The equation I think for starting a robotic business has changed and will continue to change at an accelerating pace because the upfront cost is not that high anymore. Everyone's sort of spending a lot of time in the digital world and it feels like you know now is the time to start thinking about the world of atoms. You literally just gave people the playbook for how to build a vertical robotics company. This has really been our mission from the start is to create that Cumbrian explosion.

It still like blows my mind. I didn't know if this would exist even in my entire lifetime. Welcome back to another episode of the light cone. Today we have a very special guest, Quan Vang. He's one of the co-founders of physical intelligence, which we think might be the robotics AI lab that brings about the GPT1 moment for all of robotics. Kuang, thank you for joining us. Pleasure to be here. Has been a longtime admirer of YC and our mission is to build a model that can control any robot to do any task that is physically capable of and to do so at such a high level of performance that's going to be

useful to people in all walks of life. And so GPT1 for robotics you know what is it you know is the chat GBT moment for robotic real our perspective here is that um we want to build a model that's really intelligence we want to build a platform that allow us to externalize that intelligence to the rest of world and allow them to use it to build very interesting application in all sorts of vertical and robotics and we think that it's going to be more like a peeling an onions analogy where you start from a really strong base model that have all sorts of common sense knowledge and already works to some extent on your robot. Um you have then a mixed autonomy system uh very similar for example to a

autonomous driving car today. Um and then you actually deploy that system to do a real job. That system might make mistake um it's okay. Um and then over time by actually exposing the system to the complexity and the edge case of the real world that system get incrementally even just slightly better over time every day. Um and you know one day you wake up and you certainly have a system that is just fully autonomous and just provide tremendous value. Might be helpful to give the audience a bit of a mini history lesson on why robotics is so hard. And there's been a lot of breakthroughs in the last two years. And I mean just to simplify the robotics problem is three pillars.

Semantics which I think we got a lot of unlocks and with language models that somehow we ported into robotics. Then you have the planning and then the last thing is control which needs to be done in real time and interact with a environment that changes. walk us through the seinal papers that a lot of the team of PI robotics published that gave you the inkling that the GPT one moment is near and that started in 2024. Yeah. So the dream to build general purpose robot like robots has been a longtime dream I think in humanity like you know we're not the first to say that our mission is to build a model that can work on any robot. Um and we're really

fortunate to be in this moment in time in history where we feel that it's possible to kind of walk back a little bit um a few years before there was I think the first is Seikhan which to me was the first demonstration of language model and how you can bring all of the common sense knowledge in language model into robotics and therefore that significantly kind of reduces the need to collect robot specific data. So for example, if you have a task of oh I want to go to the YC office to record a podcast you know what are the step I need to take you can ask a language model you know just show me the step and show me the plan um and that work incredibly well um and then the way kind

of language model infiltrate if you will in robotic is it start at the planning level uh at the semantic level and then but there's still the control problem you know at the end of the day you still need a mechanism to convert the plan into low-level action that can actually actuate the robot And that bring us to POM E and that bring us to RT2 which stand for robotic transformer 2. And what this two work really show is that if you start from a vision language model that is really powerful and you kind of use robotic data to adapt this model to speak robot language if you will. um then you see a lot of transfer from the kind of knowledge that exists in the language in the vision language

model down to the low-level action like um one of my favorite example when we did the RT2 project was you can have picture of celebrity on the table you have a picture of Taylor Swift you have a picture of the queen of England and you can ask the robot you know pick up the coke can and move it to Taylor Swift even though the concept of Taylor Swift it just doesn't exist in the robot data at all and that work. You can do other examples such as um kind of spatial reasoning that doesn't exist in the robot data at all. Like for example, move the dinosaurs next to the like red car and these are always just completely unseen object in robot data. And so that was RT2 and that was palm E. Now RT2 and

