All right. Good morning everybody. How are you? Ohio Gazimus. Um, great to be with you this morning. We have a incredible morning speaker. How many of you are interested in the future of robotics? All right. Well, you're in the right place then. Fantastic. You got the right guy. Brett Adcock. Come on out, Brett. Brett, great to have you. Uh, welcome to Dreamforce. >> Thanks for having >> Everybody here is interested in robotics. We're going to have a great time. You brought some friends here. Do you want to introduce these guys before we get going? >> Yeah, our friends. Um, yeah, nice to meet everyone. I'm Brett Adcock. Uh, this is our basically three generations of humanoid robots at Figure. Uh so on the right we have figure one which is our first generation that we unveiled. Um we we actually had this robot walking um in under a year since we started the company which is really fun. Uh and this is our second generation robot figure two. Uh and then last week we unveiled uh figure three uh basically our latest generation robot. >> Very good. All right. Well let's talk Brett. Come on. Let's have a seat. >> All right. Okay. So Brett, you are well, you know, I don't think a lot of people really know you and your background and kind of your Midwest roots. So why don't you just tell us a little about kind of where you're from, your background, uh kind of, you know, how you grew up and um and then before we jump into all the good stuff. >> Yeah, sounds good. Um and thanks for having me. This is like an incredible event. Uh yeah, so >> we are thrilled to have you actually so thank you for being here. >> So I I grew up in the Midwest, central Illinois. Um actually on a third generation farm uh we basically raised corn and soybeans and um and then ba basically since high school, middle school been building started building software companies uh in tech. So now it's been about about 20 years or so building technology businesses. I spent about 10 years building software companies, ended up selling one and then started a company about eight years ago called Archer Aviation uh build electric aircraft and then uh then started Figure about three and a half years ago. So we build um humanoid robots. So um >> now before you get in there I you you went pretty quick because I do think people don't really know what Archer Aviation is and how incredible that is and the work you did. So why don't you just kind of just hit exactly why that is what that is because it's so relevant before you bridge into figure. >> Yeah. So so at Archer we build electric vertical takeoff and landing aircraft. So it's like kind of a lot of words but basically we're building like a like basically a a fully electric helicopter merged with an airplane um that can basically move people in and around cities. So the goal is like roughly half the world lives in cities today. Um, we're stuck in traffic. I was stuck in traffic on the way here. Um, and um, >> I could have used one this morning, actually. >> I know we all need them. Um, so our our goal here is to basically help solve uh, this congestion by moving things into the air. Uh, you can do that now with like fully electric propulsion systems. Uh, it makes it far safer, cheaper. Um, you can basically just build a overall better product. So, we've now designed five generations of aircraft at Archer. Our current version, Midnight, is a piloted four passenger aircraft that we're working to certify here in the US with the FAA right now. Um, >> you made a point to focus on the word piloted. >> The the quickest way to get into the airspace and actually ship the product is is a piloted is where we live in a piloted airspace today. Um, I I think there's no reason that you can't do this autonomous from a technology perspective. Uh but like I think being able to then deliver product into market within a sub five 10 year like basically in a reasonable timeline uh we have a piloted airspace today we can certify an aircraft and get type certification with this. So our route is basically piloted and then over time slowly remove the pilot. So Brett, this is very important because I think this really sets up the whole conversation because here you are you're building soft, you know, you come out of the Midwest, you're building, you built this great software company. Now you start Archer. You have this huge aggressive vision for Archer. You have this, you know, aircraft in your mind. It's doing this. It's doing this, right? And it's not just piloted. It's also autonomous. >> That's something, you know, we've seen in the movies, in the cartoons. We don't have any of those flying around San Francisco exactly right now. Right. I didn't that's not how I got here today. I would have loved to been able to, you know, land right here. >> But, you know, you just you envisioned it, you built it, you built a factory to build it, you did it all right here in the United States. I just take those little pieces apart. And how how long were you at Archer? Was it four years, 5 years? What was the total period? And you took it public as well, right? It's a public company. >> Yeah. Um the the like the I guess the fun fun stuff >> tease it out a little bit for us. >> Sure. Um well, so I I basically had to go back to school to learn how to build electric aircraft. There's this funny thing here where like we found I found experts in certain areas of like rotorcraft for like vertical lift and like air like normal airplanes and then electrification, but none of them like overlapped. And so the rotocraft folks that built helicopters didn't understand the like the physics of and the aerodynamics of forward flight. The folks that were doing electric propulsion like batteries and electric motors didn't had no clue what that nothing in aviation is like really that electric. Um so we didn't find a good mix. I basically went back to school and we started I started the company out of the University of Florida where I basically