Pod Insights No Priors Sunday Robotics: Scaling the Home Robot Revolution with Co-Founders Tony Zhao and Cheng Chi About this episode The robotics industry is on the cusp of its own “GPT” moment, catalyzed by transformative research advances. Enter Memo, the first general-intelligence personal robot, focused on taking on your chores to give back your time. Sarah Guo sits down with Tony Zhao and Cheng Chi, co-founders of Sunday Robotics, to discuss the state of AI robotics. Tony and Cheng speak to the challenges they faced while developing their technology, the innovative glove system employed to scale real-world data collection, and the impact of diffusion policy and imitation learning. Plus, they talk about their 2026 in-home beta program and why personal robots are only a handful of years away from mass deployment.
Sign up for new podcasts every week. Email feedback to [email protected]
Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @tonyzzhao | @chichengcc | @sundayrobotics
Chapters:
00:00 – Tony Zhao and Cheng Chi Introduction
00:56 – State of AI Robotics
02:11 – Deploying a Robot Pre-AI
03:13 – Impact of Diffusion Policy
04:29 – Role of ACT and ALOHA
07:02 – Imitation Learning - Enter UMI
10:38 – Introducing Sunday
11:57 – Sunday’s Robot Design Philosophy
15:05 – Sunday’s Shipping Timeline
19:02 – Scale of Sunday’s Training Data
23:58 – Importance of Data Quality at Scale
24:56 – Technical Challenges
27:59 – When Will People Have Home Robots?
30:48 – Failures of Past Demos
32:34 – Sunday’s Demos
36:53 – What Sunday’s Hiring For
39:10 – Conclusion
Transcript 0:05 Speaker 00 Nobody want to do their dishes. Nobody want to do their laundry. People will love to spend more time with their family, with their loved ones. So what we believe in is that if the robot is cheap, safe, and capable, everyone will want our robot. And we see a future where we have more than 1 billion of these robots in people's homes within the decade.
0:29 Speaker 01 Thanks Memo.
0:35 Speaker 01 Hi listeners, welcome back to No Priors. Today, we're here with Tony Zhao and Chang Chi, co-founders of Sunday, makers of Memo, the first general home robot. We'll talk about AI in robotics, data collection, building a full-stack robotics company, and a world beyond toil. Welcome. Chang, Tony, thanks for being here.
0:53 Speaker 00 Thanks for having us.
0:55 Speaker 01 Okay, first, I want to ask, like, why are we here? Because classical robotics has not been an area of great optimism over time, or like massive velocity of work. And now people are talking about a foundation model for robotics, or a chat GPT moment. Can you just contextualize like the state of AI robotics and why we should be excited?
1:15 Speaker 00 I would say I think we're kind of in between the GPT moment and the chat GPT moment, like in the context of LMs. What it means is that it seems like we have a recipe that can be scaled, but we haven't scaled up yet. And we haven't set up so much so that we can have a great consumer product out of it. So this is where I mean like GPT, which is like a technology, and chat GPT, which is a product.
1:41 Speaker 02 Yeah, and so we're seeing across academia, there's consensus around what's the method for manipulation? But everybody's talking about scaling up. It's like, we know there's sign of life for the algorithms people are picking, but people don't know if we have more data, like what happened to GPT 2, GPT 3, will happen. And but we see a clear trend that, you know, there's no reason to believe that robotics doesn't follow the trajectory of other AI fields, that scaling up is going to improve performance.
2:11 Speaker 01 Maybe even if you took a step back, like, what was the process for deploying a robot into the world like 10 years ago? Like, pre-set of generalizable AI algorithms? Like, why was it so slow as a field?
2:24 Speaker 02 Yeah, so previously, you know, classical robotics have this sense, plan, act modular approach, where there's a human designing interface between each of the modules. And those are need to be designed for each specific task and each specific environment. In academia, that means for every task, that means a paper. So a paper is you design a task, design an environment, and you design interfaces. And then you produce engineering work for that specific task. But once you move on to the next task, you throw away all your code, all your work, and you start over again. And that's also kind of what happened to industry. And so for each application, people build a very specific software and hardware system around it. But it's not really generalizable. And therefore, it's just feel like we're just running in loops. We build a one system and then we build a next one.