POM E are single embodiment um exercise. Just for the audience, single embodiment meaning it worked for a very specific robot. In robotic, you can ask the question, how do you scale? Um, especially how do you scale data collections? And one of the insight that we had back then was, you know, maybe the data from one robot is not that different from another robots. Anyway, if you have enough robots in your training data, maybe what the model learned isn't to control one specific robot. what the model learned is something that's more abstract which is how do I kind of learn a general notion of what it means to control any particular robotic platform and therefore I will be better at

controlling any particular platform and that bring us to what we call uh open cross embodiment and robotic transformer X that was a big paper because it was the first that showed potential scaling laws that apply to robotics because now you could start training all these models across multiple kinds of hardware, not just one, which has never been done in robotics ever before. Because from all the research labs, they would all train with a very specific set of sensor actuators and motors and it was all very finicky with that particular hardware, right? Yeah. One of the really interesting result from um open cross embodiment and let me provide the context here is that you can take let's say 10 different

robot platform collect data from them train a policy and really optimize the policy to work well on that platform. Um so let's say you know you have that you have 10 different platform 10 different policies and now if you simply take the data and absorb it into a model that is high capacity enough to really absorb that data and you can compare you have this generalist right that learn to control how to the 10 different robot you can compare it to the specialist that has been optimized to work well on a particular embodiment how does it compare and the interesting result from open x is it was 50% better wow Um, and that was really surprising because in robotic it's hard enough to

get your model to work on one particular robot platform. And one of the reason why I say that we're really fortunate to be in this moment in time in robotic is because OpenX was really only possible because of the support that we received from the robotic community. It was a huge collaboration across the robotic community. And the reason why that's really important is there is this joke in like robotic grad school that you know if you want to add two years to your PhD just work on a new robot platform. You know by that logic if you want to have 10 robot platform that's 20 years like why is that it takes like a year or two to just get the platform um up and running to even collect the data.

Yeah. Is it fair to say that the data set that was created from embodiment X is similar to the scale of an impact that imageet did for vision because it was huge and it was the first large data set across multiple hardware huge collaboration and I still think that imageet was more impactful in the vision community and the reason for that is um a few the first is that imageet also allowed for reproducible evaluation right um you OpenX as an effort was more about making data available for kind of people to use and evaluation is a really difficult problems um in robotic that open X did not solve. Um and the second is I think open X is a drop in the bucket at this point in the robotic community. Um if

you measure in the kind of the scale and the volume and the diversity of data that the community is collecting, I think open at this point is a drop in the bucket. I mean, I guess we started talking about sort of GP1, but even GP1, you know, that was sort of this moment where you can prove, you know, Alec Radford figured out that there was a neuron based on a very specific input and output. Um, and then that allowed the scaling laws to sort of take hold. The biggest problem in robotics I've heard is basically actually exactly what we've been talking about is like it's the data problem. you know, language you could bootstrap off of like, you know, the sum total of what you could get off

the internet, which is actually quite a lot. Can you give us like a sense for um like scale? Is it like pabytes? Like, you know, what do you think is necessary as an input to you know the true GPT1 of robotics? Yeah. So, the data scarity problem in robotic there's a few way to look at it. The first way is that it's really two problem in this guys. There is the generation data generation problem and there's data capture problem and the difference is that the data capture is that there might already be lots of robotic data that is being generated but there's just never been really an incentive to capture it to make it easy for digestions in training. Um and

that's one of the goal that OpenX was trying to solve which is if you have robotic data it's a really good idea to capture it and make it possible to train on. The second way to look at it is that robotic is very different from language model. There is not a internet of robotic data that you can use. And so you see this kind of very operationally heavy effort to collect data. And there's the question is it going to scale? Well, the way that I look at it is let's take the US GDP$ 24 trillion US. Let's say if we actually solve robotics a model that can control any robot to do any task napkin math maybe contribute 10% to US GDP well that's already a massive number um and I think that promise is one of the reason

that warrants the investment into data collections um in robotics and the third way to look at it is we're very focused on cross embodiment and cross embodiment there is the data collection aspect of as well which is to really make sure that your model and your organizations and infrastructure are set up to consume data from many different sources of uh of robots and that actually allow you to scale easier. For example, I if I were to contrast our approach compared to let's say a company that have a particular hardware platform that they optimize for and they scale um it's not an approach that have really allowed people to scale. um because it's just much harder to figure out how do you manufacture like a thousand unit of

something for now compared to making sure that you yourself are ready to absorb data from like a thousand different types of robot that are already in there in the community. I mean it's a crazy problem isn't it? I mean the hardware itself even within the same design of embodiment if there's a hardware run that goes ary or like one of the servos is slightly different like you see it in the data right and then how do you control for that yeah so I think we were doing kind of like a inventory of robot in the company we were so shocked to find that there are no robot no two robot platform that are the same and if you ask people in the royal committee sometime there's

