speaker 2: to go do here, but now it's like execution that
1:32:42
Speaker 2: we gotta go do to show it works.
1:32:44
Speaker 1: Right on, right on. You wanna take a walk around
1:32:46
Speaker 1: this thing?
1:32:47
Speaker 2: Yeah, let's do it. Perfect.
1:32:51
Speaker 1: Most people blame stress, sleep, or just getting older when
1:32:55
Speaker 1: the energy starts to fade. The brain fog, the slower recovery,
1:33:00
Speaker 1: the feeling of running on empty by midday. But underneath
1:33:04
Speaker 1: all of it, there's often a cellular reason. Your body
1:33:08
Speaker 1: is running low on NAD, the molecule that powers energy
1:33:12
Speaker 1: production at its core. That's why I've been taking Rho
1:33:15
Speaker 1: Nutrition's Liposomal NAD Plus. And the difference is real. Clean,
1:33:21
Speaker 1: sustained energy, sharper focus, in better recovery without relying on
1:33:26
Speaker 1: caffeine or stimulants. NAD supports mitochondrial function, the powerhouses of
1:33:32
Speaker 1: your cells that keep you energized and your mind sharp.
1:33:36
Speaker 1: Until recently, replenishing NAD meant expensive clinic visits, but Rho
1:33:41
Speaker 1: offers a simple at-home option, just one teaspoon a day.
1:33:46
Speaker 1: Right now, you can try it risk-free with a 60-day
1:33:49
Speaker 1: money-back guarantee and Get 20% off with code SRS at Ronutrition.com
1:33:56
Speaker 1: slash discount slash SRS. Again, that's SRS for 20% off
1:34:03
Speaker 1: at Ronutrition.com slash discount slash SRS. Want more from the
1:34:11
Speaker 1: Sean Ryan Show? Join our Patreon today for more clips
1:34:15
Speaker 1: and exclusive content. You'll get an exclusive look behind the
1:34:18
Speaker 1: scenes where you can watch the guests interact with the
1:34:20
Speaker 1: team and explore the studio before every episode. Plus, unlock
1:34:25
Speaker 1: bonus content like our extra intel segments where we ask
1:34:29
Speaker 1: our guests additional questions. Our new SRS on-site specials and
1:34:34
Speaker 1: access to an entire tactical training library you will not
1:34:38
Speaker 1: find anywhere else. And the best part? Patreon members can
1:34:42
Speaker 1: ask our guests questions directly. Your insights can help shape
1:34:45
Speaker 1: the show. Join us on Patreon now, support the mission,
1:34:49
Speaker 1: and become part of the Sean Ryan Show's story.
1:35:11
Speaker 2: All right, this is our figure three humanoid robot. We
1:35:18
Speaker 2: actually unveiled it last year. My God.
1:35:22
Speaker 1: Yeah.
1:35:23
Speaker 2: It's about 130 pounds, five foot six. And we basically
1:35:28
Speaker 2: designed it to do most things, like a lot of
1:35:30
Speaker 2: things humans do.
1:35:32
Speaker 1: 130 pounds.
1:35:33
Speaker 2: 135 pounds, yeah. It's fold laundry, do dishes, do manufacturing logistics.
1:35:40
Speaker 2: I think a few things here that we made improvements on.
1:35:42
Speaker 2: This is our third time basically running through three generations
1:35:44
Speaker 2: of robots. We reduced the weight and mass. We made
1:35:49
Speaker 2: the robot skinnier, but also same strength and speeds. We
1:35:52
Speaker 2: upgraded the sensors on the robot. It basically sees through cameras.
1:35:56
Speaker 2: We have better, basically our fifth generation hands on board
1:35:59
Speaker 2: that have a camera, tactile sensors, basically improved grip. We
1:36:05
Speaker 2: also have on the robot basically more compute. on board
1:36:09
Speaker 2: for running our Helix neural network. We also spent a
1:36:12
Speaker 2: lot of time on just basically making the robot more safe.
1:36:15
Speaker 2: So they all have kind of the squishy layer of
1:36:17
Speaker 2: foam on it. So, yep, go ahead.
1:36:20
Speaker 1: So if you, let's say somebody pushed it over and
1:36:24
Speaker 1: fell over, I mean, what's the durability of these?
1:36:26
Speaker 2: I mean, it depends how hard you push it, but
1:36:28
Speaker 2: like for the most part, we fall, robot can get
1:36:31
Speaker 2: back up, just continue to do work. It depends how
1:36:34
Speaker 2: you fall. Sometimes we break necks. Sometimes it's fine. All right,
1:36:40
Speaker 2: turn around. Another thing too is like we basically, the
1:36:46
Speaker 2: robot's almost fully soft wrapped. One thing we can do
1:36:49
Speaker 2: here is we basically can make clothes for the robot,
1:36:51
Speaker 2: which we do for both our customers and internally. The
1:36:54
Speaker 2: clothes can be put on by like any person. So
1:36:56
Speaker 2: we can basically unzip it, take clothes off, put clothes
1:37:00
Speaker 2: back on. We don't need tools to do so. Can
1:37:03
Speaker 2: we see? Can we see what's in there? Yeah, basically
1:37:05
Speaker 2: it's the torso. Oh, you can't see any of the internals? No,
1:37:11
Speaker 2: they're all inside the structure. So inside of here we
1:37:13
Speaker 2: have basically a battery, GPUs, computer, power distribution. Basically the
1:37:21
Speaker 2: brains and all the energy are in the torso.
1:37:24
Speaker 1: Wow.
1:37:26
Speaker 2: And then basically the robot is left with basically 40 joints.
1:37:32
Speaker 2: So all basically electric motors. and the motors have like
1:37:36
Speaker 2: basically tons of sensors on it for balancing and doing work.
1:37:39
Speaker 2: All right, turn around. All right, we can walk with
1:37:45
Speaker 2: it for a minute.
1:37:46
Speaker 1: All right, let's do it.
1:37:50
Speaker 2: All this walking and all the robot movements are all
1:37:53
Speaker 2: done again through a neural net. There's no code helping
1:37:55
Speaker 2: us do this. Holy shit, this is awesome. You want one?
1:38:02
Speaker 1: What's that?
1:38:02
Speaker 2: You want one of these?
1:38:03
Speaker 1: I want a couple of them.
1:38:04
Speaker 2: You want a couple of them?
1:38:05
Speaker 1: Okay, great. Dude, whoa. Yeah. Let's go back this way.
1:38:12
Speaker 2: Let's turn around.
1:38:16
Speaker 1: Can it run?
1:38:17
Speaker 2: Let's see how fast it can go. We have running,
1:38:19
Speaker 2: we have, I don't know if we're on the running mode,
1:38:21
Speaker 2: but let's go as fast as we can. We do
1:38:24
Speaker 2: jog with the robots outside.
1:38:26
Speaker 1: Really?
1:38:26
Speaker 2: At campus, yeah. I think this also looks cool, right?
1:38:42
Speaker 2: Got high tops on.
1:38:44
Speaker 1: Looks awesome.
1:38:45
Speaker 3: Yeah.
1:38:45
Speaker 1: And you said there's cameras in the hands too?
1:38:47
Speaker 2: Yeah, cameras in the palms. Right here in the palm.
1:38:50
Speaker 2: So we can see the fingertips when it's grabbing objects.
1:38:52
Speaker 2: And then every single fingertip has a tactile sensor inside.
1:38:55
Speaker 2: So we can basically touch forces as we're grabbing objects.
1:38:59
Speaker 1: Can it shake my hand?
1:39:01
Speaker 2: I don't know.
1:39:02
Speaker 1: Maybe. Can it squeeze my hand?
1:39:03
Speaker 2: There you go. There you go.
1:39:07
Speaker 1: Will it crush my hand?
1:39:08
Speaker 2: No, it's not going to crush your hand.
1:39:11
Speaker 1: Dude, that's pretty sturdy. That's like, I can't move it.
1:39:14
Speaker 2: We can pick up like 40 pound boxes off the
1:39:16
Speaker 2: floor and we can also fold a t-shirt, so.
1:39:21
Speaker 1: That is wild. Yeah. Is this the power button?
1:39:26
Speaker 2: Yeah, don't push that.
1:39:27
Speaker 1: Okay.
1:39:30
Speaker 2: We had somebody in the office our day, it was like,
1:39:33
Speaker 2: I feel like I need to push this. I'm like,
1:39:35
Speaker 2: it's literally going to turn off if you push the button.
1:39:37
Speaker 1: And how long does it hold the charge?
1:39:41
Speaker 2: It depends on what we do, but anywhere from four
1:39:42
Speaker 2: and five hours.
1:39:43
Speaker 1: How long does it take to charge?
1:39:44
Speaker 2: It takes about an hour to charge. So we can
1:39:46
Speaker 2: do about four or five hours on. We can charge
1:39:48
Speaker 2: for an hour. You know, humans take breaks during the
1:39:51
Speaker 2: day to eat and do other stuff. Or we should
1:39:55
Speaker 2: do a lot of time in the office. We'll sub
1:39:57
Speaker 2: another robot in during the meantime.
1:39:58
Speaker 1: Wow. Yeah.
1:40:00
Speaker 2: We actually charge here inductively through the feet. So the
1:40:03
Speaker 2: feet basically have charging pads the robot steps onto, and
1:40:08
Speaker 2: we charge wirelessly. We can basically charge in one hour
1:40:11
Speaker 2: through that whole process.
1:40:12
Speaker 1: Holy shit.
1:40:12
Speaker 2: Just by standing. So in case the robot has a
1:40:14
Speaker 2: task where it needs to stand a lot, we can
1:40:15
Speaker 2: just stand it and charge.
1:40:16
Speaker 1: So it just stands on a mat?
1:40:17
Speaker 2: Changes on a mat. We designed it in-house.
1:40:18
Speaker 1: Like an iPhone charger.
1:40:19
Speaker 2: It's like an iPhone charger. Yeah, and it can charge
1:40:21
Speaker 2: about a kilowatt per foot. It's about two kilowatts it
1:40:23
Speaker 2: can charge. When is this going to be available to
1:40:25
Speaker 2: the consumer market? As soon as we make it work
1:40:27
Speaker 2: really well. And so I can send it to my
1:40:30
Speaker 2: house and my kids will ask for ice cream every
1:40:33
Speaker 2: single day. And yeah, so... We're working really hard. I
1:40:37
Speaker 2: think we've been testing in my home fairly recently and
1:40:43
Speaker 2: we'll be shipping these robots out to commercial customers here
1:40:46
Speaker 2: really shortly.
1:40:47
Speaker 1: Can I ask who the commercial customers are?
1:40:49
Speaker 2: Yeah, we work with BMW. We work with one of
1:40:52
Speaker 2: the largest logistics companies in the world. I work with Brookfield.
1:40:55
Speaker 2: They're one of the largest real estate companies in the world.
1:40:57
Speaker 2: They have a giant portfolio of companies. And then we
1:40:59
Speaker 2: have two more customers we'll be announcing in the next
1:41:02
Speaker 2: 60 days.
1:41:03
Speaker 1: Congratulations. Yeah, thanks. That's amazing.
1:41:05
Speaker 2: So we're going to try to ship as many as possible.
1:41:07
Speaker 2: We can this year.
1:41:08
Speaker 1: Wow.
1:41:08
Speaker 2: We also make these on site next door at Baku.
1:41:11
Speaker 2: It's our production manufacturing facility. And we make about one
1:41:15
Speaker 2: every kind of like 90, like 90 minutes or so now.
1:41:18
Speaker 1: You can make one of these in 90 minutes.
1:41:20
Speaker 2: When we run the line, when lines are running about
1:41:23
Speaker 2: every 90 minutes, we make one. And then that'll greatly
1:41:26
Speaker 2: increase even in the next like several months here.
1:41:29
Speaker 1: Wow.
1:41:30
Speaker 3: Yeah.
1:41:30
Speaker 1: Wow. Yeah.
1:41:31
Speaker 3: What do you, I mean, in full capacity, like, Our
1:41:36
Speaker 3: facility there can do maybe upwards of like 40 to 50,000
1:41:40
Speaker 3: a year at like full capacity.
1:41:42
Speaker 2: But we need to design for much higher, like we
1:41:45
Speaker 2: want to get to like a million units, like a
1:41:47
Speaker 2: million units a year. And, you know, like within this decade.
1:41:50
Speaker 1: A million units a year.
1:41:51
Speaker 2: Yeah, for sure. I mean, you sell like, we sell over.
1:41:53
Speaker 1: It's like building a country.
1:41:55
Speaker 2: It's not, I mean, like you sell over a billion
1:41:57
Speaker 2: phones a year easy. So I think it's going to
1:42:00
Speaker 2: be like a, a robot for every human, so you'll
1:42:02
Speaker 2: need like a cell phone-style manufacturing. Shit. Yeah.
1:42:07
Speaker 1: So I can push this?
1:42:09
Speaker 2: Oh yeah, it has push recovery. Give it a little push.
1:42:12
Speaker 2: I mean, a little harder than that might be nice. Harder? Harder.
1:42:18
Speaker 1: Dude, what?
1:42:20
Speaker 3: Yeah. Yeah.
1:42:27
Speaker 1: This is better balance than I do.
1:42:28
Speaker 2: Yeah, same.
1:42:33
Speaker 1: Dude, that's crazy. Yeah. This is three and a half years.
1:42:39
Speaker 2: We had this walk-in in three years since I started
1:42:41
Speaker 2: the company. It was crazy. Basically, the week of year three,
1:42:46
Speaker 2: we were walking this thing at the office. The thing is,
1:42:50
Speaker 2: this is like, we're going to go through this whole
1:42:52
Speaker 2: iPhone lineup where it's, you know, iPhone 1, iPhone 2,
1:42:56
Speaker 2: iPhone 3. It just gets better and better. And I
1:42:58
Speaker 2: think humans will take more radical steps between those. Every year,
1:43:02
Speaker 2: we're roughly building a new robot every year. We'll just
1:43:04
Speaker 2: get dramatically better than this.
1:43:06
Speaker 1: Damn.
1:43:07
Speaker 3: Yeah.
1:43:07
Speaker 2: Our step up from here even to the future robots
1:43:10
Speaker 2: will be, I think, perhaps the most dramatic step up
1:43:13
Speaker 2: we ever make.
1:43:15
Speaker 1: Wild. You want to take some pictures with them?
1:43:18
Speaker 2: Let's do it.
1:43:18
Speaker 1: Dude. That is insane. What do you think? Awesome. I
1:43:28
Speaker 1: want one.
1:43:29
Speaker 2: You want one? Yeah. Let's get you one, man.
1:43:31
Speaker 1: Wow. You said the hands can sense three grams of pressure?
1:43:40
Speaker 2: Yeah, we basically have tactile sensors on every fingertip, and
1:43:45
Speaker 2: they're really sensitive. And we have a camera in the
1:43:50
Speaker 2: hand that can detect when the fingertips are in contact
1:43:52
Speaker 2: with some surface. It could be like something we're touching.