3:08 Speaker 02 But there's like no synergy between them. And as a result, the progress has been somewhat slow.
3:13 Speaker 01 I feel like that's a good segue into some of the amazing research work that you guys have contributed over the last five years to the field. Should we start with diffusion policy? What was the impact of that?
3:23 Speaker 02 Yeah, so a diffusion policy is like specific algorithm for a paradigm called imitation learning. That's really like the most intuitive way of, you know, how to use machine learning for robotics. It's like you collect pair data, action, observation of what the robot should do. You use that to train a model with supervised learning, and then the robot do the same thing. The problem is that in the field, it's known to be very finicky. So when I talk to researchers, when I start into the field, people are like, the researcher themselves, the specific researcher, need to collect the data so that there's exactly one way to do everything. Otherwise, the robot either like, either the model training will diverge or the robot will behave some weird way. And diffusion model really allows us to capture multiple modes of behavior
4:08 Speaker 02 for the same observation in a way that's still preserve training stability. And that really kind of unlocked more scalable training and more scalable data collection.
4:17 Speaker 01 So it doesn't have to be you personally wearing, you know, a teleop set in order to make a robot learn?
4:23 Speaker 02 Yep, yep. So like, we can have multiple people, sometimes even untrained people collecting data, and the result will still be great.
4:29 Speaker 01 Where do Aloha and ACT play into this?
4:32 Speaker 00 Yeah, so DC papers are actually like super close to each other. They're like one month or two months away. That's actually how me and Chen know each other. It was by looking at each other's paper, like, and we met on Twitter, I think, when Chen is back in Columbia. Before Aloha, I think the typical way people collect data is with a like polyopera setup with Weare Haset. And it turns out to be very unintuitive to do. And it's hard to collect data that is actually dexterous. Well, Aloha brings this a very simple and reproducible setup. So it's very intuitive.
5:04 Speaker 01 Sorry, in terms of just for most people who haven't worn a teleop setup, is it the lag? Is it like just, you know, how should I compare it to like playing a video game or something?
5:14 Speaker 00 Yeah, I think Aloha make it feel more like playing a video game. Uh, normally it feels kind of disconnected that you're just like moving in the free air and Aloha is moving with some delays. Uh, but Aloha reduces that delay by a lot. And that contribute to the kind of smoothness and how fast human can react. Like, once we get those really dexterous data, what allows us to do is to investigate on algorithms that are actually solving things that are difficult. In this case is through the introducing of using transformers in the case of robotics. And there was a long period of time that I think robotics was stuck with uh three-layer MLPs and ConvNets. And as you make it deeper, it works worse.
5:57 Speaker 00 But it turns out that once you have very strong and dexterous data sets, like just throw a transformer at it and it works quite well.
6:06 Speaker 01 Actually, like just in terms of progress of the industry over time, transformers didn't make sense without certain level of data collection capability. Okay.
6:14 Speaker 00 And also auto system around it, for example, action chunking, which is to predict a trajectory as opposed to predicting single set of actions. All these things kind of combined to make dexterous task, bimodal tasks, more scalable.
6:28 Speaker 01 Why is uh chunking important here? If I think about like just the analogy to LMS and um like text sequence prediction.
6:35 Speaker 00 I think it just kind of throws the model off if you're trying to force it to react every millisecond. That's not, not, not how human act. We perceive and we can actually move quite a bit without uh looking at things again. And that turns out to make things, uh, the motion a lot more consistent and overall performance to be a lot better.
6:57 Speaker 01 And you discovered that actually transformers architecturally did apply to robotics. Cheng, you felt then that data collection was still a problem, so enter Umi.