debate about multi-root versus single robot and the argument is that you know single robot is simpler to scale. And actually that's not how it plays out in practice. Like how it plays out in practice is even if you have a single robot that you're optimizing for over time that platform is going to drift. You know maybe you want to make hardware change or you have software change you end up in a situation where it's much harder for you to reuse old data because you know in machine learning if you want to generalize from a distribution you would like many sample from that distribution and if you just have one robot platform that have a major change every 3 months maybe you have a few data

point from that distribution. Um whereas if you start from the hypothesis that if you have many robot platform in your fleet your model is going to learn something more abstract which is how do I control a robot not any particular robot then the model will be able to ingest data from you know a slightly different robot better. Yeah. And actually we're starting to see immersion property in this kind of robot large prodation model. That's good news we're doing where you start to see like interesting transfer be between different um data sources for example today it's possible to perform tasks zero shot meaning you don't collect any data and these are the tasks that last year might have required like hundreds and hundreds of hours

what are some examples yeah do we have any videos we can see that like show so you know um I get might get some flak when I come back because this is not published result hopefully this will come out soon um so you know I want to reserve the excitement for that I'm kind of like building up the excitement a little bit. Um so hopefully this will come out soon. These are not simple tasks. These are like actually difficult task that just last year require like hundreds of hours of data collections. You hear on Ly cone first that there's some emergent properties that are going to come out of PI. Can you give us a sense of like the flavor of the tasks?

It's really easy to fool yourself and so we wanted to test across like few different tasks of different flavor. a task that require precision, task that require reasoning with multiple objects in the scene, it all seems to have this property. Um, that's really nice. So, it does seems like that's something that's kind of a more general property that emerged rather than we just, you know, got lucky and suddenly the model start working on one particular test. Could you help us understand where we are now in terms of like what's working and how well it's working? Like we're not quite at the chat GBT moment yet.

Like where are we? And I think you brought some videos that you were going to show us to like help everybody visualize what the current state-of-the-art actually looks like. I think where we are is I think if you have a task where it's okay for the robot to make a mistake um and it's possible for you to set up a mix autonomy system where you have a person that takes over when the robot make a mistake and provide corrections. it is possible to get to a level of performance where it start to make sense to think about scaling robot deployment and the example that I specifically want to highlight here is this blog post that we did with weave and ultra and you know

it's great that these are al both YC company I want to provide a little bit of context here first the context is that PI is a primarily research organization we want to focus on building the best model Um but we also want to not be tunnel vision. We want to make sure that the model that we built actually going to be useful and actually perform tasks that people in society cares about. And one of the really good way for us to do so is to partner really closely with company that want to get robot out there today. And the way that these relationship work is that we treat each other like we're on the same team very free flow of information. Um and we designed a system that try to get the best possible performance for the task

that these company care about. So let me talk about we fuss. What you're seeing in this video is a system that we built together folding really diverse item of laundry in a real laundromat. In the mission you can see you know people walking outside and why this task is difficult is because there's just infinite possibility of observation space like you know um clothing are deformable and no two item of clothing here are the same and these are also unseen you know these are not like clothing item that are seen in the training data. Yeah I love this team they are some of the most cracked people out of Apple I've ever met. Gary was the partner for weave. Maybe I want to like explain like what weave is and what their like what

their like company is. Yeah, I mean they're actually you know shipping their first robots into the home. Uh we sort of talked about it as you know being able to do household tasks like this and I think they were very inspired by physical intelligence's first demos with um with laundry folding. So it's actually a total trip to hear about it. you know a bit a year ago we were talking about them doing it and then now to see them do it working in hand handin-hand with you is really awesome. I think this is a great example of like you know you need the model smarts you need the data collection and then the hardware and um the sort of system integration all working together is just hard to nail. So yeah, and to get back to your question

about why robotic is hard, it's really it's it is a really hard system problems. Um like you need everything to work well and work well together to get this result. And like we've is such an incredible team for us to work with to get this result. And it actually didn't even take us that long to get this result. It was roughly well we set a goal and maybe it was like two weeks afterwards where we got a model that was got a model and a system that was good enough at performing this task. It still like blows my mind to see a robot actually folding laundry because I remember until basically until chat GPT I didn't know if this would exist even in my entire lifetime because like

folding laundry I mean it's it's always been like the Turing test for robotics because there's no way to like deterministically program a system the way that you did like preAI to do this because the space is like so infinite and like we've shown that it's possible for us to do like basically everyone can do this like robots will be able to do everything it's only a matter of like improving it from here. There was a funny story where um when we first published Pi Zero, people thought of us as the laundry company because the demo was just focused on laundry and actually picking home task, especially task that has to do with deformable object. It's a very intentional choice on our end.