1:43:54
Speaker 2: And then within there, every joint can kind of also
1:43:58
Speaker 2: feel sense and track the position of every like, you know,
1:44:01
Speaker 2: part of the hand. So the hand's like, the hand's
1:44:04
Speaker 2: really good. Honestly, we're working on hand stuff for like
1:44:08
Speaker 2: close to four years. It's probably one of the hardest
1:44:11
Speaker 2: engineering problems we have on the hardware side. It's probably
1:44:15
Speaker 2: as hard. And we have our next generation hand that
1:44:17
Speaker 2: we kind of teased a couple weeks ago that has
1:44:19
Speaker 2: like basically full, I think it gets a full human
1:44:24
Speaker 2: level dexterity with this hand.
1:44:27
Speaker 1: Are you serious?
1:44:28
Speaker 2: It's got as many joints on the hand as a
1:44:30
Speaker 2: human hand has. There's still a lot of work to
1:44:33
Speaker 2: go do, but like it's now... It's now a huge
1:44:38
Speaker 2: step up where we actually even currently are. And the
1:44:40
Speaker 2: hand now can like fold laundry and, you know.
1:44:43
Speaker 1: Do you think it'll hit a point where it can
1:44:46
Speaker 1: outperform a human? More dexterity in a hand than a human?
1:44:51
Speaker 1: Better balance, faster, stronger.
1:44:53
Speaker 2: We already have better balance than a human. The robot
1:44:56
Speaker 2: on one leg could balance better than a human can.
1:44:58
Speaker 2: I don't know about like... Humans have a lot of
1:45:01
Speaker 2: degrees of freedom. We have like hundreds, a few hundred
1:45:04
Speaker 2: degrees of freedom. Our hands are very dexterous. I would
1:45:07
Speaker 2: say if we can do close to human dexterity in
1:45:10
Speaker 2: terms of like... That'd be a huge win. You'd have
1:45:14
Speaker 2: robots everywhere.
1:45:15
Speaker 3: And.
1:45:18
Speaker 2: You know, and then we're going to still have a
1:45:20
Speaker 2: lot of trouble getting to a full human range of motion.
1:45:23
Speaker 2: Like small things, like you would reach inside of a,
1:45:27
Speaker 2: you know, a washer and you kind of like move
1:45:28
Speaker 2: your head as you're like getting in. Or sometimes like
1:45:31
Speaker 2: some people get on like, you know, get down to
1:45:33
Speaker 2: the ground and like kind of get in the washer
1:45:34
Speaker 2: to grab someone on the back. We do a lot
1:45:36
Speaker 2: of crazy stuff.
1:45:37
Speaker 1: Yeah, that is.
1:45:38
Speaker 2: And, you know, so it's like, like even like a
1:45:41
Speaker 2: 12 year old can kind of do like most things
1:45:43
Speaker 2: in a house. You know what I mean? Like they
1:45:45
Speaker 2: can jump up on countertops and all kinds of crazy stuff. Um,
1:45:49
Speaker 2: humans would be tough, but like, I think we can
1:45:51
Speaker 2: get like very soon we'll get to like pretty close
1:45:55
Speaker 2: to most of what humans can.
1:45:58
Speaker 1: You had a pretty close relationship with OpenAI, correct?
1:46:03
Speaker 2: Yeah, they, um, they led my, so, uh, Sam and
1:46:08
Speaker 2: OpenAI led my series B, uh, co-led my series B
1:46:10
Speaker 2: with Microsoft. Uh, that was a few years ago now.
1:46:14
Speaker 2: So we raised about a little under 700 million, uh,
1:46:17
Speaker 2: in our series B, our second round of funding. And
1:46:20
Speaker 2: they joined my board and then we ended up spending
1:46:21
Speaker 2: basically a year with them working on, well, I mean,
1:46:25
Speaker 2: I'll give you the background. The goal was to like
1:46:29
Speaker 2: try to advance AI models for humanoid robots together. And,
1:46:35
Speaker 2: you know, they have some great folks that have worked
1:46:37
Speaker 2: on like LLMs and chatbots and things. And at the
1:46:42
Speaker 2: time we had like a, we, you know, we still do,
1:46:43
Speaker 2: but we had our, we had a full like AI
1:46:45
Speaker 2: team internally. So we were basically working weekly, daily on like,
1:46:49
Speaker 2: basically how do we advance a state-of-the-art kind of like
1:46:52
Speaker 2: language models for robotics. And, you know, like, yeah, I
1:47:02
Speaker 2: ended up firing them.
1:47:04
Speaker 1: I know.
1:47:05
Speaker 2: A year, a year later, but in splitting ways, but listen,
1:47:09
Speaker 2: they're a great team. I, I, the senior leadership and
1:47:12
Speaker 2: everybody there, Sam included, were great to interact with. The
1:47:16
Speaker 2: issue lied for us. Nobody's ever put advanced language models
1:47:23
Speaker 2: into these systems. We have to produce action output on
1:47:27
Speaker 2: the robot, and it's a very different thing than next
1:47:30
Speaker 2: token prediction for language models. We ended up finding that
1:47:35
Speaker 2: the team we had in place, my team lead, the
1:47:38
Speaker 2: folks we have here all from Google DeepMind or certain
1:47:40
Speaker 2: areas of top AI programs, and they're really good. The
1:47:45
Speaker 2: team now we have is over 50 or so on
1:47:47
Speaker 2: our Helix team internally. We just found that that team
1:47:50
Speaker 2: we had internally, we just found circles around them every day.
1:47:56
Speaker 2: We had a hard time getting... In robotics, you gotta
1:47:58
Speaker 2: run the robot, see how it does. You want to
1:48:02
Speaker 2: run a new AI experiment or do some ablations and
1:48:05
Speaker 2: some evals, you need to run the robot at the
1:48:07
Speaker 2: end of the day and see how it does. Sim
1:48:09
Speaker 2: is one thing, you can get certain far running simulations
1:48:11
Speaker 2: and looking at loss curves and stuff, but we need
1:48:13
Speaker 2: at the end of the day, like, do we need
1:48:15
Speaker 2: to get, like, see how the robot does? And we had,
1:48:17
Speaker 2: we just had a hard time getting them in the office.
1:48:18
Speaker 2: We had a hard time, like basically like, uh, like basically,
1:48:21
Speaker 2: you know, advancing stuff together as a team, uh, ended up,
1:48:25
Speaker 2: we like the strategy we had internally and the team
1:48:28
Speaker 2: we had was just like complete superstars. They're the best
1:48:32
Speaker 2: robot learning folks on the planet. They're sort of figure. And, um,
1:48:35
Speaker 2: it got to a point where, uh, And I got
1:48:38
Speaker 2: a call one day, just like, you know, we were
1:48:40
Speaker 2: like also week to week, like showing them how we
1:48:42
Speaker 2: were doing all this work. And I got a call
1:48:45
Speaker 2: one day saying like, hey, we're like, you know, we've
1:48:48
Speaker 2: been watching your progress.
1:48:49
Speaker 1: It's unbelievable.
1:48:51
Speaker 2: And, you know, we're thinking about doing robotics work internally.
1:48:56
Speaker 2: And I was just like, ah, this is over. Yeah,
1:48:58
Speaker 2: just get out of here. Like, this is like, we're
1:49:00
Speaker 2: like teaching you how to do like robot learning. You're
1:49:03
Speaker 2: seeing our progress. We had like a couple of, you know,
1:49:06
Speaker 2: Sam and a couple of the co-founders on site at
1:49:08
Speaker 2: one point, right before this, and they saw it and
1:49:09
Speaker 2: they were like, wow, this is like, it was doing
1:49:11
Speaker 2: like this neural network on table. And they were just like, Jesus,
1:49:14
Speaker 2: this is amazing. And I was like, you know, they
1:49:18
Speaker 2: were still at a point where they continue to want
1:49:19
Speaker 2: to work together after this. And I was like, there's
1:49:21
Speaker 2: no way we're going to teach you how to do
1:49:22
Speaker 2: this stuff anymore. And also we just like got no
1:49:27
Speaker 2: value out of the whole relationship, very little. I mean, listen,
1:49:29
Speaker 2: it was helpful having them lead the route, lead the round.
1:49:32
Speaker 2: it was like, there was some, there was like some
1:49:34
Speaker 2: good brand association there, but like beyond like that, there
1:49:36
Speaker 2: wasn't much. So we ended up, you know, we're going
1:49:39
Speaker 2: to chart our own territory. We're going to do AI
1:49:40
Speaker 2: ourselves here. It was also just became like a, to
1:49:43
Speaker 2: be frank, like it became like really hard to recruit.
1:49:45
Speaker 2: We were like, you know, I have to spend a
1:49:46
Speaker 2: lot of my time hiring like, like on the AI team.
1:49:49
Speaker 2: And we bring candidates in and they'd be like, oh,
1:49:52
Speaker 2: you guys do the robot and open AI does the models.
1:49:54
Speaker 2: And I'm like, oh no, not really. No, we have
1:49:56
Speaker 2: a whole AI team internally. We do model development here ourselves.
1:50:01
Speaker 2: You know, Like we're advancing ourselves and it just wasn't
1:50:04
Speaker 2: the perception from the outside. It was just hard. So
1:50:07
Speaker 2: that also wasn't helpful for us. Both hiring was not great.
1:50:10
Speaker 2: And we were like, you know, there was like an
1:50:12
Speaker 2: information passing back that I think wasn't really helpful for
1:50:15
Speaker 2: us long-term if we're going to be competitors. So we
1:50:18
Speaker 2: decided to split ways. I decided specifically to split ways.
1:50:22
Speaker 2: But they have a great team. I think they're doing
1:50:24
Speaker 2: robotics now internally.
1:50:26
Speaker 1: Sounds like it.
1:50:27
Speaker 2: Yeah, exactly.
1:50:28
Speaker 1: Yeah.
1:50:29
Speaker 2: Yeah, exactly. Uh, I was like, I got a call
1:50:31
Speaker 2: saying like, yeah, like, you know, partly like, uh, the
1:50:33
Speaker 2: part of the feedback I heard was like, we've made
1:50:35
Speaker 2: so much progress at figure and they seen that, that
1:50:38
Speaker 2: they were, you know, they opening, I started out as
1:50:40
Speaker 2: a robotics program. They were trying to solve AGI through robotics.
1:50:44
Speaker 2: First three, four years, they were just like all in
1:50:46
Speaker 2: on robots. If you Google like open AI robotics, it's
1:50:49
Speaker 2: like old 2016, 2017, 2018, 2019, like, you know, maybe
1:50:50
Speaker 2: like Maybe 2019, 2020, something like that, they end up
1:50:56
Speaker 2: pivoting into large language models, maybe 2021, something like this.
1:51:00
Speaker 2: But they're in robotics from, I think, 2016, 2017 for
1:51:03
Speaker 2: many years, maybe three or four years, trying to solve
1:51:07
Speaker 2: AGI through robotics. There's this other, we don't need to
1:51:12
Speaker 2: get into it, but it's unclear if you need an
1:51:14
Speaker 2: embodiment or not. At the time, it was unclear whether
1:51:18
Speaker 2: you need an embodiment or not to truly get to
1:51:20
Speaker 2: above peak human intelligence. Um, and, um, and they, they
1:51:25
Speaker 2: had a hard time in there, but there was like,
1:51:26
Speaker 2: part of their thesis was like, get back into robotics
1:51:28
Speaker 2: at some point. And I think we just, we accelerated
1:51:30
Speaker 2: that here at figure. Um, and you know, I think
1:51:33
Speaker 2: to be fair, like to be home, like somewhat humbled
1:51:35
Speaker 2: is like, it's, we made like, I think we made
1:51:38
Speaker 2: like 10, I don't know, five to 10 years of
1:51:41
Speaker 2: progress in like three years, four years. Like we just like,
1:51:45
Speaker 2: it just felt like, uh, this should have taken 10.
1:51:48
Speaker 2: Even right now it feels like we're not even four
1:51:50
Speaker 2: years old yet, four years old and into May or
1:51:53
Speaker 2: something like that. I couldn't believe when we started the
1:51:57
Speaker 2: company three and a half years ago, we'd be at
1:51:58
Speaker 2: a point where you can get a humanoid robot even
1:52:00
Speaker 2: here doing the stuff it's doing here, but let alone
1:52:02
Speaker 2: the real stuff it's doing now in 24-7 commercial work
1:52:06
Speaker 2: in the home. It's neural net driven. We can make
1:52:09
Speaker 2: them every 90 minutes when our lines are up. It's
1:52:12
Speaker 2: just like, it's crazy. So yeah, we decided to part ways.
1:52:17
Speaker 1: Man, I mean, I don't think there's too many people
1:52:21
Speaker 1: in the world that can say they fired the biggest
1:52:23
Speaker 1: AI company on earth. I mean, that's a ballsy move,
1:52:28
Speaker 1: but it makes perfect sense. And, man, again, just congratulations
1:52:34
Speaker 1: on everything. I mean, that is, that's just crazy. You know,
1:52:39
Speaker 1: I've done a, it's just very surreal for me to
1:52:45
Speaker 1: unveil some of that. I mean, I know we didn't
1:52:46
Speaker 1: unveil this, but it, It's the first podcast that's ever
1:52:49
Speaker 1: been on.
1:52:50
Speaker 2: Dude, Sean, I have not taken a robot to a podcast.
1:52:53
Speaker 2: I get asked every week to do this. This is
1:52:55
Speaker 2: the first time. And I love your show and want
1:52:59
Speaker 2: to get them here in Tennessee. It's the first time
1:53:01
Speaker 2: a bot's been out here to something like this.
1:53:03
Speaker 1: Thank you. It's really cool to be able to do this.
1:53:09
Speaker 1: Like once-in-a-lifetime opportunity type stuff. Thank you.
1:53:13
Speaker 2: No problem.
1:53:14
Speaker 1: What about military applications?
1:53:18
Speaker 2: Yeah, we've decided not to do military stuff today. Not
1:53:23
Speaker 2: to say the robots won't be good in military or helpful.
1:53:28
Speaker 2: My belief right now is it's just too difficult to
1:53:33
Speaker 2: do both. Ship into the home, ship to top Fortune
1:53:37
Speaker 2: 100 companies in the US, and then also put militarized robots.
1:53:42
Speaker 2: I think it's just too hard in one umbrella. Um,
1:53:44
Speaker 2: I think there's a huge opportunity like to save lives
1:53:47
Speaker 2: and help on the military side. Um, but I think
1:53:50
Speaker 2: it becomes like, you know, we, we do have, you know,
1:53:54
Speaker 2: we do have like a very advanced system here. The
1:53:57
Speaker 2: system can, you know, unlike a car, if a car
1:54:00
Speaker 2: became sentient, like, you know, you can like walk in
1:54:04
Speaker 2: your house, walk upstairs, go in your room. It's like
1:54:07
Speaker 2: not going to come chase you. Like a robot will
1:54:10
Speaker 2: just walk right up your stairs and open your door.
1:54:12
Speaker 2: The humanoid robot. you know, this is a very different technology.