went to college basically moved back started a small team there and we started building small like subscale aircraft for EV tall at some point I got the hang of this and we ended up moving the company out to California and then we started >> some most people say small aircraft they think it's a DJI drone or something that's not what we're talking about >> like a 12 15 foot wingspan's pretty big um but you Archers Midnight Aircraft has like a, you know, greater than 40 foot wingspan today. Six, like 6,000 lbs maximum takeoff weight. Like >> I only know it because all of a sudden I was in a hanger in San Francisco and all of a sudden one of them was in there. >> Yeah. >> And I was like, "What what is this?" >> Yeah. What is that? >> This is crazy. >> It works. Like it flies like every flight every week. So um >> Yeah. Yeah. So, basically had to like figure out how to marry like like mechanical engineering structures with aerodynamics uh controls, embedded systems. Um like is quite a hard and then like the older >> how did you do it so fast? That's what I just couldn't get my head around. I mean you it seemed like you went from Yeah. >> zero to prototype to production literally what was it a matter of two years, three years? >> Yeah. We um really spent a lot of time learn I I spent the first like year kind of as um a crazy person trying to build out this very complex like spreadsheet of how the whole aircraft work from top to bottom and size it. I called it my aircraft sizing like engineering sheet and um it was basically an input to say like how how big do the wing need to be, how many motors, how many fan blades, like how big the battery system is going to be and then you can basically boil everything down to like what what I call like we call >> are you just at home writing a spreadsheet of what >> I'm I'm at like un I'm at like a Marriott in Atlanta doing a course with a NASA professor on like a NASA guy on like electric a like electric propulsion and trying to put this all together and I had a really hard time um and ended up moving uh like down to like Gainesville, Florida and spent some time there like trying to work on that and then prototyping those the physics of that and see if it worked and I got a lot of the physics wrong the first few innings uh I built the wrong aircraft completely and um I I basically laid out the propulsion system wrong. I laid it out like longitudinally and it really wants to be lateral distribution for lift. You really want good axial flow into the fan blades. Sorry, a little too intense, but I basically figured out the physics of it, of what worked and then just worked like hell towards the vision of how to do that and chew it down. It's no different than what we do at Figure now. We're like, I don't want to jump too much from Figure, but we're three years old. We built like basically three generations of like the best humanoid hardware in the world. And so, I think it's just like we getting the right >> The reason why I think the Archer story is so important is because I think, you know, that you kind of you proved out that you could integrate hardware and software together. Yeah. And you could also deliver the electrical capabilities. >> Yeah. >> And also kind of overcome the physics. >> Yeah. >> Yeah. And then you had is there an autonomous archer today? >> Um our first few generations of Maker and early Midnights were flown autonomously without like or flown without a human in the aircraft. But our current generation of of aircraft for Midnight and the ones that we'll bring to market are all piloted. >> Because when you started Archer, we weren't really at the kind of threshold of AI that we are today. Right. I I called Michael Romanowski, the then director of the FAA. I cold called him in 2019 and I asked him, I want to bring an autonomous aircraft to market. And he he said, "Don't do it." He said, "The um airspace today is piloted. We can certify if you can get the technology built and the aircraft built, we will certify it and you can fly it and you can get it in the market. If you do autonomy, there is no road map right now for like having any definitive year that you could like you could sit there for two decades or a decade theoretically >> and just like >> that's in the United States. There's other countries there's even >> who are a little bit more aggressive right in terms of I was just in the UAE and they're they have you know I think a goal of having autonomous aircraft flying in their airspace like in the very short term >> I think it's like 60 70% of all flights are done in the Europe and US today and like YASA and the Europe and the US FAA are like the gold standards if you can get certified in one of these two regions you can get certified everywhere else >> that thinking about like what Zipline is doing and what these people are doing in terms of like drones or you or is when you started carrying people it must be it's a whole different >> these are civilian type certified aircraft space that needs to be rated to one in a billion hours to a catastrophic event 1 * 10us 9 >> you have to develop intrinsic safety on the aircraft with redundancy so that you can achieve these it's a very different ballgame you can um you can short circuit this and try to get certified in other countries um you're ultimately ultimately you need to design a safe system we like we travel today it's the safest form of transportation in the air anywhere you go better than any form of transportation you take. It's because of the intrinsic safety that the aircraft and the design process they go through and you you can short circuit it. You're just like you're just you're like short circuiting this like safety goal and you're you're going into markets that are smaller low like lower TAMs and you're ultimately you need to be in the biggest markets in the world to build something massive here and that's that's the FA and YASA and the YASA really piggybacks in