7:06 Speaker 02 Yeah, so after Aloha and diffusion policy, I was super excited about imitation learning. But at the time, both of us started still doing kaldi operation, and that just feels super limiting. I think the problem is that, you know, in your setup at a time, like a kaldi up setup, it takes a PhD student a couple of hours to set up in a lab. It pretty much restricts data collection to a lab. But in order for the robot to actually work as a product, it needs to work in the wild, in unseen environments, that requires data to also be collected in the wild. And at the time, I was thinking, okay, is there a way we can collect robotic data without actually using a robot? That like forced me to think, okay, what's the actual most essential part of a robotics data?
7:50 Speaker 02 And after diffusion policy and act, actually, the paradigm is kind of simple. You just need paired observation and action data. In our case, observation is a video clip. The action is the movement of your hand, plus, you know, how the finger moves. I realized that all this information you can get from a GoPro. You can track the movement of a GoPro in space, and you can, you know, track the motion of the, like, of the gripper and also finger through images as well. And that's why I built this Umi gripper, it's 3D printed. At the time, like, the project had three PhD students. Like, we just took the grippers everywhere. Like, I think it was two weeks before the paper deadline. Like, every time I go to a restaurant, before the waiter come in, we just collect some data.
8:33 Speaker 02 And very quickly, we got, you know, I think 1,500 video clips of this, like, espresso cup serving task. And that turns out to be one of the biggest data sets in robotics, and just simply by three people. And that's like how, that's where the power kind of shines. And then with that amount of data, allows to train the first end-to-end model that can actually generalize to unseen environments. So we can push the robot around in Stanford, actually Tony was there as well. Push a robot arm around the Stanford campus, and then anywhere, you know, the robot can serve you a drink.
9:06 Speaker 00 Yeah, I think that is the moment I was like, hey, maybe we should start a company. This is actually working so well. I remember like just falling tongue and.
9:16 Speaker 02 Times it doesn't work well.
9:16 Speaker 00 Instead, yes. I think the only exception I saw was when it's under direct sunlight.
9:22 Speaker 02 Yeah.
9:22 Speaker 00 Right? And I think the reasoning was like over that whole like two, three weeks of data.
9:26 Speaker 02 That two weeks is all raining, so there's no sunlight data. Uh, so like it fails. That also demonstrates the importance of distribution matching. So in order for a robot to work in a sunny environment, it must have seen sunny environments during training data.
9:39 Speaker 01 Yeah, this is really interesting because I remember when I first met you guys, it was like you spent like, I don't know, $200,000 across all of your academic research and yet the scale of data collection as translated to model capability is leading, right? So it's very interesting that, you know, we look at where we are, maybe going back to Tony's point of scaling and massive capital deployment. But that entire paradigm actually wasn't relevant before people realized like you should train on all of the internet data. And we just don't have that in robotics. So the entire field is just blocked on having any scale of data that's relevant.
10:16 Speaker 00 Yeah. I think these days there are still like so many debates about like what is even the right way to scale. There are like world models, there are simulations, there is tally operation, there are like all these new ideas. And I think this is the sort of area that we really want to innovate, that we want to differentiate. They want to find out something that is both high quality and scalable.
10:38 Speaker 01 And then you guys, you decided to start a company, pushing this cart around Stanford. Tell me about that decision and congratulations on the launch and sort of the the direction and team you've built.
10:48 Speaker 00 Yeah, it's a, it's a very interesting journey. Uh, I remember in the beginning, it's actually two of us, uh, in Chen's apartment on his desk. We were like, clamp the robot there and try to do some tasks. And it soon becomes like, I think, an eight-person team towards the end of 2024. And now we're at around like 30 to 40 people. We're not the best at everything, right? But starting a company allows us to find people who we really love working with. And then bring all the expertise together from mechanical engineering, supply chain, like software engineering, like controls, and to build a system together that is not like. A demo, but a real product.
11:27 Speaker 01 You've built this amazing team. What are people actually signing up for? What's the mission of Sunday?
11:31 Speaker 00 Yes, it is to put a home robot in everyone's home. I think there are a lot of AIs trying to make you more efficient during the work, but there is not enough AI that actually helps you with all these mundane things that are not creative, that it really has nothing to do with what's making us more intrinsically human. What's ideal for people to spend more time on is actually with their hobbies, with their passions, as opposed to spending more time doing chores.