1:54:15
Speaker 2: We've got to be very careful with it. So I
1:54:17
Speaker 2: think because of some of that and some other things,
1:54:20
Speaker 2: we like, we know we've drawn a line here to
1:54:22
Speaker 2: say like, you know, we want to stick with the,
1:54:25
Speaker 2: you know, consumer market, commercial market and go to harden
1:54:28
Speaker 2: the paint with that. I think there are and will
1:54:31
Speaker 2: be like incredible opportunities for companies like to go into
1:54:34
Speaker 2: the military. To be frank, these robots would be great there.
1:54:38
Speaker 2: Like they can just like, they can, you know, like
1:54:40
Speaker 2: some of the most dangerous missions are like, you know,
1:54:43
Speaker 2: going to close quarters and houses and, you know, that
1:54:48
Speaker 2: stuff is like extremely dangerous. Humans would be great at
1:54:50
Speaker 2: that stuff. Like opening doors and just making sure the
1:54:53
Speaker 2: house is, you know, cleared. Like clear a house. You
1:54:56
Speaker 2: know what I mean?
1:54:57
Speaker 1: I could see it for a whole ton of stuff.
1:54:59
Speaker 1: Not even just going on target, but sentries, gate guards.
1:55:04
Speaker 2: Yeah.
1:55:05
Speaker 1: I mean, roving patrols.
1:55:07
Speaker 2: Totally.
1:55:07
Speaker 1: I mean, all of it.
1:55:08
Speaker 2: Totally.
1:55:09
Speaker 1: Just armed security. It's... It's, wow.
1:55:15
Speaker 2: You kind of have somewhat of a treatable asset too.
1:55:17
Speaker 2: You can basically, I think you can make them relatively cheap,
1:55:19
Speaker 2: make a lot of them, just put them out to work.
1:55:22
Speaker 1: Do you think you'll get into it in the future?
1:55:23
Speaker 2: I don't know. As of now, no. But like, there's
1:55:30
Speaker 2: a part of it that you'll, there's a part of
1:55:32
Speaker 2: the story here where you're like, you could make this
1:55:34
Speaker 2: like obviously really safe for humans there. There's a whole
1:55:38
Speaker 2: part of the story where it's like, I think it
1:55:39
Speaker 2: just becomes, you know, to be frank, like the, When
1:55:43
Speaker 2: we sell to commercial customers, even homes, it's not like
1:55:46
Speaker 2: selling a robot arm on a stand. It's like these
1:55:49
Speaker 2: commercial customers need CEO approval. We can't get them through
1:55:53
Speaker 2: without the CEO of these major companies coming to see
1:55:55
Speaker 2: the robots and saying, we're going to announce this relationship
1:55:58
Speaker 2: with figure and we're going to announce humanoid robots in
1:56:01
Speaker 2: our facilities. And it's just like a, it's a very,
1:56:04
Speaker 2: you know, it's like a, there's, I'll.
1:56:08
Speaker 1: Make that announcement.
1:56:10
Speaker 2: I think it's fucking awesome. It's awesome. And then that
1:56:14
Speaker 2: makes it that much harder than if we have a
1:56:16
Speaker 2: new military side of things.
1:56:18
Speaker 1: Why do you think they're hesitant? Is it replacement of
1:56:21
Speaker 1: human jobs? I mean, Jack Dorsey just let go of what?
1:56:26
Speaker 2: 10,000 people.
1:56:27
Speaker 1: Yeah, almost half of his personnel because of AI. And
1:56:32
Speaker 1: it's stocked one up because.
1:56:33
Speaker 2: I think it's probably because the robot is human-like and
1:56:36
Speaker 2: can do human-like work. So I think it's just scary for...
1:56:38
Speaker 2: It's a scary thing that it can do what humans can.
1:56:41
Speaker 2: I think you have similar scariness folks have around digital
1:56:45
Speaker 2: AI and how that will basically manifest in the future.
1:56:49
Speaker 2: So I think it's a real thing. I think the
1:56:52
Speaker 2: robots can do human-like work. And it will continue every
1:56:55
Speaker 2: year to do more and more human-like work. But like that,
1:56:59
Speaker 2: we just got to We just want to be very
1:57:01
Speaker 2: careful about how we position this and what we do
1:57:04
Speaker 2: and also how we communicate it.
1:57:05
Speaker 1: Yeah.
1:57:06
Speaker 2: Yeah. What's next for the robots? We want to solve
1:57:11
Speaker 2: general robotics, I figure. We want to... We think of
1:57:15
Speaker 2: ourselves truly as like at the frontier of like this
1:57:19
Speaker 2: robotics AI lab that needs to build common sense reasoning
1:57:24
Speaker 2: in a robot that can put in every home. How
1:57:27
Speaker 2: do we... drop it into your home it's never been,
1:57:30
Speaker 2: and you can just communicate with it and get to
1:57:32
Speaker 2: start doing work. That's the problem we wanna solve here.
1:57:35
Speaker 2: That's the problem if you solve it, you can ship
1:57:37
Speaker 2: billions and millions of robots. There's also a business where
1:57:40
Speaker 2: if you don't wanna solve that, you can definitely ship robots.
1:57:43
Speaker 2: You can ship them in the commercial workforce, you can
1:57:45
Speaker 2: ship them in the military, as you mentioned. There is
1:57:48
Speaker 2: a path to go build a business doing that. But
1:57:51
Speaker 2: the biggest business in the world is if you solve
1:57:54
Speaker 2: general purpose robotics, where just through speech, and talking to
1:57:59
Speaker 2: the robot, it'd be, it'd feel like you had like
1:58:01
Speaker 2: a human in a body suit that can like understand you, nod,
1:58:05
Speaker 2: like go off and do things now after your task.
1:58:08
Speaker 2: That's the problem we want to solve at Figure. That's
1:58:10
Speaker 2: a large-scale, it's like an AI lab problem at this point.
1:58:15
Speaker 2: We talk a lot about how we're trying to give
1:58:19
Speaker 2: AI a body here at Figure. And so we have
1:58:24
Speaker 2: this embodiment. We need to put really sophisticated AI into
1:58:26
Speaker 2: it to be able to command it. And that's the
1:58:30
Speaker 2: biggest problem we're trying to solve. If you're with me
1:58:33
Speaker 2: in the office every day, I am working that down
1:58:38
Speaker 2: with no sleep basically as hard as I possibly can.
1:58:41
Speaker 2: And it's a very, very difficult problem. At this point,
1:58:45
Speaker 2: it's largely constrained by getting the appropriate data into the network.
1:58:53
Speaker 2: I think if we could snap our fingers and get
1:58:56
Speaker 2: a pile of data that we really needed, into Helix stack,
1:59:00
Speaker 2: I think we would solve general robotics right now.
1:59:03
Speaker 1: Wow. What should I be asking you that I haven't
1:59:06
Speaker 1: asked yet?
1:59:07
Speaker 2: About figure or general?
1:59:08
Speaker 1: About figure.
1:59:17
Speaker 2: I mean, there's a lot of stuff going on with
1:59:18
Speaker 2: China and manufacturing and a few other things, but I
1:59:21
Speaker 2: think maybe to summarize, I think where we're at is
1:59:28
Speaker 2: I think if I had to like, if I was
1:59:30
Speaker 2: like watching this and I wasn't following the story, I
1:59:35
Speaker 2: think the one thing I would like to convey that,
1:59:37
Speaker 2: you know, is like, we are so close to like
1:59:41
Speaker 2: making this happen now. And it's only until, you know,
1:59:45
Speaker 2: people can come online and like watch our stuff we
1:59:46
Speaker 2: put out, you know, but when people come to the
1:59:50
Speaker 2: office and experience it and see the robots and you
1:59:52
Speaker 2: can talk to them and some of the stuff you're
1:59:53
Speaker 2: doing here today, it's just like a full, like, emotional
1:59:57
Speaker 2: experience that is really hard to convey.
1:59:59
Speaker 1: It's just crazy to just... It feels like we're living
2:00:05
Speaker 1: in the future.
2:00:06
Speaker 2: It just feels like we're living here. Yeah. It's crazy
2:00:09
Speaker 2: it works. It's crazy it's working. But we're now in the...
2:00:14
Speaker 2: We now have line of sight to make this happen,
2:00:18
Speaker 2: which is exciting in my perspective. Super exciting. I think
2:00:21
Speaker 2: it's going to be super transformative for the world. And
2:00:24
Speaker 2: I think what we're going to try to do over
2:00:26
Speaker 2: the next year or two is try to get this
2:00:27
Speaker 2: out further at scale and get everybody to feel this
2:00:30
Speaker 2: more and more. You feel it when you come to
2:00:32
Speaker 2: our office and you feel it when you're next to
2:00:33
Speaker 2: the robots, but it's hard for the... We're such early
2:00:37
Speaker 2: innings about this yet for takeoff that it's hard for
2:00:40
Speaker 2: the whole world to really feel this.
2:00:42
Speaker 1: Yeah. Have you seen... Do the robots interact with each other?
2:00:47
Speaker 2: Yeah.
2:00:47
Speaker 1: What does that look like?
2:00:49
Speaker 2: Right now, they communicate with each other. So we have
2:00:53
Speaker 2: robots that are running these 24-7 shifts. When one robot
2:00:56
Speaker 2: gets down to low state of charge, let's say it's
2:00:58
Speaker 2: like 10%, and it's a few percentages away from... We'll
2:01:02
Speaker 2: dock it before it's at 1% or something like that.
2:01:05
Speaker 2: It's at 10%. The other robot will get ready to
2:01:08
Speaker 2: sub in. It will come walk over, sit right behind it.
2:01:12
Speaker 2: And then when the robot is ready and knows that
2:01:14
Speaker 2: it's there, it will then back away, and then the
2:01:16
Speaker 2: robot will go in to do operations and do work.
2:01:18
Speaker 2: That other robot will then go over and start charging.
2:01:22
Speaker 2: If any of those robots have any problems throughout, it
2:01:24
Speaker 2: could be hardware or software, they will go and go
2:01:27
Speaker 2: to basically the hospital in our office. So they'll go
2:01:30
Speaker 2: to a certain place. When they know they're going to
2:01:33
Speaker 2: the hospital, we have another robot coming in to the
2:01:35
Speaker 2: main docks to start subbing in and getting ready to go.
2:01:39
Speaker 2: All this communication is happening robot to robot. And it's unbelievable.
2:01:44
Speaker 2: And the robots are getting really robust. A year or
2:01:48
Speaker 2: two ago, there would be certain motors that you would
2:01:51
Speaker 2: lose communications with or other types of comms or could
2:01:55
Speaker 2: be hardware failures or software failures, whatever. Let's say it's
2:01:58
Speaker 2: the knee. Lose your knee, you can't stand anymore. You
2:02:01
Speaker 2: know what I mean? You fall. Today, that doesn't happen.
2:02:05
Speaker 2: We can lose a knee. We can lose full comms
2:02:10
Speaker 2: with the knee. We can stiffen the joint and we
2:02:13
Speaker 2: can limp off. Holy shit. Yeah. Actually, I'll post some
2:02:18
Speaker 2: of the next week publicly about this. I've never... It's like,
2:02:22
Speaker 2: holy shit. So we can lose like a lower body
2:02:24
Speaker 2: motor and it literally limps off stage, like off like the,
2:02:28
Speaker 2: you know, the main like line it's on, heading to
2:02:31
Speaker 2: the hospital. It'll limp all the way there. While it's
2:02:34
Speaker 2: limping there, another group from like the healthy part of
2:02:37
Speaker 2: the hospital will then come in and re-sub it in
2:02:39
Speaker 2: on the dock while another one undocks while it just
2:02:41
Speaker 2: lost its knee to go in and do work. all
2:02:44
Speaker 2: that's happening through robot communication levels. You can be like
2:02:48
Speaker 2: literally asleep while this is happening. We run them 24 seven.
2:02:50
Speaker 2: It can be at three in the morning and it
2:02:52
Speaker 2: will happen. It's insane. This is happening like, I saw
2:02:58
Speaker 2: this in the last like few months, this happening out
2:03:00
Speaker 2: right now. This is not even like the future stuff.
2:03:02
Speaker 2: Future stuff is gonna be robots building robots. We're designing robots.
2:03:08
Speaker 2: We will have robots building robots here. And then they
2:03:12
Speaker 2: will go out and they will just do autonomous work
2:03:15
Speaker 2: And they will like charge themselves. They will go do work.
2:03:19
Speaker 2: You'll speak to them sometimes. Sometimes you won't need to
2:03:21
Speaker 2: do and they'll just do work and they'll just be
2:03:22
Speaker 2: like everywhere. I say this again, but I think we'll
2:03:26
Speaker 2: walk out. It'll happen first in probably the Bay Area.
2:03:29
Speaker 2: We're based in the Bay and a lot of companies
2:03:31
Speaker 2: are in there for robotics. But I think you'll go
2:03:33
Speaker 2: to the Bay Area at some point and you'll see
2:03:35
Speaker 2: more humanoids and humans in the next 10 years for sure.
2:03:40
Speaker 1: That is, I can't even imagine what that's going to
2:03:43
Speaker 1: be like.
2:03:44
Speaker 2: It'd be weird.
2:03:45
Speaker 1: Do you think that they will bring, do you think,
2:03:47
Speaker 1: do you think manufacturing will come back to the US?
2:03:51
Speaker 2: Yeah, we're going to bring back.
2:03:52
Speaker 1: Because of this?
2:03:53
Speaker 2: My view is that we don't want to bring back
2:03:54
Speaker 2: manufacturing that's already overseas. We don't want to like, you know,
2:03:59
Speaker 2: like make shoes, make toys, like things like that. I don't,
2:04:02
Speaker 2: I don't think we want, I don't think we have
2:04:03
Speaker 2: the will to do this. I don't think we have
2:04:06
Speaker 2: the know-how to do this as well as like some
2:04:09
Speaker 2: of the Asian manufacturing groups. So I've walked a lot
2:04:13
Speaker 2: of the high volume consumer electronics lines and stuff overseas.
2:04:17
Speaker 2: Some of the most impressive things I've ever seen in
2:04:18
Speaker 2: my life. It's like you walk these lines and they're
2:04:23
Speaker 2: just shipping electronics like crazy and every line they have
2:04:25
Speaker 2: this box of automation inside of it, like a little
2:04:29
Speaker 2: tiny robot inside of there that's moving some whatever, a
2:04:32
Speaker 2: phone enclosure or something like that. And it's doing it
2:04:35
Speaker 2: through an automated way and moving it around, a little conveyor,
2:04:38
Speaker 2: and it's moving to the next station. Maybe a human's
2:04:39
Speaker 2: doing something and it's going down the line, it's going
2:04:41
Speaker 2: to the next station that's got a robotic system in there,
2:04:43
Speaker 2: completely customized in a different way than you just saw.