the FA. So you really didn't have the FA as the gold standard for the globe on aircraft certification. >> When do you think half the people here will be coming to Dreamforce in some kind of autonomous aircraft? >> Oh man. Um at this point it's just like largely a certification level timeline. So the the the only thing left is basically like to get certified and get what they call like a type certification. Once that's done, you can start taking like civilians and start charging money for for flights. Well, it's it's it's governed like federally in the in the US. So like you have to get federal certification and then you there's some other types of certification you have to get, but largely you have to get a type certification and that's taken anywhere between like three and seven years for most flybywire aircrafts here in the US. So we're in like a like we're in the middle of that now just going through that whole process. So it's like um it's largely a regulatory timeline now. So what do you think the time frame is for you know let's saying like I said like here's large conference 50,000 people they all came in you know with pilots when when would they come in >> like 50 50% without pilots >> hopefully in the next like well it depends then it also depends on like how long it takes to build enough aircraft and saturate these markets and enough people here to take it that's a different story >> but like when can we >> my friend is a pilot is sitting in the front row So he's listening very carefully. It's >> like when can I fly this? Um like certainly under five years, maybe as soon as two or three. >> That was kind of be the balance I would say. >> Yeah. Yeah. >> All right. Well, this was I this was important for me cuz before we get into the robots, I really want people to kind of hear that you you have a you know deep understanding of not just th this category and where this is going, but also this this idea that it's >> Yeah. Can I say something real quick? You can absolutely do it. >> The when you think about the the components needed to build an electric aircraft, you basically need batteries, electric motors, control software, embedded systems, and sensors and structures. Those are the same things we need to build a humanoid robot. Like exactly the same stuff. It's all different components and all different ways to design it. They're design. It's it's orthogonal in the design space. There's not a single thing that we'll like use or want to use from aircraft, but it's the same stuff. It's the same team members, >> but you laid out also that there is a whole ecosystem though around it of governance and compliance and all these other pieces as well, right? So the stack is not >> it's more than just a technical stack. There's also a governance stack and a and a societal stack and a you know >> an an awareness and and it and do you think that is going to be different though between kind of the rapid kind of adoption of robots and where even kind of where we are right now in autonomous aircraft. Is is that is that the right way to think about it or is that wrong about it? Yeah. I know it'll take a lot longer to like get certified with Archer and and basically like scale that uh human robots is almost like the opposite land. There's like almost no regulatory >> reg. There's no regulatory. >> These aren't coming with pilots, right? >> No pilots. >> Okay. >> Sorry. >> And is that exciting you that there's no pilots? Is it easier from a governance perspective and compliance? Is that is that is that is that freeing for you in some way? >> It's exciting for me not to have this like binary regulatory body in here saying like you can go no-go. For an entrepreneur, you want to like get product out. >> There's no FAA for robots. >> There's no FA for robots. So like as soon as we like can build something intrinsic, >> should there be Um, >> I read the book, by the way. >> I know you did. >> There were some bad things. >> Yeah. Watched the end of the movie. Um, >> saw the movies. Our futurist wrote Minority Report. >> Yeah. And I saw a few other movies, too. I won't go through the names. >> I mean, it's like, >> you think there will be an FAA of robots? >> The the like there's a difference between a robot and a humanoid robot. Like we talk about like there's like all these articles coming out of like China like how many robots they have like these are industrial robots. >> We saw some nasty videos backstage. >> I saw. Yeah, that was crazy. Uh but like you know we think about like what real robots have been or we've been around. It's like we've been around these like you know like giant cuca like seven degree of freedom like you know warehouse industrial manufacturing robot. >> Let's not go down this rabbit hole. Let's just step back and come back into reality. So here you are. You're from the Midwest. You're in Florida. You've got the entrepreneurial bug. You're doing the software thing. You're like, I can do more than software. I can do engineering. You've got your spreadsheet. You have your NASA guy. You're defining the future of of airflight. And you're the future of flight. And here it is. And now you're seeing it. You're going now. You build the company. It's public. You've done it a few years. And you go, "No, no, I'm not going to do that anymore. New company." Boom. Figure. Now, that was only like 36 months ago, right? Okay. >> Yeah. >> Is that right? Am my time flame right or wrong? >> It's like a little over three years. >> Three. Okay. >> Yeah. >> Three years. Sorry. So, three years ago, all of a sudden, you say, "Oh, I'm new company. I'm going to do autonomous robots." Now, we have three generations of robots. Just you're going super fast. What has happened? What is it? How did you get to figure one so quickly? Everyone all of a sudden said, "Wow, here we are in figure one." >> Yeah, the the short answer is we when I started the company, we thought about all of those things you just