We're not just after the home. We really want to make it broadly applicable. But picking home task for us to start with has a few benefits like one is relatable. you know, you can see the laundry folding demo and you can kind of like grock how this is going to be useful and you can get a sense of why it's hard. And the second is that it's really easy to set up to test generalization. You can talk about uh Ultra, which is your company, Jared, a demo of it. Yeah, this is Ultra. The thing that I love about this video is you see, you know, it's bright outside and you see this is 4x speed and it's 100 minutes.

If I scroll to the end, the sun has set. Oh, wow. Ah, that was one of the big problems in robotics where it would be so sensitive to the environment in lighting and mess up the vision system, the semantics and part of it. Yeah. And the interesting uh thing here is that it is possible to get to the level of autonomy that the robot is just performing the task. This is autonomy at scale. Like this is ready to be scaled. Quan, because this task is less familiar than laundry folding, do you want to explain what the robot is doing here and what Ultra is like doing as a company?

Ultra is a company that want to makes it really easy to adapt robot to, you know, new task. Um and right now they're focusing on logistic space which is really important because you know there's lots of labor shortage in logistic and the task that we focus on together here is you know if you order an item from Amazon you sometime get this soft pouch that item get shipped from and the task here is you have a tray of these items here and the robot is supposed to pick one of them at a time and place it inside this pouch. The machine would then close it and then pick up the pouch and put it um on the left here to be ready for shipping. Now, this heart is hard because there are many different types of object that can

be in this tray. And the opening here is actually very narrow. So, you see this interesting example of the robot kind of nudging the item to go into the pouch. And that's that's really hard. like that require very good understanding of the scene and like very precise motion to nudge the object into the pouch. Um the other thing that's hard about this task is the level of autonomy that's required like this is running for an entire day. There is still human intervention I want to say in um this like full day operation. Um but the level of intervention is actually quite minimal.

This is not just like some like demo station, right? This is actually recorded in an actual e-commerce warehouse where they're actually shipping real products to real customers. This isn't just like a lab. This is packaging real customer uh real order for customer to be shipped out in a real warehouse. So, this is real operations. So, I think this is really cool because I think when people think about robots, they tend to think of the consumer use cases like weave because that's, you know, what we're familiar with in our daily life. What I find really interesting is that there's like a million applications like this ultra thing that you wouldn't think of as

obviously like oh who packs the like soft pouch of things that you get from like Amazon. Well, there's some person like who does that and this is like a job that we can now build a robot to do. The interesting thing about the approach is that you're converting it from a very difficult engineering problem into a operation problems of how do I identify the use case and how do I collect the right data which is in some sense more scalable because you can build the system that allow you to collect data for many different tasks. So you know it's now a problem of how do I scale data collection rather than you know for every new task how do I design a really

difficult engineering system to solve it. YC Starter School is back. We're hand selecting the most promising builders in the world and flying them out to San Francisco for July 25th and 26th to discuss the cutting edge of tech. Apply now for a spot. Okay, back to the video. I think one thing that the audience may not know is that you have a very unique technical insight that in the past robotics folks would have kind of gasp and be shocked because robots need to run in real time. A lot of times all of the compute runs in on device but you guys have done something very different. Can you tell us more about that so that this works in real time with large models and really well? So the context here is that you know we

talked to many companies that would like to deploy robots and one of the first question we get is what compute unit should we get on the robot? You know it's expensive it's going to increase the bomb cost and they're worried that it's going to go out in fashion very quickly because the model change the model gets bigger. How do I make sure that the hardware that I'm going to commit to today is going to be viable for you know a couple of years? It's a very difficult questions. People often really surprise when I tell them that almost all of the robot evaluation that we run at PI today including the really complicated demo that we have shown making coffee folding laundry mobile

robots navigating around the model actually hosted in the cloud. Um and you know this is not like a cloud isn't a server in the office. It's a real cloud. The model is hosted in a data center somewhere and within this high frequency control loop that um is controlling the robot. The robot is actually querying an API endpoint that hosts the model sending it images and language command and getting back action that then execute it directly on the robot. And this is surprising because of precisely the reason that you mentioned you know how do you actually make it work? This is why it's really important for pi to couple