2:04:45
Speaker 2: And they have lines and lines and floors and floors
2:04:48
Speaker 2: of this, and then buildings and buildings. And you're like,
2:04:50
Speaker 2: holy shit, each one of those boxes is like a
2:04:55
Speaker 2: figure-style complexity. And they have, like, hundreds of them. And
2:05:00
Speaker 2: they need to run them at high rate. It's just like,
2:05:03
Speaker 2: it's unbelievable, actually. It's not trivial. It's very complex. And
2:05:07
Speaker 2: they've been doing it for several decades on these lines.
2:05:11
Speaker 2: So I think one is like, I don't think that
2:05:13
Speaker 2: stuff we want to move back. I think we want
2:05:15
Speaker 2: to move back the high-end robotic stuff that's going to
2:05:19
Speaker 2: be like super transformative for us in the future.
2:05:22
Speaker 1: All the futuristic.
2:05:23
Speaker 2: Yeah. We want to bring back flying cars. I want
2:05:25
Speaker 2: to bring back like, like humanoid robots, like the stuff
2:05:28
Speaker 2: that's like highly dynamic, very intelligent systems, like the next generation,
2:05:32
Speaker 2: like manufacturing 2.0 stuff.
2:05:34
Speaker 1: Gotcha.
2:05:35
Speaker 2: So we're doing that right now in California on our campus.
2:05:38
Speaker 2: We have a fairly large campus in, in the Bay area.
2:05:42
Speaker 2: And we manufacture right now, like whenever 90 minutes or so.
2:05:46
Speaker 2: And we'll continue to spin that up. And then we'll put,
2:05:49
Speaker 2: you know, we'll talk more about it. But we'll put
2:05:53
Speaker 2: more investment here into U.S. manufacturing for the future. Right on.
2:05:58
Speaker 2: So we're going to design humanoids here.
2:06:00
Speaker 1: So these are all manufactured?
2:06:01
Speaker 2: We manufacture those in California.
2:06:03
Speaker 1: Right on, man.
2:06:04
Speaker 2: Yeah, man. They walk off the lines. They walk over.
2:06:07
Speaker 2: It's like 90 days ago, you come, we're making a
2:06:12
Speaker 2: little bit. But now we make like, There's seven robots
2:06:15
Speaker 2: that are all doing end-of-line checkout by themselves for an
2:06:17
Speaker 2: hour and a half. They do their own burn-ins, all
2:06:20
Speaker 2: OEOL checks. They're self-looking at each other, self-calibrating. They're doing
2:06:25
Speaker 2: burpees and other shit to make sure they're okay. If
2:06:28
Speaker 2: they fail, they go into a triage place. We understand
2:06:30
Speaker 2: why to fail. That shouldn't happen. We should always fix that,
2:06:32
Speaker 2: and it should not fail again. How do we fix
2:06:34
Speaker 2: the manufacturing process so the next one doesn't come out
2:06:36
Speaker 2: and ever have that failure? Now we've gotten that process
2:06:39
Speaker 2: really dialed in. We still have issues, but it's fairly
2:06:43
Speaker 2: dialed in. And so the robots come out, do a
2:06:45
Speaker 2: couple-hour check, and when they're done, they just walk over.
2:06:49
Speaker 2: And at some point, we'd love for them to get
2:06:51
Speaker 2: inside their own box, and another one get it ready
2:06:54
Speaker 2: to go and put it on a pallet, and we
2:06:56
Speaker 2: can just start shipping them out.
2:06:57
Speaker 1: It will get in its own box.
2:06:59
Speaker 2: For sure.
2:06:59
Speaker 1: And another one will throw it on the pallet and
2:07:01
Speaker 1: ship it out.
2:07:01
Speaker 2: For sure. Yeah. That's not hard things, though. These are like...
2:07:05
Speaker 2: That's not that hard. You know what I mean?
2:07:08
Speaker 1: Yeah, it's just interesting to think about. I don't know.
2:07:10
Speaker 1: I feel like.
2:07:10
Speaker 2: Comes off.
2:07:11
Speaker 1: The line, gets in its own box... gets loaded on
2:07:14
Speaker 1: by another robot and then shipped off.
2:07:17
Speaker 2: The scary thing for me is like, those are like
2:07:18
Speaker 2: very like rigid body things like cardboard and like moving
2:07:22
Speaker 2: boxes and maybe using machines and stuff. Like those are
2:07:25
Speaker 2: like easy. The scary stuff a couple of years ago
2:07:27
Speaker 2: was like laundry that like literally moves. It's like literally
2:07:31
Speaker 2: never in the same spot. It's like when you touch it,
2:07:33
Speaker 2: it's like actually moving or we do like these like
2:07:35
Speaker 2: packages on this manufacturing conveyor system that like you grab it,
2:07:39
Speaker 2: it's literally moving. It's moving because the conveyor is moving
2:07:41
Speaker 2: down and then the packages are squishing each other. And
2:07:43
Speaker 2: then the package itself is moving because it's plastic when
2:07:45
Speaker 2: you're grabbing it. Those are the hard things that are
2:07:47
Speaker 2: compliant that are like really difficult for robotics. Because they're
2:07:50
Speaker 2: not like stationary when you touch them. So those are
2:07:52
Speaker 2: the things that were like, man, that's going to be
2:07:54
Speaker 2: really tough to fold laundry. And with code, it's been impossible.
2:07:59
Speaker 2: The reason you haven't seen like package logistics and stuff,
2:08:01
Speaker 2: some of this stuff automated is because like these bags
2:08:03
Speaker 2: are just like hard, they're compliant. They're just tough. You
2:08:05
Speaker 2: can't model them. And now we have like, we put
2:08:08
Speaker 2: it all in a neural net. They basically instantly worked.
2:08:11
Speaker 2: When we were working with our, we have a logistics
2:08:13
Speaker 2: customer we were working with. They did like soft packages.
2:08:16
Speaker 2: And we sign them, they're like, we want you to
2:08:17
Speaker 2: move these packages on the Symbara system. And we've put
2:08:19
Speaker 2: videos out about it and stuff. And the first month
2:08:22
Speaker 2: we signed them, Inga who runs accounts was like, we
2:08:27
Speaker 2: need to do this for them or they're gonna be
2:08:30
Speaker 2: really unhappy. And I was like, damn, that's like a
2:08:33
Speaker 2: compliant material that is moving while you're touching. Some of
2:08:38
Speaker 2: them touch and there's something hard inside. Some of them
2:08:40
Speaker 2: are squishy. They're like, there's tons of them. We're gonna
2:08:42
Speaker 2: move every three seconds. We gotta find the barcode, put
2:08:44
Speaker 2: it down. and put it in the middle of the
2:08:45
Speaker 2: conveyor every three seconds of package. I was like, it's
2:08:49
Speaker 2: 50-50 shots works. And it's gotta be with a neural net.
2:08:53
Speaker 2: And we got a bunch of data, trained it, policy,
2:08:57
Speaker 2: and right away it worked. And I was like, holy shit,
2:09:00
Speaker 2: this is like, it worked really good. And for some reason,
2:09:04
Speaker 2: the neural nets do extremely well under those like high
2:09:07
Speaker 2: variability environments that's like extremely diverse. They can learn the
2:09:11
Speaker 2: representations extremely well across like a wider distribution. And they
2:09:15
Speaker 2: just love it. Folding T-shirts, towels, like packages, like no problem.
2:09:21
Speaker 3: Wow.
2:09:22
Speaker 2: Stuff that would like, you know, you're replanning very fast
2:09:26
Speaker 2: as these things are all moving. It's doing that in
2:09:28
Speaker 2: real time. It's just like, it just works. Deep learning
2:09:32
Speaker 2: just works on humanoid hardware. Yeah.
2:09:39
Speaker 1: Crazy, crazy stuff. We've all seen it. The Department of
2:09:46
Speaker 1: War is operating in a world that's changing faster than ever.
2:09:50
Speaker 1: That's why so many guests on my show talk about
2:09:53
Speaker 1: the importance of continued innovation and technology in the military.
2:09:57
Speaker 1: But here's the problem. Working with the Department of War
2:10:00
Speaker 1: can be complex. For many companies, the process isn't always transparent.
2:10:06
Speaker 1: It's hard to know who the right stakeholders are, where
2:10:09
Speaker 1: the decisions are made, or how funding actually moves. Even
2:10:13
Speaker 1: when you get the right conversation started, you often hear
2:10:16
Speaker 1: the same responses. We like the technology, but funding is
2:10:20
Speaker 1: already allocated. Or check back next fiscal year. That leaves
2:10:25
Speaker 1: a lot of capable teams with strong products, but no
2:10:28
Speaker 1: clear path forward. That's where my friends at SBIR Advisors
2:10:32
Speaker 1: come in. They've built a team of over 60 former
2:10:35
Speaker 1: acquisition officers who spent their careers inside that black box.
2:10:40
Speaker 1: They help you find the right buyers in the Department
2:10:43
Speaker 1: of War, find the money, and write winning proposals so
2:10:48
Speaker 1: you can get on the right contract and fast. Since 2020,
2:10:51
Speaker 1: they've helped small businesses like yours win over $ 600 million
2:10:56
Speaker 1: in government contracts. And they're a 100% veteran team dedicated
2:11:02
Speaker 1: to one thing, getting the best technology into the hands
2:11:06
Speaker 1: of the people who need it the most. the war fighters.
2:11:09
Speaker 1: If you're serious about selling to the Department of War,
2:11:12
Speaker 1: go to sbiradvisors.com. That's sbiradvisors.com. And if you mention my name,
2:11:21
Speaker 1: you'll get the first month free. Crazy, crazy stuff. Let's
2:11:31
Speaker 1: talk about your venture to save kids and schools. You
2:11:34
Speaker 1: ready to move on to that?
2:11:35
Speaker 2: Let's do it.
2:11:37
Speaker 1: I love this. Like, yes, yes. Can you give us
2:11:41
Speaker 1: the synopsis?
2:11:42
Speaker 2: Yeah. So back when I sold Bettery, I mean, I
2:11:47
Speaker 2: mentioned I got obsessed about a few different areas of
2:11:49
Speaker 2: like working on, you know, I always want to work
2:11:51
Speaker 2: on flying cars, but like the macro environment turned like
2:11:56
Speaker 2: extremely poor for like school shootings. Like I went from like,
2:12:00
Speaker 2: you know, it's really hard to track, but we went
2:12:01
Speaker 2: from like 30 to 40 events per year in the
2:12:04
Speaker 2: US to like 300. And that was over like a
2:12:07
Speaker 2: span of 10 years. And it's also really hard to
2:12:10
Speaker 2: understand why. It's like another thing that we could like
2:12:12
Speaker 2: spend time on, but it's, there's just like a, you know,
2:12:16
Speaker 2: a 10X, mostly in the US, you didn't really see
2:12:18
Speaker 2: this like a lot internationally. And, you know, we started
2:12:23
Speaker 2: looking at it, looking at it. I basically started reading
2:12:28
Speaker 2: a bunch of like research reports and other things. And
2:12:30
Speaker 2: I stumbled upon this technology, this like basic technology and
2:12:36
Speaker 2: kind of like terahertz radar, so basically like, or sometimes
2:12:39
Speaker 2: also called millimeter wave technology, where it's basically like, basically
2:12:44
Speaker 2: high frequency, like, it's basically like, it's radio RF, it's
2:12:49
Speaker 2: like radio frequencies, but basically done at a very high frequency,
2:12:52
Speaker 2: like in the two to three, 400 gigahertz. And it's
2:12:57
Speaker 2: basically like similar to, you know, when you're in an
2:12:58
Speaker 2: airport and you go in there and you like hold
2:13:00
Speaker 2: your hands up and like the LG system scan you,
2:13:02
Speaker 2: they're like a couple of feet away, they can see
2:13:05
Speaker 2: like anything, anything you have, like, you know, if you
2:13:09
Speaker 2: have a knife, gun, vape pen, whatever. I read a
2:13:12
Speaker 2: research report that showed... And my goal is, like, if
2:13:16
Speaker 2: you want to put it in schools, you can't scare
2:13:18
Speaker 2: the kids. You have to be able to... So, sorry,
2:13:22
Speaker 2: back up. My view in schools is if you want
2:13:24
Speaker 2: to solve it, you have to solve it from a
2:13:25
Speaker 2: perception perspective, meaning you have to see if people have...
2:13:29
Speaker 2: You have to understand if people have guns on them
2:13:31
Speaker 2: or not. You can like change, like there's like a
2:13:34
Speaker 2: regulation side that some people chase, which we're not chasing.
2:13:38
Speaker 2: And then there's like a, how do we actually like
2:13:39
Speaker 2: know if people have guns on them? Because if you
2:13:41
Speaker 2: know a kid has a gun on them, you can
2:13:43
Speaker 2: go like take it away. And then majority of all
2:13:47
Speaker 2: school shootings are unplanned. Most of them, like almost all
2:13:51
Speaker 2: of them are some kid bringing a gun in habitually.
2:13:55
Speaker 2: It's like their uncle's gun and they bring it into
2:13:57
Speaker 2: school like every day for like three months. They get
2:14:00
Speaker 2: in a fight at recess and they shoot something, they
2:14:02
Speaker 2: shoot the gun. sometimes shoot somebody, sometimes shoot it. And
2:14:06
Speaker 2: that is majority of all gun events. The ones where
2:14:09
Speaker 2: you see like a planned event that's like on like CNN,
2:14:12
Speaker 2: where somebody is like coming in with a machine gun
2:14:14
Speaker 2: or automatic weapon, it happens like one or two times
2:14:17
Speaker 2: a year. It's on the front page of the news.
2:14:20
Speaker 2: The majority of all the cases, like 90 something percent
2:14:22
Speaker 2: is all happening from unplanned. Folks are bringing in guns
2:14:25
Speaker 2: all the time and then they're shooting it. So you basically,
2:14:28
Speaker 2: what you can do is you can stop all those
2:14:30
Speaker 2: The planned ones are very difficult and maybe impossible to stop.
2:14:34
Speaker 2: But the 90-some percent of all other shootings, you can
2:14:36
Speaker 2: actually avoid. I think you can avoid those. Meaning, like,
2:14:39
Speaker 2: you can prevent them by knowing if somebody has a
2:14:41
Speaker 2: gun on them. You can do it the old-fashioned way,
2:14:45
Speaker 2: which is, like, metal detectors and all this other stuff. But, like,
2:14:47
Speaker 2: it's just, like, that's not... We don't want the kids
2:14:49
Speaker 2: going to school like that. That's just, like, not how
2:14:50
Speaker 2: I want my kids growing up. So, basically, the reason
2:14:54
Speaker 2: why I got obsessed with, like, terahertz imaging is you
2:14:57
Speaker 2: could basically do this at, like, at a larger offset, 10, 20,
2:15:01
Speaker 2: 30 meters away. You can do it at a high
2:15:03
Speaker 2: frame rate, and you basically get back a point cloud.
2:15:06
Speaker 2: You basically get back an image. It's like a three-dimensional
2:15:10
Speaker 2: camera image almost, but it's done in a radio frequency.
2:15:15
Speaker 2: You can look at it almost like an optical image.