mentioned and felt like how do we do this better than ever before? Um, we started with like here's our culture, here's our mission, vision, values, here's what we care about, here's where we want to go in the short term, here's where we want to go long term. We care about things different than most companies. We care about like turbo like like turbo velocity and we want to move extremely fast. We want to build a extremely flat org. We want to be really hands-on. So everybody in my or here at Figure CADs are codes. Uh we don't have like executive rooms like we don't have anything else. I sit on the floor next to engineering and uh our goal is to >> you zoom with people like me. Little robots are walking around behind you. >> Yeah. You're like that's cool. Um we should have that at the event. And um yeah the and the goal was to basically like work on something extremely challenging and the only way through that problem is to work extremely fast to unblock the critical path of like lessons learned about what works what doesn't. >> Did you were you able to take a few engineers from with you or were you you know did you just kind of start with a green sheet and your own just your own skills? >> Yeah. know we have um we have a whole crew of folks that I've been working with now for like almost 15 years. Uh actually like the my first employee uh at my vetery and early folks are here today. Lee and Logan, they're right there in the front row. >> Would you guys stand up if you're here from figure and just be recognized? >> Fantastic. >> Yeah. So we have uh and then we had a whole crew from Archer join. So we had like the whole core team from Archer there, core team from Better. Uh yeah, like Dana took me public at Archer. She's right here now from Figure. So uh yeah, we have a we we started with like a incredible team. I I was lucky enough to have enough money to like selffund the whole thing. So we were just like like we went to 30 people as fast as possible and then we got to million month of burn in like month four. So we were like it was like we were just at the at the start line like we want to walk the right architecture of humanoid within 12 months since the sequence. >> So day one you had how many people there in the office with you? >> Um it was me like Lee and Logan. We a few we were at the we work in Palo Alto. >> 5 to 10. >> No like yeah five to five something like that. >> Three to five. >> Pretty early on. Not like day one but like call it month one or something like that. >> And then after 30 days you had how many people? And within like 4 months we had like 20 to 30 people. So it took some time to recruit the best talent in the world. We were we were hiring from like the best robot. >> And literally within what a year or is it 18 months you had figure one? >> No. No. We we we like we had our CC Corp like filing date in like May of 2022. We walked that first robot you see on stage dynamically around the office in under a year from when we started the company. And it was incredible. And it worked. And we made a lot of decisions right. We made some wrong. And then we just like revved it into rev version two. And then we got a lot of things right. We got some things wrong and we revved it to get into version three. Version three for us is a robot that we think we can ship at scale. It's designed for high rate manufacturing. It's 90% cheaper in cost. It's it's just it's a it's a it's a it's a robot that can be fully general purpose. >> Okay. Now Elon Musk kind of says, "Look, there's three things here. You got the software, you got the hardware, and you have the manufacturing. And you're going to have to he you know he wants to press out a million of these and that's like obviously he's got the Tesla kind of factory experience. You you got kind of the same thing with the Chinese where the Chinese have this kind of you know factory experience high volume manufacturing. Then, you know, I mean, that's like what a lot of people are like thinking, well, the Chinese right now, you know, they're they're pressing these cars, the BYD cars or the Huawei cars or the whatever it is, and they're coming over on these huge, you know, um, tankers, and uh, god forbid we ever turn in any kind of military situation or at least turn into military objects cuz all of a sudden they're going to get loaded onto those same tankers that are bringing those cars over they're going to be bringing over the robots. So just address that. How are you gonna build a million of these or five million of these or 10 million of these? >> Can I like take a little like detour from this for one second? >> You absolutely can interview. >> There's like I get this question a lot like how you going to make a lot of robots and how's it going to be low cost and all this stuff. It's it's today that's not the issue. The issue today is you're watching the market and like everybody's out there doing like theatrics with robots, like open loop theatrics. They're like dancing or they're teleyoperating the robot and it's like it's like a movie set. And we're not at a point in society where anybody can be at a point where they if if you could make a million, you can make them work. You're you're talking about like one of the hard things about humanoid robots is just like as a little tangent is there's 40 degrees of freedom on this robot. The degree of freedom is a joint. A joint can theoretically spin 360 degrees as a motor. So the amount of states the robot can be in, the amount of positions it can be in are 360 degrees to the power of 40. That means there's more states of the robot, the more positions it can be in than atoms in a universe. You can't code your way out of this problem of getting that thing to do general purpose work. You have to use a neural net to learn those representations. It's the only that's the invention society made this blobby tissue that can like learn that stuff. And you need to then put in >> you see this as the