system hardware and model development and research like very tightly to together because like it allow us to solve for this problem. So for example, one of the insight that we have here is that you can actually bury the inference time within the robot control loop because you know if I'm a robot, I have enough action for me to execute for the next 100 millconds. Like there's no reason for me to wait until I finish executing that action to ask my model for a different action. You know, I can do it as fast as um inference essentially. Um, and so you know maybe when I only have 50 milliseconds of action worth left I can ask for the next sets of action and when the current 50 millisecond is over like I have

something that's ready for me to continue with you know my next 100 milliseconds. Um so that's one of the inside. The other uh kind of algorithmic improvement um we refer to them as real-time chunking design inference in such a way that you know there's going to be a delay in how long it takes to query the model on the cloud. Basically like the problem here if I get uh a little bit more technical is an action chunk is a sequence of action that I can execute on the robot. So you know it's not just one action. And if I have an action chunk that I can execute for 100 millisecond and 50 milliseconds in I want to predict another action chunk and I'm going to transition to that new action chunk after my current 50

millisecond is over. How do I make sure the two are consistent? Like you know how do I make sure that if I'm moving this way the next action chunk is going to continue me to allow me to continue to be smoothly moving this way. You can premputee. Yeah, you can premp compute and like that's one of the algorithmic improvement that we've made to make inference using model hosted in the cloud possible. I studied computer engineering so I'm not really an algorithms person but when it comes to systems like that like pipelining like get me all over that. That sounds great. That's so interesting.

I mean this simplifies it's kind of it's a brilliant choice because it simplifies so much of the system for the robots. You don't need all these clunky I don't know people have two operating systems at sometimes for robots embedded arts and then the regular one and all these complex giant compute and power and this is what the initial versions of Whimo used to run basically a server on the trunk and you can't afford to do that with general day robotics which is brilliant that you figure out how to do it. Yeah, you don't have to. I mean you can do things some of it there obviously has to be some compute there but a lot of the compute can happen elsewhere and then is there must be a video like

this thing that we're looking at in the top left like how much of that is sort of like video feed back how much of it is like local processed I mean is there any compute locally on this robot or is it just like a dumb like video camera that streams data to the cloud for this I am not 100% sure but I am inclined to believe that it's just a dumb computer. Like for this specific video, um I don't remember, but I'm just 100% confident that we can make this work with a dumb computer and the robot. And the one other interesting thing about our collaboration with Weven Ultra is one, I've never seen the robot in person.

Oh wow. Um two is I have very little idea about how the robot actually works. Interesting. Um and that's a very intentional choice. Like I want to stay away from that as far as possible. I also don't know how they collect data like I intentionally don't ask them this question to understand whether it's possible for an organization like PI to parachute into their existing system and to work really closely with them on the thing that actually matters to get the system to work and not have to learn about how they've set up their system because in a way that's like a more scalable recipe. Yeah, you completely

decouple a lot of the hardware control loop choices from the semantics and planning which is just works which is brilliant. Yeah, it I mean I'm really surprised that it works and when we started the company, we thought that real deployment is going to be a is only going to be in the conversation like 5 years um into the life of the company cuz the problem is really hard and we're two years in and you know this is the result that we have and real like deployment and scaling the number of robot is a really serious consideration today and so the pace of progress has just been very pleasantly much faster than we expected.

Originally, often on this podcast, we talk about like what all this means for startup founders. I think that might be an interesting question for us to explore here. So, if you imagine someone was listening to this podcast, maybe they're like a college student that's studying computer science and they think robots are really cool and they want to do something like this, how should they get started and what are the skills that they need? Do they need to be a mechanical engineer to be able to build a robot like this? Can they just buy an off-the-shelf like robot arm and camera system and like what and load pie and load piec more context. The first is that robotic is traditionally really hard because

it's an extremely vertically integrated business. You need to have your own customer relationship, your own hardware, your autonomy stack, your own safety certification, your own everything. And the barrier to entry is just really high because of that. And one of the thing that we're trying to change is that we're trying to provide a foundation of physical intelligence that the community can build on top of that allow them to onboard autonomy onto their robot and their task much quicker than before. So that's the first you know we want to provide that kind of seat of intelligence that allow people to move much faster so that they can you know focus on other problems. Um the second thing is that the I think the recipe for starting a