2:15:18
Speaker 2: And the reason that's interesting is because if it's basically
2:15:22
Speaker 2: people bringing guns in habitually, and you can scan them
2:15:25
Speaker 2: at entrances, you're always coming in through a few doors
2:15:26
Speaker 2: at a school. you're not going in anywhere anymore. And
2:15:29
Speaker 2: the schools also have all procedures now for this stuff.
2:15:32
Speaker 2: You basically can do like offset scanning at, you know,
2:15:35
Speaker 2: five or 10, whatever, meters away. You can scan people
2:15:39
Speaker 2: as they're walking in passively. Like just like, as like,
2:15:41
Speaker 2: you know, just walking in, don't need to stop anybody.
2:15:44
Speaker 2: And you can basically scan, you know, most guns that
2:15:47
Speaker 2: are brought in schools are either in your pocket, waistband,
2:15:50
Speaker 2: or backpack. It's like most of all guns are being
2:15:52
Speaker 2: brought in there. And you can basically find them. And
2:15:55
Speaker 2: if you know that, you can basically like.
2:15:56
Speaker 1: Stop It will find a gun in a backpack? Yeah.
2:16:00
Speaker 1: No shit.
2:16:01
Speaker 2: Yeah, so... It's amazing.
2:16:03
Speaker 1: It'll find a concealed weapon anywhere.
2:16:06
Speaker 2: There's like, you know, yes. There's print in a backpack. Yes,
2:16:10
Speaker 2: you can find them in backpacks. You can find them
2:16:11
Speaker 2: in waistbands and pockets. So the story is, so I
2:16:14
Speaker 2: found this research report done by a few of these guys,
2:16:18
Speaker 2: you know, that were at NASA Jet Propulsion Lab. I
2:16:21
Speaker 2: write these two guys and they said, you know, sure,
2:16:24
Speaker 2: we'd love to have you over. I get over there
2:16:27
Speaker 2: And they're like, they tell me the whole backstory. They're like, listen,
2:16:29
Speaker 2: we developed this technology for standoff distance detection for the
2:16:33
Speaker 2: Iraq and Afghanistan war. It was funded by the US government.
2:16:37
Speaker 2: We worked on it for 10 years. And when the
2:16:39
Speaker 2: war stopped, like funding dropped to zero and we like,
2:16:42
Speaker 2: we were done. We didn't work on it anymore. And
2:16:45
Speaker 2: I'm like, oh, it sucks. And then I'm like, okay, well,
2:16:48
Speaker 2: I guess keep me posted if, you know, if this
2:16:50
Speaker 2: thing ever works out. And then they're towards the end, like, oh,
2:16:53
Speaker 2: you want to go see it? I'm like, what do
2:16:54
Speaker 2: you mean see it? They're like, it's in the basement,
2:16:56
Speaker 2: it's done. We did it. And this is in 2017, 2018.
2:17:01
Speaker 2: So this is like, I was like, oh yeah, let's
2:17:04
Speaker 2: walk down. Walked into the basement. There's like this tarp
2:17:07
Speaker 2: over this machine. Took a tarp off. They had a
2:17:11
Speaker 2: guy with like a mannequin that's sitting there with a
2:17:13
Speaker 2: gun underneath a shirt, like, I don't know, three or
2:17:16
Speaker 2: four meters away. They turned this machine on. It was
2:17:20
Speaker 2: built like 10 years ago. It had like a computer
2:17:22
Speaker 2: tower inside of it. And And then it had like
2:17:26
Speaker 2: a little screen next to it. So I started this
2:17:28
Speaker 2: machine and they basically moved over to the screen and
2:17:33
Speaker 2: the screen showed like as clear as day, like a
2:17:36
Speaker 2: photo of the, you can see the exact gun. You
2:17:38
Speaker 2: could see it in 3D, you could see in 2D,
2:17:40
Speaker 2: you could see in power. There's a bunch of other
2:17:42
Speaker 2: ways we can look at the data, but it was
2:17:43
Speaker 2: just crystal clear.
2:17:45
Speaker 1: Wow.
2:17:46
Speaker 2: And I was like, what happened here? We basically got
2:17:49
Speaker 2: to the end of this program and we don't have
2:17:51
Speaker 2: any more funding, so it's done. And I basically made
2:17:55
Speaker 2: the decision, long story short, I ended up chasing Archer
2:17:58
Speaker 2: at the time. I went and built Archer. And at
2:18:00
Speaker 2: the time, this was a big endeavor for me, going
2:18:04
Speaker 2: from software to deep tech hardware. So I basically decided
2:18:08
Speaker 2: to put Cover on hold. And, you know, Chase Archer.
2:18:12
Speaker 2: And then about two years ago, somebody came to my office,
2:18:16
Speaker 2: one of my investors, and was like, hey, I'm like
2:18:17
Speaker 2: looking at like trying to solve school shootings. I was
2:18:19
Speaker 2: just back from LA and I'm like trying to solve
2:18:22
Speaker 2: it with CCTVs, like the security cameras. He's like, the
2:18:26
Speaker 2: problem is like, you can't, you won't know until the
2:18:28
Speaker 2: gun goes off. And you won't like brandish a gun.
2:18:30
Speaker 2: You won't pull the gun up until you're like trying
2:18:32
Speaker 2: to like shoot it. So it's just like way too late.
2:18:34
Speaker 2: And I told him the story about how I went
2:18:36
Speaker 2: down this path and here, and he kind of like,
2:18:37
Speaker 2: he looked me dead in the eyes. He's like, I
2:18:40
Speaker 2: have kids and you have kids. You have a fiduciary
2:18:42
Speaker 2: duty to go build this. And it was right when
2:18:46
Speaker 2: my daughter was also applying to first grade and we
2:18:48
Speaker 2: were worried about it at schools. You know what I mean?
2:18:50
Speaker 2: Just looking at the fence and what goes in. Just
2:18:52
Speaker 2: like kind of anybody can go in. You know what
2:18:53
Speaker 2: I mean? So I was like, shit, I got to
2:18:56
Speaker 2: go do this. I ended up spinning the technology out
2:18:59
Speaker 2: of Jet Propulsion Lab at Caltech. And I own it.
2:19:03
Speaker 2: And started Cover two years ago. The OG team that
2:19:07
Speaker 2: built it is with me now.
2:19:10
Speaker 1: No way.
2:19:11
Speaker 2: We put an office in Pasadena. The main office is
2:19:13
Speaker 2: right next to JPL. And we've been working on this
2:19:16
Speaker 2: now for two years. I've been self-funding the whole thing.
2:19:19
Speaker 2: And we have a prototype that already works last year.
2:19:23
Speaker 2: And we'll have a full-scale prototype out, I hope by summer,
2:19:29
Speaker 2: in our lab. And then hopefully, if all goes well,
2:19:31
Speaker 2: by the end of the year, we're beta testing it
2:19:32
Speaker 2: in school.
2:19:33
Speaker 1: Wow.
2:19:34
Speaker 2: And we'll put it in that figure campus first even.
2:19:36
Speaker 1: Wow.
2:19:36
Speaker 2: This is like an optical play. Can you see it?
2:19:39
Speaker 2: This is an AI play saying, can you detect it now?
2:19:44
Speaker 2: There's 130,000 K-12 schools in the US. There's like 60
2:19:48
Speaker 2: or 80 million K-12 students. It's huge. But it's not
2:19:53
Speaker 2: just schools. It's stadiums and airports.
2:19:56
Speaker 1: It's everywhere. Hospitals, malls. Any venue, you can movie theaters.
2:20:02
Speaker 2: I had my last baby a year ago. Just like
2:20:03
Speaker 2: anybody can walk in the hospital. It's just like, it
2:20:06
Speaker 2: doesn't matter. They don't check you in. It's just like scary.
2:20:08
Speaker 2: And so anyway, we're getting close here. And the technologies
2:20:15
Speaker 2: we designed are incredible. Actually, we designed all of it.
2:20:18
Speaker 2: Like we designed the whole system that I saw, redesigned
2:20:21
Speaker 2: the whole system I saw seven years ago last year,
2:20:24
Speaker 2: but it was just too expensive. The systems we were
2:20:27
Speaker 2: using were like, certain parts on it were like 50, $ 60, 000.
2:20:30
Speaker 2: So we moved all of that into a, into a chip.
2:20:35
Speaker 2: And we spent last year and a half doing that work.
2:20:37
Speaker 2: Those chips are in our office now and working. Those
2:20:40
Speaker 2: chips are like $ 7 instead of $ 50, 000. There's only a
2:20:46
Speaker 2: few groups in the world that can make them and
2:20:47
Speaker 2: design them. We co-design them. We worked on the design
2:20:49
Speaker 2: with them, made them, fabricate them. And we have them
2:20:54
Speaker 2: now in our office. They work. We didn't like a lot,
2:20:56
Speaker 2: you know, we use many different, we use like a
2:20:58
Speaker 2: lot of chips, but they're like really cheap. And that's important.
2:21:03
Speaker 2: So we, you know, K through 12s will have like
2:21:05
Speaker 2: a large budget and we need to be able to
2:21:08
Speaker 2: get the cost down and make it affordable for every school.
2:21:10
Speaker 1: That's what I was going to ask. I mean, how
2:21:11
Speaker 1: are you going to, how are you going to get
2:21:12
Speaker 1: this in school? A lot of, a lot of schools
2:21:14
Speaker 1: won't do, they won't even hire a security guard. Yeah.
2:21:18
Speaker 2: There, there are big budgets, both at the federal and
2:21:20
Speaker 2: municipal level, like a lot of money to put that
2:21:24
Speaker 2: are going into school, like schools are getting subsidized for
2:21:26
Speaker 2: to put in a lot of stuff.
2:21:27
Speaker 3: Good.
2:21:28
Speaker 2: They're putting in CCTVs, cameras. They're putting in ballistic chalkboards,
2:21:34
Speaker 2: all kinds of stuff in the schools. There's a lot
2:21:37
Speaker 2: of cash there. The schools also spend a decent amount
2:21:39
Speaker 2: per student. And I think we get the cost down
2:21:42
Speaker 2: a reasonable amount per student that both public and private
2:21:46
Speaker 2: schools can afford. But we could have already had our
2:21:52
Speaker 2: systems beta testing in some schools by now if we
2:21:55
Speaker 2: didn't pivot a year and a half, year ago. It's
2:21:58
Speaker 2: been the last year trying to like 90% decrease the
2:22:01
Speaker 2: bill of materials, like the cost.
2:22:03
Speaker 1: Wow.
2:22:04
Speaker 2: It's just like that's needed to go big and make
2:22:06
Speaker 2: this really work well.
2:22:07
Speaker 1: I think we'll... Are you going to put anything else
2:22:11
Speaker 1: into it? Like here's an example. When I think of this,
2:22:17
Speaker 1: would there be a way to maybe facial recognition? Who's
2:22:24
Speaker 1: enrolled here? Who's not? Just for example, like the shooter
2:22:27
Speaker 1: that happened up at Nashville a couple of years ago,
2:22:29
Speaker 1: the Covenant School. Went to school there, but not at
2:22:32
Speaker 1: the time. And so if they would have had some
2:22:35
Speaker 1: type of facial recognition on top of what you had,
2:22:37
Speaker 1: that's like, this person doesn't go here. This person hasn't
2:22:40
Speaker 1: gotten this.
2:22:41
Speaker 2: Yep, 100%. We'll have cameras, maybe even some audio, like mics.
2:22:46
Speaker 2: Cameras will be really huge. You can really do a
2:22:48
Speaker 2: lot with just RGB cameras and understand what's really going on.
2:22:51
Speaker 2: You'll also get a lot of semantic understanding because guns
2:22:54
Speaker 2: are like, they're hidden somewhere, they're concealed. People are not
2:22:57
Speaker 2: walking in with handguns and shotguns into school. They're in
2:23:01
Speaker 2: a waistband, in a pocket, backpack. We can be really
2:23:04
Speaker 2: thoughtful about if somebody clearly doesn't have anything in their
2:23:09
Speaker 2: pockets when they're walking in, but they have a backpack,
2:23:11
Speaker 2: we can be thoughtful about we probably need to scan
2:23:13
Speaker 2: a backpack. And we maybe need to spend more time
2:23:18
Speaker 2: getting higher frame rate on this area. And then as
2:23:21
Speaker 2: you mentioned, like a lot of understanding about like, does
2:23:24
Speaker 2: this person belong here or not? Is this a weird
2:23:27
Speaker 2: time for somebody to be like leaving and walking back
2:23:30
Speaker 2: into the school? So there's just a lot of semantic
2:23:33
Speaker 2: grounding we can put into the models to really help
2:23:35
Speaker 2: like understand if there's threats or not. The schools are
2:23:37
Speaker 2: set up really well to do like random locker checks
2:23:40
Speaker 2: now and like, okay, this doesn't look okay or not.
2:23:43
Speaker 2: Like the schools are really well equipped for that. It's
2:23:45
Speaker 2: just like, we don't know what's happening. We actually think
2:23:48
Speaker 2: now that there's probably like like perhaps like tens of
2:23:52
Speaker 2: thousands of guns that are being brought into schools in
2:23:54
Speaker 2: the US across, you know, 130,000 schools every year. I
2:23:59
Speaker 2: think we're finding, what we're finding now is a very,
2:24:02
Speaker 2: very small percentage of them are found, that are brought in.
2:24:06
Speaker 2: And then from there, what we're also finding is actually
2:24:09
Speaker 2: a similar small percentage are actually being reported. Because if
2:24:13
Speaker 2: you report like a student that has a gun, they're
2:24:16
Speaker 2: going to juvie. So we're also finding out, we think
2:24:20
Speaker 2: a large percentage of guns are even found. And then
2:24:25
Speaker 2: of that, we think a large percentage are not even
2:24:27
Speaker 2: reported because, you know what I mean? It could wreck
2:24:30
Speaker 2: this kid's life, which is unclear what we should do
2:24:36
Speaker 2: here for that. That's terrible. But we think there's maybe
2:24:40
Speaker 2: tens of thousands, maybe hundreds of thousands of guns that
2:24:43
Speaker 2: are being brought in every year through the schools.
2:24:44
Speaker 1: Wow.
2:24:45
Speaker 2: You're reporting thousands. And you're seeing hundreds of shootings. So
2:24:51
Speaker 2: our view is that we think it's actually happening, like,
2:24:52
Speaker 2: you know, as a percentage, it's low, but as an
2:24:55
Speaker 2: absolute number, it's quite high. Yeah, so I'm excited about this.
2:25:01
Speaker 2: It's in some way, like, you know, I write this
2:25:03
Speaker 2: prediction every end of every year for, like, what, you know,
2:25:06
Speaker 2: will happen in the spaces I'm in, which is, like,
2:25:08
Speaker 2: flying cars, like, robotics, like, AI, and weapon detection and
2:25:12
Speaker 2: stuff like that. And, like, I did a post in December, like,
2:25:15
Speaker 2: here's what I think on these four areas. And, like, overwhelmingly, like,
2:25:19
Speaker 2: the most support I got, like, publicly, um, was just
2:25:22
Speaker 2: for cover. It just, like, um, I think it, I
2:25:25
Speaker 2: think it, you know, just, like, hits, I think it
2:25:27
Speaker 2: hits in a really good way with a lot of folks,
2:25:29
Speaker 2: maybe especially parents. So, um.