11:56 Speaker 01 So if you guys are going from, you know, these amazing research breakthroughs to we're actually going to ship a home robot, you know, and that's a product you have to talk about, cost and capability and robustness, like what's the design philosophy?
12:11 Speaker 00 As these AI models becomes more capable, and as hardware cost continues to go down, the home robots or all kinds of robots will be everywhere. So if we start from the most surface level, which is design of the robots, when we design it, we think about what should the robot look like if it is ubiquitous? You need to see it every single day. What should it look like? And what we end up with is that we really think the robot should have a face. It should have a cute face and it should be very friendly. So in- instead of like a Terminator doing your dishes, we want the robot to feel like it's out of a cartoon movie. And then the huge decision is like, how many arms should the robot have?
12:56 Speaker 00 Should it have like four arms? Should it have like one arms? Should it have legs? Should it have like five fingers, two fingers, three fingers? It's a huge space.
13:03 Speaker 01 Why isn't the obvious answer, it should just be like a full human arm?
13:07 Speaker 00 I think the core motivation for us is how can we build a useful robot as soon as possible? So whenever we see something that we can accelerate it with simplification, we'll go simplify that. So one example of that is the hand that we designed, which has three fingers. We kind of combined the three of the fingers that we have together. And the reasoning there is just that most of the time when we use those fingers, we use it together. Let it be like rasping a handle, let it be opening the dishwasher. So it really doesn't make sense to add the cost like multiplied by three X. To have separated into three, when we can do one with most of the benefits. So this is how we think about the whole robot as well.
13:51 Speaker 00 It's kind of with the constraint that we're building a general purpose robot that can eventually do all your chores. And we'll simplify everything we possibly can, so that the robot can be as low cost and as easy to repair as possible.
14:03 Speaker 02 Yeah, just want to add a little bit more on to the actuator and like mechanical design. Traditionally, most robots are designed for industrial use cases.
14:10 Speaker 01 Uh huh.
14:11 Speaker 02 And the robots are very fast, they are very stiff, and they're precise. Uh, the reason is because all the industrial robots are blind, so they're blindly following a trajectory that's programmed by someone.
14:21 Speaker 01 It's not reaction to perception.
14:23 Speaker 02 Correct, correct.
14:23 Speaker 01 Okay.
14:24 Speaker 02 Um, but because of the breakthrough we had in AI, like now robots have eyes. So it can actually correct its own mechanical and hardware inaccuracies.
14:32 Speaker 01 Uh huh.
14:33 Speaker 02 So that kind of like opened up a new different space of design.
14:37 Speaker 01 And intuitively, it should be like, I can't tell you exactly what the distance is here on a millimeter scale, but I'm gonna get to the cup because I can stop.
14:45 Speaker 02 Yeah, exactly. So that allows us to use these low cost actuators that's achieved, that's compliant, but they're imprecise. But uh, because of the AI's algorithms and the systems we build, it allows us to build robots that's mechanically inherently safe and compliant. While simultaneously be able to achieve the sufficient accuracy we need for the home tasks.
15:05 Speaker 01 Where are we in that timeline? You said we're between GPT and ChatGPT. Um, and so like, when do consumers get ChatGPT and when will you guys ship something?
15:14 Speaker 00 Yeah, it's actually a really exciting time because like we have so many prototypes internally. What we will do next year, twenty twenty six, is actually start doing beta programs. We'll have these robots, all kinds of different ones, into people's home and see how they react to it. That will be when we learn the most about like how people like people want to talk to their robots. Do people want to have their robots uh maybe teach their kids some uh new knowledge about the world? And this will inform us what the eventual product should look like. Internally, we just have an extremely high standard of what is the minimal consumer product we want to ship. It needs to be extremely safe, it needs to be extremely capable, and low cost.
15:58 Speaker 01 Do you feel like you know something now that you didn't when you started the company?