vertical robotic business today is one have a really good understanding of the existing workflow because the robotic system needs to fit into existing workflow and the second is to be very meticulous about identifying where the opportunity is. you know, if there's a workflow that need X number of work today, you know, where is the robot when you insert it? It's going to make the biggest difference. And two is to really be scrappy when it comes to hardware and data collections. You don't need a incredibly expensive robot that is capable of very precise motion today to be able to do this task. And the reason why is this model are really reactive and so they can compensate for some of the inaccuracy in the actual

robot movement and to ensure that you have the ability to collect data and to run evaluation especially evaluation in real deployment. The next step after that is to get a mixed autonomy system that allow you to get to the point where it's break even like break even economically. Break even economically because the reason why that's important is because it allow you to then scale the number of robot because if you lose money in every robot it's very hard to scale. That has been historically one of the biggest challenges for robotic companies that they go into growth stage. It's just the payback c period is just doesn't make sense.

Yeah. So the equation I think for starting a robotic business um has changed and will continue to change at an accelerating pace because the upfront cost is not that high anymore. And now you know what is the upfront cost? The upfront cost is much cheaper hardware, ability to collect data, ability to collect um evaluation and ability to kind of like understand the use case to see where they should insert the robot. you know, it's not about having incredibly expensive hardware. It's not about having your own proprietary, I think, autonomy classical uh stack anymore to be able to do this task. Um, and so it allow company to focus on the component that will actually allow them to differentiate

themselves from the rest of the space. Now that you've sort of unbundled it and you no longer need to build this fully vertically integrated company in order to build a robotics company, are we on the precipice of a Cambrian explosion of vertical robotics companies where there's going to be like a thousand companies like Ultra going after, you know, every like menial job in the economy and like getting a deep understanding of the customer, building a robot that can solve that problem, doing a like mixed human machine deployment until it like can run fully autonomously and building a company in every sector. Is that the future that you see people building on top of pi?

It's funny that you mentioned Cumbrian explosion because when we wrote this blog post there was that term that was very kind of like hotly debated. We are I think academics at heart and we want to be kind of very measure when we communicate but you know myself personally I believe there's going to be a Cambrian explosions of um robotic company across the entire world and across many different vertical um just because it's just so much cheaper to build and it doesn't require um you know someone with 20 years of experience in robotic to start anymore you know it require someone that is really scrappy that can move really quickly um can do the system integration um can understand customer what they want to start the deployment

I mean what's coming up for me is obviously we work with a lot of robotics companies and meet a lot of founders and it feels like there's this continuum um one is to use an analogy to you know personal computing you could argue that industrial robotics today is basically like mainframe or uh minicomp computer level like you know if you look back in the 70s huge public companies like digital computer that you know just did like these sort of very expensive deployments but like they were very specialized and it was all extreme enterprise like you know the idea of a personal computer was ridiculous right you know it took the altar and then Apple 1 and Apple 2 and then IBM PCXT to

like create personal computing and then like the traditional advice for robotics for many years is like go after like dirty and dangerous. And then of course those are sort of the industrial cases like you know you have these giant Tesla robots in the Gigafactory and things like that. It feels like what you said around profitability is really big. So, you know, does that mean that the people who do the vertical robot cambrian explosion sort of moment uh the people who are sort of first in that like it sounds like they would be the first to be profitable and not dirty and dangerous?

I think this is already happening today. I think um we have the fortune of having lots of visibility into the robotic community because um you know people would like to talk to us, people would like to learn you know what it's like to build a foundation model for robotic and people would like to know how do I get the same level of autonomy and there are so many companies and businesses that we talk to that would love to put the robot into that space that you know it's okay for the robot to make a mistake and they just need it so much. I really believe that the recipe that I mentioned earlier of identify where the robot can fit in, focus on cheaper hardware, collect data, run evaluation, mix autonomy, break

even, scale robots will work across many different vertical and I'm I'm seeing it play out today and it's just incredibly exciting to see. And this is pretty cool that you literally just gave people the playbook for how to build a vertical robotics company. like this is a playbook that could possibly be followed successfully hundreds or thousands of times. And the reason why I want to mention it is because I do want to see that Cambrian explosions and we want to help enable it. You know, for pi if if we talk about why pi is going to fail, it's probably going to be because the problem is just way too hard. You know, maybe it take 50 more years to solve the robotic problem and you know, not couple of years, five, 10. Um and so

we want to enable the community. We want to accelerate progress. And that's why we're very open like we publish our research. We open source PI 0 and PIO5. And people also shock when they asked me you know is there any difference between PI 0 and PI5 that you open source versus the model that we use internally PI 0 and PIO5. And the answer was actually no. It's the same model. like the pre-trained model weights that you're using that we open source is also the pre-trained model weights that our researcher internally use for PI 0 and PIO5 and so we really want to help accelerate progress in the community um and to create that Cambrian explosions.