2:25:31
Speaker 1: Everybody's word. We're homeschooling. Yeah. Because of this shit. That's
2:25:35
Speaker 1: a big reason why we're homeschooling.
2:25:38
Speaker 2: Yeah. I, uh, I hear you. We're, my wife and I,
2:25:40
Speaker 2: when we think about where we're, put our kids and stuff, too,
2:25:43
Speaker 2: it's just, like, something we talk about every time, too,
2:25:44
Speaker 2: and it's, like, uh, You know, and it's like, it's
2:25:46
Speaker 2: probably like a low occurrence rate, but if it did happen,
2:25:48
Speaker 2: it's just like, you can't recover from that.
2:25:51
Speaker 1: You know what I mean? I mean, it's just every
2:25:52
Speaker 1: school I go to, I'm like, man, like you guys
2:25:54
Speaker 1: got to, it's just happened down the road.
2:25:57
Speaker 2: I had a good buddy, like had his house breaking into,
2:25:59
Speaker 2: he's got a family and stuff. It was like six
2:26:01
Speaker 2: months ago. And he just told me when I was
2:26:03
Speaker 2: talking to him last, he's like, the sense of security
2:26:06
Speaker 2: we have now in our home is just like, we'll
2:26:07
Speaker 2: never get back. And I just like, I didn't know,
2:26:08
Speaker 2: I didn't know what that felt like, like feeling like
2:26:11
Speaker 2: we were secure before. but we lost it now. And
2:26:14
Speaker 2: now we can definitely see it and feel it. And
2:26:15
Speaker 2: just like, we just never, we're never going to be
2:26:17
Speaker 2: able to get it back. And I've had like, you know,
2:26:21
Speaker 2: I've had like some close people I know that have
2:26:24
Speaker 2: been involved around this stuff. And it's just like, it's terrible.
2:26:28
Speaker 2: And so, you know, my agenda here is I think
2:26:31
Speaker 2: it can be prevented. I don't know if you're going
2:26:33
Speaker 2: to prevent all of them. I think you can prevent
2:26:34
Speaker 2: a lot of them. And then if, and then if
2:26:35
Speaker 2: you even have like, there's no real security there at
2:26:37
Speaker 2: all right now, but even if you have security, there's
2:26:39
Speaker 2: also like a sense of like, I got to bring
2:26:42
Speaker 2: a gun in here now. And there's real sophisticated AI
2:26:44
Speaker 2: that's in all these schools that can catch it. I
2:26:46
Speaker 2: think that's another big thing. You have that at TSA
2:26:49
Speaker 2: when you go to pre-check. Yeah, it's a deterrent. So
2:26:51
Speaker 2: you have that, but we can also find it. We
2:26:53
Speaker 2: can see underneath through backpacks and stuff. It happens at
2:26:56
Speaker 2: specialized radio frequencies. It happens at 200 to 300 gigahertz,
2:27:02
Speaker 2: then it happens to get at 600 gigahertz. And in
2:27:05
Speaker 2: between those bands, there's either FCC rules that prevent you
2:27:08
Speaker 2: from doing it, or there's atmospheric attenuation, meaning sometimes there's
2:27:13
Speaker 2: enough moisture in the atmosphere at certain radio frequencies that
2:27:16
Speaker 2: the radio frequencies don't do well and perform well. They
2:27:19
Speaker 2: perform well at these certain radio frequencies for the imaging
2:27:22
Speaker 2: stuff we do. So it's actually quite a hard technical feat.
2:27:25
Speaker 2: One of the reasons I didn't do it and did
2:27:27
Speaker 2: Archer is because I thought the cover stuff was actually
2:27:30
Speaker 2: harder than doing flying cars.
2:27:32
Speaker 1: No, it didn't.
2:27:35
Speaker 2: I actually think it still is. We still like, it's
2:27:38
Speaker 2: hard to, it's like, it seems like pretty obvious. Like,
2:27:40
Speaker 2: it's like a, what airports have, but do it 10
2:27:42
Speaker 2: times higher. So it seems like, and you look at Archer, like, shit, man,
2:27:45
Speaker 2: that looks really complicated. Or like at Figure. Cover is
2:27:50
Speaker 2: just like super niche area of folks that haven't, like, there's, like,
2:27:54
Speaker 2: the folks spending, like, in this space are like, they're doing, like,
2:27:59
Speaker 2: They're doing work in weather and space, and they're not
2:28:01
Speaker 2: doing this for shootings and security and stuff. There is
2:28:04
Speaker 2: no industry for this. And luckily, we have the world's
2:28:08
Speaker 2: best terahertz experts at Cover that are working every day
2:28:12
Speaker 2: on this, and they're really passionate. They're probably not getting
2:28:15
Speaker 2: paid enough, and they're just super passionate about solving this problem.
2:28:19
Speaker 2: And so anyway, I think the through line for Cover
2:28:22
Speaker 2: is I think it'll work. I think we'll be able
2:28:23
Speaker 2: to demonstrate it hopefully by end of this year. We'll
2:28:26
Speaker 2: be able to say, we'll have it at figure campus first,
2:28:29
Speaker 2: and then we'll put it in schools, hopefully on the
2:28:32
Speaker 2: West Coast, and maybe one or two, and we'll see
2:28:34
Speaker 2: how it goes. How do we market this? What do
2:28:38
Speaker 2: we tell the students, parents? There's a lot of stuff
2:28:40
Speaker 2: here we need to get right. But if it goes
2:28:42
Speaker 2: well from there, and we're getting low false positives, what
2:28:46
Speaker 2: we really want to do is make sure we don't
2:28:47
Speaker 2: freak the kids out. We don't want to think it's
2:28:50
Speaker 2: a gun, but it's a crayon box. That'd be terrible.
2:28:53
Speaker 2: So that's really an AI problem. So we basically want
2:28:57
Speaker 2: to make sure we have low false positives around the
2:29:00
Speaker 2: whole sys stack. That's a really hard problem to solve,
2:29:02
Speaker 2: especially for us, where you can be partially occluded on
2:29:05
Speaker 2: certain areas of the person or the weapon. And then
2:29:09
Speaker 2: we need to know what we see. It's actually, is
2:29:12
Speaker 2: it real or not? And so funny enough, if you
2:29:14
Speaker 2: come to our lab, we have just guns everywhere. And
2:29:19
Speaker 2: they're all bricked. You can't actually shoot them. But all day,
2:29:23
Speaker 2: we try to figure out how to put guns on
2:29:25
Speaker 2: humans or mannequins, and we try to figure out how
2:29:27
Speaker 2: to detect them.
2:29:28
Speaker 1: So how does it work? Is it shooting frequency and
2:29:32
Speaker 1: then detecting the response when it hits something solid?
2:29:35
Speaker 2: Yeah, that's exactly what it's doing. It's basically shooting out
2:29:39
Speaker 2: a radio frequency. It's like electromagnetic. It's like a little
2:29:41
Speaker 2: wave format that goes out. Very similar to how your
2:29:46
Speaker 2: Wi-Fi works in your home or 5G. Same type of concept.
2:29:50
Speaker 2: It's just on a higher frequency. like different radio frequency level.
2:29:54
Speaker 2: But think about your Wi-Fi and like you want to like,
2:29:59
Speaker 2: you know, an order of like 20X or so, like
2:30:01
Speaker 2: the radio frequency level, it's like operating in a few gigahertz,
2:30:03
Speaker 2: something like that. But we operate at much higher frequencies,
2:30:06
Speaker 2: like 300 gigahertz. So you want to like, whatever, call
2:30:09
Speaker 2: it 50, maybe it's 50, 100 times this. And then
2:30:13
Speaker 2: basically it shoots us out and this waveform comes back
2:30:17
Speaker 2: and we review it. And we look how long it
2:30:20
Speaker 2: took to come back. and we use beamforming and a
2:30:23
Speaker 2: couple other techniques to figure out what happened. But you're basically,
2:30:26
Speaker 2: it's the same as traditional radar technology, but we can
2:30:30
Speaker 2: shoot out, it comes back, it's not ionizing, it won't
2:30:32
Speaker 2: hurt you, it's perfectly fine to be around your Wi-Fi.
2:30:35
Speaker 2: And we can basically get both a 2D image of
2:30:40
Speaker 2: what's happening and a 3D point cloud. The 3D point
2:30:42
Speaker 2: cloud is what's really important. So if you have a
2:30:45
Speaker 2: weapon on you in your pocket for whatever, or say
2:30:48
Speaker 2: I have one in my chest, for example, we will
2:30:51
Speaker 2: start getting back the signals back from the top surface
2:30:54
Speaker 2: of the gun before we get your chest stuff back.
2:30:57
Speaker 2: In the case of your chest, you have a lot
2:30:58
Speaker 2: of water in your skin, and it'll somewhat attenuate in
2:31:01
Speaker 2: your chest. So then we'll get back an image from this,
2:31:04
Speaker 2: and we'll reconstruct it really fast. And we can reconstruct
2:31:07
Speaker 2: into somewhat of a three-dimensional point cloud like you do
2:31:09
Speaker 2: a camera. So you can basically get almost like it
2:31:11
Speaker 2: looks like an op. I showed you earlier today the
2:31:15
Speaker 2: vision from this. It looks like kind of a camera
2:31:16
Speaker 2: image is what you get back. And so from there,
2:31:19
Speaker 2: you can kind of like visually see what's really happening.
2:31:21
Speaker 2: In the case of the gun, you can see the gun,
2:31:23
Speaker 2: you can see the trigger in some cases. Yeah, and
2:31:25
Speaker 2: sometimes it might just be like the handle or the
2:31:27
Speaker 2: side of a gun or different places of it, but
2:31:29
Speaker 2: you can see it through materials. Like it could be backpacks,
2:31:31
Speaker 2: it could be clothing or a jacket, but most guns
2:31:35
Speaker 2: are all in the waistband pockets or backpacks, which makes sense, right?
2:31:39
Speaker 2: You're not wearing it around your neck on a, outside
2:31:41
Speaker 2: your shirt or... Things like this. That's where most weapons
2:31:45
Speaker 2: are entering a school. We have probably one of the
2:31:48
Speaker 2: best data scientists in the world that is obsessed with
2:31:52
Speaker 2: school shooting. He puts up the best school shooting analytics.
2:31:54
Speaker 2: He does it daily. He's done it for five years.
2:31:58
Speaker 2: He's working with us through this and we've done so
2:32:00
Speaker 2: much work on how students enter schools, how they exit,
2:32:04
Speaker 2: emergency responses, what solutions are on campus now for this. uh, like, uh, where, what,
2:32:10
Speaker 2: what data, uh, we find on, like, where guns are at,
2:32:13
Speaker 2: what type of guns and weapons are there. There's, um,
2:32:16
Speaker 2: I think it was, like, 200 knife stabbings last year.
2:32:18
Speaker 1: 200?
2:32:19
Speaker 2: Yeah. It's, like, it's so high and so dangerous, like,
2:32:23
Speaker 2: we're trying to, like, we can detect knives. Like, there's, um,
2:32:28
Speaker 2: you know, vape pens, whatever, like, whatever. It's not a
2:32:30
Speaker 2: metallic thing. It's, like, we can, it just doesn't matter
2:32:32
Speaker 2: what the object looks like. Different metallics will actually, like, um, uh, like,
2:32:37
Speaker 2: come back to the radar system a little bit differently.
2:32:39
Speaker 2: So you can kind of maybe sometimes tell if there's
2:32:40
Speaker 2: like a gun, like a metallic signature or not coming
2:32:42
Speaker 2: from the material. But the technology is like really kind
2:32:45
Speaker 2: of straightforward in a sense of like it's RF technology,
2:32:47
Speaker 2: like radio frequency technology. And you get back like an image.
2:32:51
Speaker 2: And we can use that image to build like a
2:32:53
Speaker 2: neural network to then look at it and say like,
2:32:55
Speaker 2: what is this thing? What time of day is it?
2:32:58
Speaker 2: Who is this human? Like, is this a dangerous threat
2:33:00
Speaker 2: or not? And we need to do a really good
2:33:02
Speaker 2: job of making sure we're accurate in those readings or not.
2:33:04
Speaker 2: If we're not, we're going to cause havoc. and if
2:33:07
Speaker 2: we're right a lot of times, we could basically start
2:33:09
Speaker 2: saving lives. There's on average one shooting every single day,
2:33:15
Speaker 2: more than that. There's over 300 or more or so
2:33:19
Speaker 2: shootings roughly a year, if you look back the last
2:33:21
Speaker 2: couple of years. Every single day, there's less school days
2:33:25
Speaker 2: in a year than 365, but roughly every day there's
2:33:27
Speaker 2: a school shooting in the US. That's just at K-12,
2:33:31
Speaker 2: not colleges. That's looking at the 130,000 K-12 schools in
2:33:33
Speaker 2: the US.
2:33:36
Speaker 1: I can't, you think this will be out in a
2:33:38
Speaker 1: couple of years?
2:33:38
Speaker 2: Yeah, I think we'll get it out in a couple
2:33:39
Speaker 2: of years. We have a team working day and night
2:33:41
Speaker 2: on this. We'll, you know, we'll probably, I'll probably increase
2:33:44
Speaker 2: funding into it this year significantly. And we'll, we'll take
2:33:50
Speaker 2: a bigger push in headcount. Yeah. But like right now,
2:33:55
Speaker 2: like right now, all things are on, like, can we
2:33:56
Speaker 2: get the first full system in a really stable spot
2:33:59
Speaker 2: that works? And we've had to do a lot to
2:34:03
Speaker 2: increase the field of view because like schools are, you know,
2:34:06
Speaker 2: several meters wide, multiple doors, sometimes double doors to get in.
2:34:10
Speaker 2: Like we need to scan all of that all the
2:34:11
Speaker 2: way through. So it's like a natural aperture that students
2:34:14
Speaker 2: are walking into, which is good. You're not walking inside
2:34:17
Speaker 2: of a building, you know, like through a brick wall.
2:34:19
Speaker 2: You have to walk into a door entrance. And we're
2:34:21
Speaker 2: trying to, yeah, basically we're trying to get that fully
2:34:25
Speaker 2: complete this year.
2:34:26
Speaker 1: Man, that is solid work. Real solid work. Let's talk
2:34:30
Speaker 1: about ARC.
2:34:32
Speaker 2: Let's do it. Ready? Okay, so I think my pitch
2:34:39
Speaker 2: here is I've been working on one of the hardest AI...
2:34:43
Speaker 2: I think humanoid AI is one of the hardest AI
2:34:45
Speaker 2: technologies on the planet. It's just an incredibly difficult problem
2:34:50
Speaker 2: that my team and I have been working through day
2:34:52
Speaker 2: and night for the last four years. So it's like, okay,
2:34:54
Speaker 2: we want to go build a crazy sci-fi future with
2:34:58
Speaker 2: flying cars, AI humanoids, and then on the other half
2:35:02
Speaker 2: of my life, I'm using an AI chatbot. like a
2:35:05
Speaker 2: frontier lab, like Gemini or ChatGPT. And it's so stupid.