hardest seinal the hardest problem that >> the manufacturing problem is like problem number like 50. It's not even close. >> And everybody's like saying who can make more. Nobody can even make a robot that can't like operate without a teleoperated human. In most cases, what you saw in the field he unveil last week for us is every task the robot was doing in manipulating objects was done by a neural network. We did it over like over a dozen different types of use cases. We did it fast. We did it under 30 days, maybe in a few weeks. >> And that's also what you demonstrated right at UPS, right? We have them on the assembly line now for quite a long time where they're sorting packages. >> So it's it's a BMW. We've been now there for five months now since April. This we run 10 hours a day. We've run every single working day. And the same robot that's running right now in Spartanberg was running in April when we started. And so we're getting like operational readiness and making sure the robots can actually do real work every single day. So at some point this will turn for us of like how do we manufacture at real scale which is what we're working on now. So if you came to our office, we have a whole facility called Bot Q. It's kind of like HQ but for bots. And we're trying to make we're trying to make thousands of robots as we speak right now. And it's hard. And the whole place is like a just like this soft it's advanced like software layer for like manufacturing. Everything's likeworked. All the tools talk to each other. We we we can scan a robot and know every single torque we put on every single bolt. It's unbelievable. The place looks it's it's phenomenal actually. It's one of my favorite places to visit. So we're learning how to do that. But the big next step is who can solve it's a race right now for humanoids of who can solve general robotics. Who can solve an intelligent like humanlike intelligence in the physical world? That's the race. That's what I would be scared of. >> And when will you be there? >> We feel we're at the we feel we just said to all hands this week. We all feel it's like now we can see it and it's like this little light at this tunnel and we don't know how far it's going to take. if it's going to take a month or if it's going to take a year. We see it now. >> Okay. But put a put a stake in the ground. >> I think we'll be able to do general purpose work with a humanoid by just through speech and have it do everything you'd want it to do in unseen places like a home it's never been in next year. >> Okay. When will five vendors exactly like yours be at that same spot? >> When will what? >> When will five vendors exactly like you be at that same spot? It looks at the as of right now we're multiple one or two years beyond anybody else in the world. >> So in three years like when will five vendors exactly like you be at that spot? >> Certainly been under five years. >> Five years >> certainly. >> Okay. When will five vendors exactly like you be able to press a million robots? >> I think you could figure out how to build a million robots fairly quickly if if you had general purpose today. >> Okay. But if you're able to deliver the but let's say within 5 years is the manufacturing this going to be some hard to like >> no >> crossbridge or not? >> No. It's like people like generally so when we started we were like oh we're we're going to do car manufacturing. Have you guys ever been to a car manufacturing company? It's like in it's like the most complicated thing I've ever seen in my entire career. >> You're already your company's already worth more than the every car company that I can think of. the like >> or your last founding last funding round I think was like at 30 billion >> a year ago it was at three billion >> and three years ago it was at I don't know what your first round was at >> 100 million 500 million a billion >> a little under 500 >> 500 so you went 500 3 billion 30 billion so you're already worth all the car companies you go buy you could probably go buy whatever you want >> you're going to build a company 200 times more valuable than Apple if you solve this it's half the world's GDP is human labor >> your goal is to build a company two and a half times more valuable than Apple. >> You will build a company like maybe hundreds times like you you'll build like definitely tens of more valuable than Apple. You will build a company worth tens of trillions for sure if you can solve this problem. >> Tens of trillions. >> Like like half of GDP is human labor. You could like deploy almost an infinite amount of synthetic humans. It's not just about it's not about just doing what humans do today. It's about like >> your goal is to build an infinite amount of synthetic humans. >> You're building a new spec. I think that is that what you said >> the goal. Yeah. >> And but it's not and will there be other forms? I mean we're looking at the human form but in the back we're looking at all kinds of like we saw the transformer movie. I don't know if you saw it. There were different forms. They do different things. They look like car. They look like that. >> Yeah. >> You know, is that where we're going? >> I don't think so. I think humanoid robots will be so much of a plurality of like of robots out there you won't even recognize the other ones. I think for various reasons. One is you want to generally manufacture one product at really high scale to get cost down. That's like a pretty easy. The second is you want to be able to understand what the training set is for neural nets. Our training set for humanoids >> is an unfair advantage which we're we're we're taking human data >> for like navigating around. We we can take data from just humans walking around an apartment and we can navigate around now >> just from that human video data and we can touch things like humans and grab things. We have five fingers. We have hands. Like we have the same affordances as a robot as