Yeah, that's very inspiring. I mean, I feel like that's uh everyone's sort of spending a lot of time in the digital world and it feels like, you know, now is the time to start thinking about, you know, the world of atoms and uh this is sort of the perfect mix of actually like, you know, how do you take electrons and turn it into abundance in the, you know, Adam's world and I think about Dario Amade's essay um all watched over by machines of loving grace. And when you really think about the perfect manifestation of that, it's not like, you know, perfect uh agents that look over you just like in the electronic world. It's, you know, actually something a little bit more akin to what we're seeing here.

Yeah. And this has really been our mission from the start is to create that Cambrian explosion. Um and you know this is why we choose to focus on the model because we believe that is the borrow neck to just really make robot useful across many different tasks in the world and that's why we also focus on cross embodiment. You know success for us is not defined as only our model on a robot performing task that is useful. The surface area for success is actually much larger which is our model performing really useful tasks on somebody else robot out there. maybe that we don't even know what that robot is like in a way that's like useful to the end consumer.

Could we maybe talk a little bit about um like the humans behind the robots here? Like um how did the company get started? Like who are the who are your co-founders? How do you all get together and what skills you each bring to such a complex problem? Sometimes the joke I make here is that the human behind the robots are also robots. Not really. Um yeah, so Pi is a very I would say untraditional company. We have a like larger than average founding teams and some of us work really closely together when we were at the robotic team at Google. And the robotics team at Google was I think a really great environment for seeing the sign of life and creating the

relationships and the community that allow the robot community and like these advances to flourish. There is Locky uh which we met when we uh were thinking about starting the company and has just been really instrumental in making sure that we're a good business and there is Adnan our hardware lead um that came over from Andro and Adnan has a really difficult job because if you want to work on cross embodiment you remember my uh joke about how if you want to add two years to your grad school you bring on one more robots the hardware problem and the operational problem for us is how do We built improve and scale a fleet of had the joinious robot you know it's just not one robot platform and

because we built the organization from scratch in the beginning to support that like I think we're able to do it but it's just a really hard uh problem um because there's just like no two different robots in the fleet like how do you make sure that everything runs smoothly. We're really good at diva and conquer if you ask. Um, but so how many co-founders are there in total? We have Brian, we have Chelsea, Sergey, myself, Locky, and Adnan. Is it just necessary to have that many co-founders to solve a problem as big as this? Or was it a case like you were already sort of like a unit together?

You'd already worked together and you just what whatever you started, you would all have wanted to work together. One common question that we have is, you know, why band together? And, you know, the first is that we really enjoy each other company. um we spend a lot of time at work and it's you know in some sense give meaning to life and so we really want to enjoy the relationship we have at work. Um and the second is that you know any one of us could have started um a company and be successful but the problem is just so incredibly hard and the chances of success is just so much higher that we bent together and we can divide and conquer the problems. Um and you know that's that I think one of the

main reason why the progress has been much faster than we expected. What were the differences of um you working before in either academia or a big industry big company like Google and as opposed to now in a startup. This is the first time for a lot of you doing a startup right? Yeah this is the first time for a lot of us. Um one of the really surprising thing that we learned when we started the company is that the infrastructure for supporting large scale general purpose robot were just not there and you know this start from the software itself. How do you collect data? What device do you use to collect data? How do you manage the data? How do you uh annotate the data? How do you get

visibility into the data? How do you run evaluation? How do you build operational process? like there wasn't company that offer this kind of services which is very different from software and we were really surprised um to find out and so we end up writing a lot of the software at PI um oursel but I think this is another area of incredible opportunity of kind of building services for robot company like you know if you can offer remote teleout for example if you can offer data collections if you can offer annotation service because you know these are functions that doesn't need to be repeated from one company to the next. So I think there's lots of opportunity to build um kind of support