2:35:11
Speaker 2: It doesn't know me at all, doesn't remember anything I'm saying,
2:35:13
Speaker 2: can't see what I'm doing. It can't use tools very well,
2:35:16
Speaker 2: uses the internet like really poorly, can't even order me
2:35:18
Speaker 2: a sandwich if I needed one right now. And like,
2:35:25
Speaker 2: it doesn't feel very futuristic. It felt futuristic three years ago,
2:35:29
Speaker 2: but now anymore, it's just like, it's just not very good.
2:35:34
Speaker 2: It feels like I'm in an incognito window searching Google.
2:35:36
Speaker 2: That's all I can do. It doesn't have access to
2:35:38
Speaker 2: my accounts. It doesn't know any of this stuff. Meanwhile,
2:35:41
Speaker 2: I think for me, I've just been sitting here for
2:35:44
Speaker 2: three years thinking, we're going to get Jarvis out of
2:35:47
Speaker 2: this from Iron Man. We're going to get something crazy
2:35:50
Speaker 2: out of AI. It's going to move to a point
2:35:52
Speaker 2: where it can listen and speak naturally like a human.
2:35:56
Speaker 2: It can see the world. It can use tools like
2:35:59
Speaker 2: a browser and terminal. It can do real work for
2:36:01
Speaker 2: you and help you out. It'll know you really well,
2:36:04
Speaker 2: it'll know Sean, it'll know everything you're ever doing, all
2:36:06
Speaker 2: your stuff and be really personal to you. We don't
2:36:09
Speaker 2: have anything like that now. I got like this stupid
2:36:11
Speaker 2: chatbot that doesn't remember the last thing I said to it.
2:36:14
Speaker 2: And so I decided to like, I said like there's
2:36:18
Speaker 2: two things here that are extremely broken. One on the
2:36:22
Speaker 2: AI side, we have like extreme, we have like a
2:36:26
Speaker 2: lot of gaps to get to, to get to like,
2:36:30
Speaker 2: extremely personalized AI intelligence. There's a lot of missed opportunity now,
2:36:38
Speaker 2: last few years. A lot of gaps there. The second
2:36:43
Speaker 2: thing is we're interacting with these AI systems through old
2:36:47
Speaker 2: pre-AI computers. If you're pre-ing up your phone or your
2:36:51
Speaker 2: Mac or your computer, they're all designed 20 years ago.
2:36:56
Speaker 2: It's like an really old interface. The chatbot's an old interface.
2:37:01
Speaker 2: It's the wrong interface to AGI. You're not going to
2:37:04
Speaker 2: get to Jarvis with those. So we have to go
2:37:07
Speaker 2: rebuild all the hardware from scratch.
2:37:11
Speaker 1: Holy shit.
2:37:12
Speaker 2: Yeah. And I don't see anybody... I've been sitting here
2:37:15
Speaker 2: for like a year and a half being like, somebody's
2:37:18
Speaker 2: going to do this really well and I can't wait
2:37:20
Speaker 2: for it. And nobody's doing it. I mean, look at Apple.
2:37:24
Speaker 2: They're just like, what are they doing? So I started
2:37:29
Speaker 2: a new lab, last summer called Hark. And it's an
2:37:34
Speaker 2: AI lab. And we're going to basically design what comes
2:37:38
Speaker 2: after the iPhone for AI. And we're going to design
2:37:41
Speaker 2: new models that are extremely multimodal that can solve this.
2:37:45
Speaker 1: No shit. Yeah.
2:37:47
Speaker 2: And we have like the world, some of the, I
2:37:49
Speaker 2: think some of the world's best AI folks of all time.
2:37:51
Speaker 2: And we have, we have like, we have the lead
2:37:55
Speaker 2: designer from the iPhone, Abadur on the team. I mean,
2:37:59
Speaker 2: he signed iPhone 15, 16, 17.
2:38:03
Speaker 1: So this is going to wind up being a device?
2:38:06
Speaker 1: Is it going to be a device?
2:38:07
Speaker 2: A family of devices.
2:38:09
Speaker 1: Yeah.
2:38:10
Speaker 2: And this will go really far. It'll replace your phone
2:38:15
Speaker 2: and computer and, you'll have, like, native AI systems that
2:38:19
Speaker 2: are always on, always thinking, always understanding, always there to help, like,
2:38:24
Speaker 2: doing stuff in the background. Like, we'll have near-perfect memory.
2:38:28
Speaker 2: We'll know everything about your life and what you like
2:38:31
Speaker 2: and don't like and be able to even, like, act
2:38:35
Speaker 2: as a coach and say, like, hey, you said you'd
2:38:37
Speaker 2: do this over 90 days and you're not doing this
2:38:38
Speaker 2: over here. It'll just, like, it'll just.
2:38:40
Speaker 1: Whoa, it'll hold you accountable.
2:38:41
Speaker 2: Hold you accountable. Oh, shit.
2:38:44
Speaker 1: Yeah.
2:38:45
Speaker 2: Yeah, we have, we've been We have hardware in our lab.
2:38:47
Speaker 2: We have... We work on AM models now. Like, stuff
2:38:51
Speaker 2: is crazy cool. And, um... Yeah, we're gonna, uh.
2:38:59
Speaker 2: I think we'll probably come out of stealth by the
2:39:00
Speaker 2: time this thing airs here between you and me.
2:39:02
Speaker 1: Holy shit.
2:39:04
Speaker 2: And, um... We're... I'm self-funding it right now.
2:39:08
Speaker 1: You're self-funding this one, too?
2:39:11
Speaker 2: Yeah, I'm self-funding it right now.
2:39:13
Speaker 3: And, um.
2:39:15
Speaker 2: And yeah, the team's great, man. I think it's going
2:39:21
Speaker 2: to be a massive opportunity. And I see the Frontier
2:39:25
Speaker 2: Labs heading in a really great place for them, but
2:39:29
Speaker 2: a very different place than where we're headed. Yeah.
2:39:36
Speaker 1: What are you most excited about?
2:39:38
Speaker 2: I just want to wake up to a world that
2:39:40
Speaker 2: I'm excited and inspired. It's like, you know, I love
2:39:51
Speaker 2: doing this stuff. I could have retired like 10 years,
2:39:54
Speaker 2: 12 years ago, 15 years ago. So like, I think
2:39:57
Speaker 2: I just want to work on cool, crazy shit. And
2:40:04
Speaker 2: I'm just excited for a world of flying cars and
2:40:06
Speaker 2: humanoid robots and helping prevent school shootings and Jarvis.
2:40:10
Speaker 1: I mean, how do you keep it all together?
2:40:14
Speaker 2: I don't see it.
2:40:14
Speaker 1: You're innovating. you just, the trick is just four major things.
2:40:19
Speaker 2: The trick is just to not sleep and always work.
2:40:22
Speaker 1: I'm good at that.
2:40:23
Speaker 2: You know what I mean? You get me. Like just, that's,
2:40:25
Speaker 2: that's how you do it. It's super simple. Um, no,
2:40:29
Speaker 2: I mean like, listen, I, I, uh, I mean, to
2:40:32
Speaker 2: be honest, like I've had to make some like tons
2:40:33
Speaker 2: of personal sacrifices. Like, you know, I think 10 years
2:40:37
Speaker 2: ago I would have like a part of my life
2:40:38
Speaker 2: that would like be dedicated to like golf trips and,
2:40:41
Speaker 2: you know, doing the annual, like, with, like, my friends
2:40:45
Speaker 2: and stuff. Like, I don't do that anymore. I spend,
2:40:47
Speaker 2: I have, like, my family and I have my companies.
2:40:50
Speaker 2: And that's all I do. And, um, you know, I really,
2:40:54
Speaker 2: I do, like, a few podcasts a year, not much.
2:40:56
Speaker 2: I'm excited to come here because, like, I love your show.
2:40:59
Speaker 2: And get the story out, too. You're great at it. And, um, so,
2:41:02
Speaker 2: you know, I just, like, protect my time and just, like,
2:41:04
Speaker 2: I go all in on these things. Like, I have
2:41:06
Speaker 2: my kids and I have my work kids, you know
2:41:08
Speaker 2: what I mean? Like, and so, like, um, which are, like,
2:41:10
Speaker 2: you know, these are my, like, they're, like, kind of
2:41:12
Speaker 2: like babies. Like, I go, you know, make them, You
2:41:13
Speaker 2: need constant care and attention. So I have like this
2:41:16
Speaker 2: family and I need to like, I go all in
2:41:18
Speaker 2: on it and I do everything else less good. You
2:41:21
Speaker 2: know what I mean? I'm like a shitty college friend,
2:41:23
Speaker 2: if you like, you know what I mean? If I
2:41:24
Speaker 2: had seen you in a while, like just like not
2:41:26
Speaker 2: gonna spend the half a day with you on Saturday
2:41:28
Speaker 2: if you're in town, I haven't seen you in 10 years.
2:41:29
Speaker 2: So it's just unfortunate. I wish, but like, you know,
2:41:32
Speaker 2: I care about these things more. I care about doing
2:41:34
Speaker 2: this stuff really well. And I'm really happy at it.
2:41:37
Speaker 2: You know, I'm happy with family and happy with like
2:41:39
Speaker 2: things going to work. And, you know, to be frank,
2:41:42
Speaker 2: I just, I, I was born and raised on a farm, man.
2:41:45
Speaker 2: And I get to do, like, work on this cool
2:41:47
Speaker 2: shit every day. And, you know, I got billions behind it, like,
2:41:53
Speaker 2: going for it. Great teams that work, like, their asses off. Like,
2:41:56
Speaker 2: teams that are, you know, here, came with. And it's great.
2:42:00
Speaker 2: And I, like, fired up to come every day and
2:42:02
Speaker 2: work to try to make this thing happen. And I
2:42:05
Speaker 2: hope these things all work. It's just, but, like, I
2:42:06
Speaker 2: don't know. These are also hard businesses, so.
2:42:10
Speaker 1: It's pretty incredible. I mean, a farm boy from a
2:42:12
Speaker 1: town of 700 people now building that thing, flying cars,
2:42:17
Speaker 1: keeping kids safe in park. I mean, American Dream is
2:42:25
Speaker 1: still very much alive and well. That's fucking cool to see.
2:42:28
Speaker 2: It's cool. I feel just internally grateful to have a
2:42:32
Speaker 2: shot to do this. I feel like young entrepreneur Brett
2:42:36
Speaker 2: 20 years ago had been like, No fucking way you
2:42:38
Speaker 2: get a shot to go do this stuff, you know?
2:42:40
Speaker 2: And it's great. I just, yeah, I just, I'm taking a,
2:42:45
Speaker 2: I'm probably at, I feel like peak career and my
2:42:48
Speaker 2: team with me is like peak team, peak resources. The
2:42:52
Speaker 2: stuff I'm working on, I feel like is very important
2:42:54
Speaker 2: for the world, which is also great. I didn't, you know,
2:42:56
Speaker 2: doing vetery is like, there was a part of me
2:42:58
Speaker 2: saying like, okay, is this like the thing I want
2:43:01
Speaker 2: to spend my whole life doing? And I have that here,
2:43:04
Speaker 2: which is great. These are like the things I want
2:43:05
Speaker 2: to spend all my time on for the next like 20, 30,
2:43:07
Speaker 2: 40 years. Um, so it's good. I'm just like, uh,
2:43:11
Speaker 2: I just don't wanna, don't wanna screw it up now,
2:43:12
Speaker 2: you know?
2:43:13
Speaker 3: Oh, yeah.
2:43:14
Speaker 2: Make them work.
2:43:15
Speaker 1: We're doing a pretty damn good job, I think. All right,
2:43:18
Speaker 1: we're wrapping up the interview. I got a hot question
2:43:20
Speaker 1: to ask you. You ready?
2:43:22
Speaker 2: Let's do it.
2:43:24
Speaker 1: For decades, movies taught us to fear robots becoming self-aware
2:43:28
Speaker 1: and turning on people. But in the real world, we
2:43:32
Speaker 1: still don't have public evidence of conscious machines. What we
2:43:36
Speaker 1: do have are real cases of robots harming people, from
2:43:39
Speaker 1: Robert Williams being killed by a Ford industrial robot in
2:43:43
Speaker 1: 1979 to the viral 2025 Unitree H1 malfunction that showed
2:43:49
Speaker 1: how violently a humanoid system can lose control, plus longstanding
2:43:53
Speaker 1: research warnings that robots in homes can create privacy and
2:43:57
Speaker 1: security vulnerabilities in ongoing global debate over autonomous weapons. So
2:44:03
Speaker 1: is the bigger threat not conscious machines at all, but
2:44:06
Speaker 1: obedient machines that can still malfunction, be hacked, surveilled through
2:44:10
Speaker 1: remotely controlled, or turned into tools of intimidation, assassination, or
2:44:15
Speaker 1: state power?
2:44:18
Speaker 2: I don't know how that person gets up and goes
2:44:20
Speaker 2: outside every day if you're that scared. So I think like...
2:44:26
Speaker 2: The future is... this future, it can be molded and
2:44:37
Speaker 2: morphed and it's what we want to do with our time.
2:44:43
Speaker 2: If we want a future full of robotic systems that
2:44:45
Speaker 2: can help us out and free us of our times
2:44:47
Speaker 2: and things like this, we're going to will our way
2:44:49
Speaker 2: to make that happen. I'm a pretty optimistic person. I
2:44:57
Speaker 2: feel that having millions and then billions of humanoid robots
2:45:01
Speaker 2: on the planet. It's just gonna be such a magical
2:45:04
Speaker 2: and important thing for the world. Um, are we gonna have, like,
2:45:09
Speaker 2: you know, bumps along the way? Like, for sure. Uh,
2:45:13
Speaker 2: are they gonna, you know, hurt somebody at some point? Like,
2:45:15
Speaker 2: I think that's bound to happen at some point with
2:45:18
Speaker 2: enough scale. Um, but I think, like, the... the spirit
2:45:25
Speaker 2: here for humanity to get this done, uh, I think
2:45:29
Speaker 2: is here and I think it's going to be one
2:45:31
Speaker 2: of the most important technologies of our lifetime. I think
2:45:34
Speaker 2: in some way this AI stuff of like, we're generating
2:45:38
Speaker 2: AI systems that can be embodied and can use computers.
2:45:41
Speaker 2: It's going to be one of the most transformative technologies
2:45:43
Speaker 2: we've ever been through. We're building synthetic humans at scale.