a human. The robot has other robots don't have that and they're not intelligent. So you're going to have a situation where a robot goes out and every day while it's doing things and it's messing up, it's getting better about those mess ups. It's making less of those. It's sharing that back to the mother ship and they're basically doing we're basically training all that information to every robot. So unlike humans where I have to watch all my kids learn how to walk by basically like they don't listen. They just like want to do their own thing, right? And figure things out. And humanoids are different. Once one robot knows how to move packages or do your dishes or whatever, every robot in the fleet knows this. So you have a situation happening where you're basically going to have like recursive intelligence will be climbing and you'll be decreasing costs and a wheeled robot, a one-ar it'll just won't be able to compare because you're going to have a humanoid 10 grand, 20 grand or in the limit cheaper because they're going to make some cells and then you're going to have intelligence getting smarter every single day. And I don't think you have a world full of like different types of affordances given the world today we live in is made for humans. Like we had to come upstairs and stuff to get here. Like we it'd be like we can a general purpose robot will will will just beat out everything in my view. >> In the movie we are Legion, we are Bob. These humans have evol You already know where I'm going. >> Yeah. >> Okay. Go ahead. In the in the book, we are Legion, We Are Bob, the humans evolve into these robots and then launch themselves into space and it's the end of humanity. And do you see that? Do you see that kind of world? How is in your fan in your science fiction mind? >> You know, you read the book. >> Yeah. It was funny. I'm like laughing because like I was we were behind stage talking about like the what is like if you keep drawing out the future. I'm a big sci-fi nerd like you. Um, like one of the things I'm excited about is like like like self-replicating like von Newman probes in space like where robots can basically build themselves and go out and mine methane and do the right things on different planets and build more of themselves and keep like basically like colonize the the galaxy this way. >> And I think like this will happen. It looks like it might even happen in our lifetime which is like funny. >> Is it? I think it'll happen in our lifetime. I mean, we're like um we're compressing all the world's knowledge. >> So, we are we are Legion. We are Bob. This is >> factual. It's not science fiction. >> It's near-term reality. >> Feels like it's possible, my guys. >> In your mind, it is possible. >> Oh, it for sure will be possible. Yeah. >> In your mind, for sure it is possible. We are Legion. We are Bob. >> Yes. Shaking. >> It's just a matter of time. I mean like we'll approach this. Yeah, it'll work. >> Self-replicating. >> Yeah, we'll I mean we're like we have we have >> in outer space. >> I mean, boom. >> We have robots going >> with human consciousness. You've woken up inside one of these. >> Yes. >> Yeah. I think you could like >> Are you going to wake up inside of one of these? Which color did you like that you want to be? >> We have like this really cool like Dune outfit concierge robot in our release. I want to wake up in that outfit. Um I was just like I'm like laughing >> pointed to one of his employees. Yeah, I was laughing because like we were talking backstage about this and one of one of the guys David's like, "Do not talk about any of that. They don't want to hear about this." >> Yeah, don't talk about any of that. We won't go there. Don't worry. But we're going to We Are Legion, We Are Bob. Infinite numbers of synthetic humans in outer space self-replicating. >> Yeah, we're we're building a new species here and it's going to be like you like we'll start with, you know, thousands and millions and billions and like >> we're not building a robots, we're building a new species. Really? >> I think you're building a new species with onboard intelligence. >> Onboard consciousness. Yeah. I mean, >> it's not intelligence at this point, right? It's beyond intelligence, isn't it? >> Uh what's the difference then between intelligence, consciousness, and then what we're talking about when all these robots are talking to each other and communicating in ways that somehow maybe we don't even understand how we communicate and share like we are talking at many different levels right now, aren't we? So now explain that in terms of the world of the robot. One thing I want to say is like we're talking about self-reclicating robots is uh we will be also putting our first robots on the manufacturing line this year which will be really cool as we think about like ultimately closing the sloop on like robots that are truly self-replicating. >> Now let I just want to come back to our short short-term needs in our society. Now as you know or you may not know but anyway we have 50,000 people here. >> Yeah. And we're a little bit we're we're we just gave the SFPD a million dollars for sign on bonuses because they need to hire a lot more cops because we only have about 1,500 cops in our city. And uh we used to have 2500 when Gavin Newsome was the um mayor. Today we have about 1,500. So I was just wondering while you're talking see and looking over here and you know you're a narrative around us synthetic humans do you see this is that you'd be selling these to SFPD and saying look you're down 500 or a thousand I can offer you um robots to do some of these jobs even if they're not armed or not militar not militaristic is that a role that you see them playing in cities. >> Um like >> I mean you're looking at the video you could see a lot of use cases. >> It it just does everything a human can. It'll just do everything a human can. >> Our our whole product road map is designed more and more