for growing robotic business. Um so that's one thing like one surprising thing that I learned and the second is I think one of the reason why we have managed to achieve such progress is that there is a really tight loop of collaboration in the entire life cycle of model development. um going from what task you collect data for. If you collect data for that task, how do you do it? What hardware do you use? Once after you collect the data, how do you get visibility? How do you ensure data quality? Um how do you then make sure that you can easily train on that data? If after you train on that, how do you run evaluation? Evaluation is really hard problem in robotic because it scale

super linearly to model capability. Like let's say you have a model that can perform a two-minute task. running evaluation for that is very different from running evaluation for a task that's 20 minutes like it's not 10 times harder. It's it's it's more than 10 times harder. um after you run evaluation how you can how do you can like dis dist distill the learning from that evaluation to know how to improve the model further like one of the really side project I would love to take on is to build a automated robotic research scientist um which is really one of the bottleneck we have today because this is a really difficult skill set um that require intuition about the entire stack so you know I would love it if there is a model

that can ingest multi model data such as this and analyze filler modes um you know understanding oh is the robot performing this way because of the data that was collected or the way that it was annotated or the way that we train the model and then you know suggest idea and actually try them to figure out if you know those hypothesis are correct so that's something that I would love to have and would like dramatically unlock us some sometime I make the joke in the company that we should record all of the meetings and then yes train a model to basically just make prediction about what is the next set Oh, you could. You totally could.

What if it's OpenClaw and um Obsidian and Markdown files and like you know a brain.md with like ontology that's custom to your use case and what if it's 100 open clause in the background that you orchestrate. I think there's two sides to this. The first is that we already see a little bit of a side of life where for simple failure modes um during evaluation if you can describe the way that the robot fail in text very precisely and very clearly then you know you can ask a language model to make very reasonable recommendation about what the next step is. Um but the flip side is that this only works for simple cases today. And the reason why that's the case is because I think it's pretty um

fundamental limitation of the model that we have today which is that they are not at the core model that take action in the world and see the consequences of its own action especially action that changes the physical world. Um and so I think this kind of very fundamental understanding about how the physical world works is missing from the really large foundation model. Um and I think that's that's one of the ingredient that's missing to be able to build this automated robot research scientist. What's interesting about openclaw? I don't know. I mean basically it can go and it can just do things which is interesting and then at that point it's on the research lab to provide like you know CLI MCP endpoints to the things that might control robots or uh

reconfigure rooms or I mean I think Karpathy feels like he's he's starting to talk a bunch about this where you know if you mix auto research plus what he's been talking about with markdown files like it might just happen in the open like it, you know, there's this sort of sense that you have to make something much more complicated to make it work. But what if that's just wrong? What if we just have markdown files and agents and, you know, you could make it yourself with, you know, literally clawed code and MCP today? What if it's not an algorithm problem? It's just literally an integration challenge.

We have a version of this internally that I use a lot. There was a point when I was spending a um embarrassingly large amount of money on API queries. Yeah. Um and you know the my team was like Juan what are you doing? Oh I'm that guy at Y Combinator right now. So uh the to give you an example um we have a uh clot skill that essentially serving the role of a pre-training on call today. Um so you know we have these pre-training runs that are really large. Um it's very I think a difficult exercise to keep them alive to you know for them to continue to churn just because there's so many things that can

go wrong and we have um a prototype a pre-training on call that kind of babysit the run and have the permission to take action to remedy error that it see um and the one of the surprising outcome of that exercise is that it leads about 50% improvement in compute usage like just overall compute utilization for that large pre- training run which is huge for us. Um and you know this is just a small simple prototype that I built and I think like there's a lot more to be done. Quan this is incredible. Thank you so much for everything. Thank you for making physical intelligence. Thank you for showing us these incredible demos. And uh honestly like the thing that gives me the most hope is this idea that

there's an entity there's a you know research lab out there that is focused on giving this to the world you know about to create this Cambrian explosion of robotic startups. So someone watching right now will be inspired by this and uh you know start playing with your models and they might create a robot that uh touches billions of people's lives in for the good. Thank you for having me been a pleasure. Um to the listener the one takeaway that I want you to have is I think robotic has changed a lot and the cost of building in robotic has decreased and I think will continue to dramatically um decrease and it also require a very different kind of scrappy skill set um that young startup like needs. We hope to enable really an explosion of many

many different robotic use case and you know always reach out to us if you want to collaborate. Thanks man. Thanks so much. Thank you.