2:45:49
Speaker 2: And it's both scary, but also very... I'm very excited
2:45:55
Speaker 2: about that future. So I think my view here is, yes,
2:46:03
Speaker 2: there's a lot of really difficult things that could go wrong,
2:46:09
Speaker 2: that perhaps maybe will go wrong, but I think we
2:46:13
Speaker 2: need this, just like we need cars. And I think
2:46:16
Speaker 2: just like we need a lot of things in life,
2:46:21
Speaker 2: airplanes and things. I think these are important technologies that
2:46:23
Speaker 2: really move society forward. So anyway, I happen to believe
2:46:30
Speaker 2: that this is like extremely important, will save lives and like,
2:46:35
Speaker 2: I think increase prosperity across like all of human civilization.
2:46:38
Speaker 2: And I think, uh, I'm excited to be working on it,
2:46:41
Speaker 2: but I think there is a lot of truth to what,
2:46:43
Speaker 2: you know, like I said, it's going to be a
2:46:44
Speaker 2: really hard road.
2:46:45
Speaker 1: Yeah. I mean, it's a, it's just a incredible advancement that,
2:46:49
Speaker 1: and you know, I know there's a lot of fear
2:46:51
Speaker 1: around AI. I have a lot of fear around AI,
2:46:53
Speaker 1: but we're going to go through it one way or another. Yeah. And, uh,
2:46:58
Speaker 1: I do think things are going to be a lot
2:46:59
Speaker 1: better on the other side of that.
2:47:00
Speaker 2: You're not stopping it now.
2:47:01
Speaker 1: Exactly.
2:47:02
Speaker 2: It's go time. It's going to happen for sure. I
2:47:04
Speaker 2: think it's going to be fine. I use AI every day.
2:47:06
Speaker 1: It's fine.
2:47:06
Speaker 2: It's a chatbot. There's different paths to go down from
2:47:20
Speaker 2: here that could be good or bad. I think my
2:47:24
Speaker 2: bets on high probability of really great. There's obviously always
2:47:28
Speaker 2: paths that could like not go well, but like being
2:47:31
Speaker 2: conscious of that and like basically doing everything possible to
2:47:35
Speaker 2: steer it in the right direction is like what we
2:47:38
Speaker 2: have to do at this point. Like, This is not
2:47:40
Speaker 2: something we can turn off. We can't turn off the internet.
2:47:43
Speaker 2: You're going to stop people from trying to build systems
2:47:46
Speaker 2: that make us more productive and do more work? I
2:47:48
Speaker 2: don't think that's happening. All we can do is basically
2:47:51
Speaker 2: do it the right way that has the best positive
2:47:54
Speaker 2: effect on the world.
2:47:56
Speaker 1: Another thing that comes to my mind is when we're
2:47:58
Speaker 1: talking about interacting with the humanoids. People, and I've had
2:48:05
Speaker 1: this discussion on other podcasts too, but People are going
2:48:09
Speaker 1: to look at that for advice, you know, relationship advice.
2:48:15
Speaker 1: And I mean, I think there's a, you know, a
2:48:17
Speaker 1: lot of important things are going to be talking to
2:48:20
Speaker 1: this thing too, about advice, certain people. And I think
2:48:25
Speaker 1: that's a big fear of a lot of, a lot
2:48:27
Speaker 1: of folks too. It's already happening with chat GPT and
2:48:29
Speaker 1: all these other cloud and all these other things anyways,
2:48:32
Speaker 1: but who are they getting advice from before that? Probably.
2:48:36
Speaker 1: I mean, you know what I mean? It's, it's, I
2:48:39
Speaker 1: think it's the caliber of person.
2:48:40
Speaker 2: Yeah, totally. Yeah, it's who you spend time with.
2:48:43
Speaker 1: Yeah. Last question. What advice do you have for future founders?
2:48:49
Speaker 2: I have a few things I think are, I wish
2:48:56
Speaker 2: I could like, maybe say differently also, like pass down
2:48:59
Speaker 2: to like young Brett, like 20 years ago. I think
2:49:03
Speaker 2: one is like, just go, just start building. I feel
2:49:08
Speaker 2: like a lot of folks get too caught up in
2:49:10
Speaker 2: this thing that's like going to be hard and might
2:49:13
Speaker 2: not work. And you can just like, it's just so
2:49:17
Speaker 2: easy to start a company these days. So many great tools.
2:49:20
Speaker 2: Just go learn. I think there's never been a situation
2:49:24
Speaker 2: where I haven't like done something and then learned a
2:49:27
Speaker 2: bunch and then haven't reset from that feedback. So almost
2:49:30
Speaker 2: like little stairs I'm climbing over and over throughout time.
2:49:34
Speaker 2: And so if I just wouldn't have started and wouldn't
2:49:37
Speaker 2: have moved, like I wouldn't have learned this information. So
2:49:40
Speaker 2: it's like a lot of information coming in, recursively self-improving
2:49:43
Speaker 2: and getting better over time. This could be simple things
2:49:45
Speaker 2: like hiring and doing accounting or running an engineering team
2:49:49
Speaker 2: or like trying to ship a product or getting feedback
2:49:51
Speaker 2: from customers. Like I'm just getting, I think I'm getting,
2:49:54
Speaker 2: it's like a sports player. You're getting better with more practice.
2:49:58
Speaker 2: And so I think the most important thing is just like,
2:50:00
Speaker 2: just go. I also think... The thing I learned a
2:50:04
Speaker 2: lot in my lifetime is what you work on is
2:50:07
Speaker 2: really a defining moment for founders. And it could be
2:50:11
Speaker 2: founders in any industry, tech, non-tech, or whatever. You're generally
2:50:16
Speaker 2: going to go and just try to have this kid
2:50:19
Speaker 2: that needs a lot of attention. And then at some point,
2:50:22
Speaker 2: it's like, you just can't abandon this thing. And you
2:50:24
Speaker 2: got to keep spending more time with it. And it
2:50:26
Speaker 2: needs a lot. And it's constantly working on the problems
2:50:28
Speaker 2: with it. So it's not the fun things. You're working
2:50:30
Speaker 2: on all the hard things. It's like this problem funnel.
2:50:33
Speaker 2: where I worked on the hardest, most pernicious problems at
2:50:36
Speaker 2: the company. So you gotta really love it. And it's
2:50:39
Speaker 2: not like you can be there for a year or two.
2:50:40
Speaker 2: You have to be there for sometimes a really long time.
2:50:43
Speaker 2: And even if you're successful and even if you sell
2:50:44
Speaker 2: your company or whatever, go public, you're getting your stock
2:50:47
Speaker 2: locked up or you're vesting out over many periods of time.
2:50:49
Speaker 2: You've gotta be in it for quite a while. And
2:50:53
Speaker 2: I find that for me, the things I work on
2:50:56
Speaker 2: as probably the most important things I could be doing
2:50:59
Speaker 2: with my decision-making And that's happening at a micro level
2:51:04
Speaker 2: inside the companies with what I work on week to week,
2:51:05
Speaker 2: month to month. But it's happening at a macro level
2:51:08
Speaker 2: where like, where do I spend my time? Like, I'm
2:51:10
Speaker 2: 39 right now. Like, where do I spend my time
2:51:12
Speaker 2: as 39-year-old Brett? And where does like 20-year-old Brett and
2:51:15
Speaker 2: 25-year-old Brett spend my time as an entrepreneur? And I
2:51:17
Speaker 2: generally have this philosophy that harder things are easier. Like,
2:51:21
Speaker 2: meaning there's like a nonlinear effect here for like starting
2:51:29
Speaker 2: companies that are like easier versus harder.
2:51:31
Speaker 1: Meaning.
2:51:32
Speaker 2: starting something that could be like a hundred times higher
2:51:34
Speaker 2: outcome is generally not a hundred times harder. So like
2:51:39
Speaker 2: doing figure is not a hundred times harder than doing
2:51:41
Speaker 2: another robot company. It's probably like three times harder, maybe
2:51:45
Speaker 2: five times harder. But the total rest of the market
2:51:47
Speaker 2: and opportunity is probably millions of times bigger than another
2:51:50
Speaker 2: like robot that's like on assembly line moving back and forth.
2:51:54
Speaker 2: And so I think there's like this nonlinear effect to
2:51:57
Speaker 2: like decision-making here. That is extremely important where harder things
2:52:01
Speaker 2: that have like larger outcomes are like usually easier to
2:52:03
Speaker 2: recruit the best talent in the world. That gives you
2:52:04
Speaker 2: a better lift to build a better product and a
2:52:06
Speaker 2: better team. That team and better product and maybe even
2:52:09
Speaker 2: a bigger industry, because it's harder, will give you like
2:52:11
Speaker 2: more capital coming at you for disposal to be able
2:52:14
Speaker 2: to like make the right investments you need and the
2:52:17
Speaker 2: right equipment or people or personnel or whatever marketing to
2:52:20
Speaker 2: basically make you more successful. And then you're generally worth
2:52:24
Speaker 2: working inside of bigger addressable markets like TAMs that you know,
2:52:29
Speaker 2: potential acquirers or public markets or other folks that really
2:52:31
Speaker 2: want to see and have like basically a disproportionate outcome.
2:52:34
Speaker 2: They want a high risk reward. They want to, you know,
2:52:38
Speaker 2: investors and things, and even people, they want to like
2:52:40
Speaker 2: go in and like, if it works, they want like
2:52:42
Speaker 2: a hundred X or a thousand X. They don't want
2:52:44
Speaker 2: like a two X. And generally for venture, like 90,
2:52:47
Speaker 2: 95% of people fail. So if it works, you really
2:52:49
Speaker 2: want to go, you want to hit a grand slam.
2:52:51
Speaker 2: So I think my philosophy is like spin, like, Choose
2:53:00
Speaker 2: wisely what you work on, young entrepreneurs. And then I
2:53:04
Speaker 2: would try to be as ambitious as possible. There's capital
2:53:07
Speaker 2: for that and there's humans for that that want to
2:53:10
Speaker 2: work at really crazy shit. We have them at my
2:53:12
Speaker 2: companies and they're incredible. You met some of them today.
2:53:15
Speaker 2: They're just like my design lead and a bunch of
2:53:19
Speaker 2: other folks here that are just unbelievable at what they do.
2:53:21
Speaker 2: They're the best in the world at what they do.
2:53:23
Speaker 2: But they want to come here and they want to
2:53:24
Speaker 2: try to do something that's never been done before. They
2:53:26
Speaker 2: don't want to go off and design the next car,
2:53:29
Speaker 2: or do the next AI product everybody else is doing.
2:53:31
Speaker 2: They want to be here, designing something revolutionary. So I
2:53:34
Speaker 2: think that's something not stressed enough. And I think the
2:53:37
Speaker 2: last thing is there's no rule book for this, which
2:53:41
Speaker 2: is really unfair. And there's a lot of people out
2:53:46
Speaker 2: there that will teach, like, here's what to do. And
2:53:49
Speaker 2: they're generally coming from folks that haven't done it before.
2:53:52
Speaker 2: And the signal and noise out there is just so high.
2:53:55
Speaker 2: I mean, you get a lot of noise out of
2:53:58
Speaker 2: like in the market, it's very noisy about like what
2:54:00
Speaker 2: to do and what successful means for building a team
2:54:03
Speaker 2: or hiring engineers or like executing a product. It's very difficult.
2:54:09
Speaker 2: And very few people in the world know how to
2:54:10
Speaker 2: do it really well consistently. And so I found over time,
2:54:14
Speaker 2: it's been really hard for me to get the right advice.
2:54:17
Speaker 2: And so I think it's been a lonely path. And
2:54:23
Speaker 2: for folks out there that are on that path, it's lonely,
2:54:28
Speaker 2: but I believe in you. You can do it. And
2:54:31
Speaker 2: I think that's, I've never had somebody for 20 years
2:54:33
Speaker 2: I could call and just like, what should I do
2:54:35
Speaker 2: in this situation? I never have had it. And I
2:54:39
Speaker 2: wish I had, but there's no book. There's nobody to call.
2:54:43
Speaker 2: And I think that makes it really hard, but it's possible.
2:54:48
Speaker 2: You can just like go do these things and it works.
2:54:51
Speaker 2: So for the folks out there that really want it,
2:54:52
Speaker 2: and it filters out like everybody who doesn't really want it.
2:54:56
Speaker 2: And you can tell the folks that want it. If
2:54:58
Speaker 2: I talk to people, they say, well, this is hard,
2:55:00
Speaker 2: that's hard. I'm like, you just don't want it. You
2:55:01
Speaker 2: shouldn't be doing this. You're going to get completely wiped out.
2:55:04
Speaker 1: You are.
2:55:05
Speaker 2: And it's like, it's the great filter. It's the folks that,
2:55:07
Speaker 2: you know, you went through buds. Like it's the great filter.
2:55:11
Speaker 2: It's 95% of everybody will fail and you'll devote your
2:55:13
Speaker 2: life into it and time and maybe all your money
2:55:15
Speaker 2: and your brand and you'll be embarrassed and you'll fail.
2:55:18
Speaker 2: Most will fail. And it's only for the folks that
2:55:22
Speaker 2: will like, I will like, you know, I will do
2:55:24
Speaker 2: whatever it takes to go make sure I make this happen.
2:55:27
Speaker 2: There is no failure. Those are the folks that do
2:55:29
Speaker 2: well here. And you can bend the world and you
2:55:34
Speaker 2: basically can mold the future to how you kind of
2:55:37
Speaker 2: want to, if you try hard enough. And the end
2:55:42
Speaker 2: of the goal, the goal at the end of the
2:55:43
Speaker 2: day is just to not die. If you don't quit,
2:55:47
Speaker 2: you won't die. So like, anyway, I think it's, listen,
2:55:52
Speaker 2: I've been playing this now for 20 years, still playing it.
2:55:54
Speaker 2: I feel like I'm in the early innings of my
2:55:56
Speaker 2: career now. I want to go ship at scale these systems.
2:56:00
Speaker 2: I haven't done that yet. We're like an inning, we're
2:56:02
Speaker 2: at inning one. Wow. And so like, but for everybody
2:56:06
Speaker 2: out there that's in that, I just think it's, I
2:56:09
Speaker 2: believe in you here. You can do it.
2:56:12
Speaker 1: That's great advice, man. Cool. Well, Brad, fascinating interview. Love
2:56:16
Speaker 1: everything you're doing, man.
2:56:18
Speaker 3: Like.
2:56:19
Speaker 1: Incredible stuff. Huge advancements.
2:56:22
Speaker 2: Sean, I'm a huge fan of you and just everything
2:56:24
Speaker 2: you do. So, I mean, having me here and, I mean,
2:56:27
Speaker 2: going through all this is just, it's been great. Thanks
2:56:30
Speaker 2: for having me.
2:56:31
Speaker 1: Thank you. It's been an honor.
2:56:32
Speaker 3: Great. Cheers.
2:56:38
Speaker 2: Yeah.
2:56:45
Speaker 1: No matter where you're watching The Sean Ryan Show from...
2:56:49
Speaker 1: If you get anything out of this at all, anything,
2:56:52
Speaker 1: please like, comment, and subscribe. And most importantly, share this
2:56:59
Speaker 1: everywhere you possibly can. And if you're feeling extra generous,
2:57:05
Speaker 1: head to Apple Podcasts and Spotify and leave us a review.