into what humans can do. Figure 3 was designed almost specifically for our neural nets. The heads, hands, everything was designed to learn better from humans. And that trend is going to continue for like that trend that'll go a long ways. >> How many times has the DoD showed up at your headquarters already? We um yeah, we get hit up like all the time for this. The we put in like our like online like our you know master plan when we started like we won't do anything like militarized right now and I don't think we like we have >> I saw that movie by the way and how it started. >> Yeah. Exactly. Um they're just like oh just do a little trial. >> We're not going to do the DoD and we won't do anything. Right. >> I just I don't think it's it's not what we want. >> Google also used to say that by the way. >> Yeah. It's just like not interested for us. We want to the civilian is the area of shipping robots in the home and the workforce. That's what gets us excited. That's what we want to really help out with. And I we don't really there's a whole other picture over here. It's like >> But when they're synthetic, self-replicating, will they make the same decision? >> Because you said they're self-replicating. So at some point, don't they're going to make their they can choose on their own. Why are you deciding for them? >> Yeah. Um, and one of the like really core reasons like also like wanted to work on this project and start this company is I feel like if we didn't >> It's a cool career that you've decided to work on for like How old are you now? >> I'm 39. >> You're 39. >> Yeah. >> It's going to be a cool life ahead, right? >> I'm excited. We're like just worked on something incredible. Like what I was going to say is like if we were like if we were not doing this it feels like we're approaching we are like we like you know we like we kind of approached artificial super like we're approaching like artificial super intelligence and these systems even digitally are going to get so smart and so capable in the future here that if if it's not like um if it's not humanoids doing the work it's going to be humans >> doing >> how will we know that we've achieved artificial super intelligence or AGI or ASI or whatever we're calling it today >> we've like we've just like did this thing with the world and did this like LM thing and we just like drove by AGI so fast and everybody just kind of waved by and we're kind of already past that. I think if you you surveyed everybody four years ago and said like is that thing AGI be like yep that thing's for sure AGI. Um so I think >> originally it was really defined as what you said earlier that is we you painted the picture the we are Bob we are legion picture of the self-replicating right >> today the AI is not >> no >> it's not self-engineering self-replicating that which I think was the definition I mean I'm not going to put words in your mouth but I think that the AGI ASI like it's self-replicated self-engineering it's not just it's not just claude you know kind of helping you to make your code more efficient. It's building the new claude. We're We're not exactly at that level, are we? Or are we? >> We're like on the exponential, so it's really hard to tell. It feels like uh I wouldn't bet against it. >> In your mind, we are. >> I mean, we're heading towards it right now. >> In your mind, you already see that happening. >> Yeah. Like >> you have total clarity for that vision. >> We like we internally we get to see at least >> What's the one thing that you see that you think nobody sees? >> I mean, we listen. We sit here internally. We get to see a year or two. we feel and see a year or two out at least in the robotic side of things like it's what we see and we see this stuff only getting extremely capable very cheap and humanlike intelligence in the home things like this we see that now like we we we we do some of it we do like pieces of it we we can't do all of it yet we can't like drop a humanoid in and just say clean my house I'll be back at 5 and and and it's done but we see the path to that that's for sure going to happen and we're just like how do we get there faster now and that's what we're doing every day in And like you know we go back to the office today we're just like we're sitting there obsessing over that one problem is how do we solve general robotics >> in our industry people are always overestimating what you can do in a year but underestimating what you can do in a decade and certainly two decades in two decades and 20 years a Dreamforce 2045 when you come back for this follow-on interview what will we be discussing and what will be happening or will I just be talking to a figure? while you're all flying here in an archer aircraft. And um one of the things we have at company goals internally that you should have, it'll be good for me is like having more humanoids at our office than humans. And we want to do that really fast. And I hope by way before then we come in here and there's like there's what 40 50,000 people here. I hope there's more humanoids than humans here at the event. Should be crazy. And the crazy thing is like you'll see at one point more humans. >> There'll be more humanoids here than humans. >> Yeah. Yeah. Yeah, for sure. They'll be doing everything. There's like a ton of humans here doing stuff all day long. It' be a humanist doing all that work. They'll be talking to them in real time speech to speech, which we do at our office. And >> this is the short term. >> This is way before 2045. >> This is way before 2045. This is really what >> this is for sure like sub five, seven years for sure. Like >> five to seven years. >> We'll have more humanoids in our than humans in our office next year for sure. >> So in seven years, I can see that there's, you know, people here of all ages in the audience, right? And um there's also, you know, people in their 20s, in their 30s, in their 40s. The life that they're going to experience. It's going to it's like hard to comprehend because we're like literally building sci-fi right now. The whole world is. And we're like, you watch sci-fi movies, you're like, "Oh, that's happening right now." It's like, "What's the ne like what's beyond this?" It's like hard. I don't know. Like um but it's cra it's definitely the most entertaining and interesting time to be alive in history. So yeah, I mean I think one thing you see from like at least the demonstrations we're doing is like you can see all these things. You can see incredible hardware now it's getting cheaper. We can make them. We're trying to make them in even higher volumes. Like you can see pockets of this doing like the hardest things possible you'd want it to do in a home. Like laundry and dishes. It's doing it with neural networks that we trained on human like data and we can do like pockets of it now. You showed it in the release. We can navigate around. We can talk to humans. We can do the things. It's not all quite there where it's like all perfect where package it up and you can get it and do all that. But it's like that's coming not it's not coming in 10 years. >> Let me ask you one more question that we didn't address. >> Are you going to have to build your own Manhattaniz data center to drive all these things? Um or are you going to use others data centers? How do you look at it? Do you see this as a vertical stack that you need to own the data center or you going to have to partner with somebody? Cuz all of a sudden I'm sure that you know like Sundar is coming here tomorrow. I'm sure he's not, you know, >> yeah, >> thinking, "Oh, I'm never going to do this." You know, you've got all these different all these different, you know, cast of characters and they're all like trying to acquire as much energy as they can acquire, build the biggest data center they can build because they see a world where all these fast food re like if you go to Burger King, two blocks from here, there's still humans making the burgers. Yeah. >> That's not your reality, is it? that there's >> do humans want to be making burgers. >> Okay. Exactly. So there we go. You just So your reality is within a short period of time, you said five to seven years, we'll be going to the Burger King a couple blocks from here. There is let's say the figure robots making the burgers or other kinds of technology doing that. Um but if it is figure, is it on your data centers? Is it other data centers? How do you think about your architecture in the long term in terms of and data? And then of course, you know, that must shift because you're about to launch all these robots into rockets and become self-replicating in the universe. So, I don't know how you're going to they're going to have to connect to some kind of data center. >> I mean, they'll probably have to do fully embedded systems there like onboard compute to do that really well. Okay. So, um, what, okay, one lesson learned I've had from Figure last three years is that I figured there's like a pretty mature efficient world. We can go out there and get things like motors and batteries and cameras and all this stuff and we would just put it into the robot. Maybe we do some stuff, maybe we don't. And I was like, we'll do some stuff and we'll go buy some stuff. And like that thesis is just like thrown out the window at high speeds. We we have to design we design everything now. We we like design our cameras for gen for generation 3. We do our motors, battery systems, like like sensors, uh like everything we like most of the things we do the software, >> rare earth metals. >> Um yeah, we we actually um >> that'll be easier because once you launch them out into the space, they can just mine the asteroids and then you don't have to worry about that anymore. >> Um >> get on other planets. >> So >> we don't have to worry about this planet. We can just mine whatever we need. >> Yeah, we should be able to do that, right? >> Yeah. It's very limiting being here on Earth, isn't it? >> It's pretty. >> Got to get the hell out of here. Um, I mean, think about all the stuff we got out there. We can just get get this done. >> There's a lot of >> Ladies and gentlemen, was this great? Please thank Brett Adcock.
Get an instant AI-powered summary of this video
Marc Benioff and Brett Adcock, founder of Figure AI, explore how AI-powered humanoid robots will transform the workforce, reshape industries, & redefine what’s possible in the age of intelligent tech. This video comes to you from Dreamforce, the conference that helps you unlock unprecedented success with the power of trusted AI Agents, real-time data, and CRM. This is where you become an Agentic Enterprise. Watch Marc Benioff's opening keynote here: https://www.salesforce.com/plus/experience/dreamforce_2025/dreamforce-main-keynote-2025 Watch all of our content from Dreamforce 2025 here: https://www.salesforce.com/plus/experience/dreamforce_2025 Continue your learning journey on Salesforce+: https://www.salesforce.com/plus Subscribe to Salesforce’s YouTube Channel: http://bit.ly/SalesforceSubscribe #Salesforce #Dreamforce Learn more about Salesforce: https://www.salesforce.com Facebook: https://www.facebook.com/salesforce Twitter: https://www.twitter.com/salesforce Instagram: https://www.instagram.com/salesforce LinkedIn: https://www.linkedin.com/company/salesforce About Salesforce: Improve customer relationships with Salesforce, the #1 AI CRM where humans with agents drive customer success together. Our integrated platform pairs your unified customer data with trusted agents that can assist, take action autonomously, and hand off seamlessly to your employees in sales, service, marketing, commerce, and more. Disclaimer: This video is for demonstration purposes only. All account data depicted is fictional and does not contain personally identifiable information (PII). This content is intended solely to illustrate product functionality in a simulated environment.