16:02 Speaker 00 Absolutely. So I think at the beginning, I would describe it as like we see light at the end of the tunnel. Of there are two axes, there's sexuality, there's generalization. When we add more data, things work better. And what this company about is the cross product of these two. How can we scale and have both sexuality and generalization? And this is something we're able to show in our generalization demo, which is like we can pick up these like very precise like forks, like actual metallic forks on a ceramic plates with very high success rates. And honestly, this is not something that like we thought that would work so easily just by having so much more data.
16:42 Speaker 02 Yeah, so actually, just want to expand a little bit, you know, it's actually the process was long and painful. So, you know, there are so many issues, like just scaling up a system, a robotic system is very, very hard. There are mechanical issues, like reliability issues, there's like data, you know, quality issues that come out of it. In the beginning, I actually thought it's gonna be much easier than this. But really, just it takes time and effort to grind out all the little details for this to work. I also, I feel compared to Teddy Up is much harder to get this system scaled up. But once it's scaled up, it's very powerful, very repeatable.
17:18 Speaker 01 So it is both harder than you thought it would be to get to here, and you are further than you thought you would be?
17:23 Speaker 00 Yes. And I remember in the beginning, we're having this like funny conversation of we're like, if we build this, someone can just like take our glove and they'll build the same thing, like what mode do we have? Are we worried about that? And at the beginning, actually, we were a little bit worried because we thought like, oh, you know, they can probably just replicate it. But as we go along the path, it turns out things are so much harder than we thought it was.
17:48 Speaker 02 There's so many the small details.
17:50 Speaker 00 Yeah.
17:50 Speaker 02 Yes.
17:51 Speaker 01 And when you say scaling up the robotic system, you mean the data collection to training pipeline and the hardware itself?
17:57 Speaker 02 Yeah, so actually, like for yeah, for this to work at all, you need the data collection system.
18:02 Speaker 01 Yeah.
18:02 Speaker 02 You need the robotic and control system to be able to deliver the hand to where we want to go.
18:06 Speaker 01 Yeah.
18:06 Speaker 02 You also need the data. Focusing pipeline and data cleaning pipeline and the training pipeline. And all these things need to be iterated together. So actually gone through several loop of these. It's kind of hard to imagine without having a full stack team in the house, how this can even be done.
18:21 Speaker 00 Yeah. The glove we're using right now is we call it like V5. And for V0 to V5, each version has like around 20 iterations.
18:30 Speaker 01 Okay.
18:31 Speaker 00 And.
18:32 Speaker 01 So 100.
18:33 Speaker 00 Yes.
18:33 Speaker 01 Yes.
18:33 Speaker 00 Um, and also like just when you make these at scale, right now we have more than 500 people using these gloves in the wild. Like all the things that could go wrong will go wrong. For example.
18:45 Speaker 02 They did, they did, yes.
18:48 Speaker 00 For example, like how things are assembled. If you don't specify exactly how it should be done, people will assemble it in creative ways. And the creativity doesn't help us here because I really want the data collection device to be extremely precise.
19:02 Speaker 01 So you guys can't obviously know everything that's happening in every company in academia industry. But from what you know, how would you compare the scale of um training data that you have today relative to the industry?
19:16 Speaker 00 At this point, we are almost 10 million trajectories being collected in the wild. And those trajectories are not just like, oh, pick up a cup. It's these long trajectories with like walking, with navigation, and then like doing these long horizon tasks.
19:34 Speaker 01 Tony, as you mentioned, like it's an open question, actually, what the right way to scale data up is. And so there are strong theories around tele up, around like pure RL, um around video and world models. Like, how did you think about all of these?
19:50 Speaker 02 Yeah, so from our perspective, actually, it's kind of somewhat surprising. So in the beginning, we worried that, you know, the data from glove or um like data that has higher quantity, but lower quality compared to tele up. Because for tele up, you're using exactly the same hardware and software stack between training and testing. It's perfectly distribution match. But what we realized is actually, you know, this glove form factor encourage people to do more adventurous and more natural movement. And those actually result in a like a more intelligent behavior. On the on the on the modeling side. And in terms of, you know, data quality, we don't really see a difference in terms of, you know, how much like like there's a gap between Teddy Up and Glove data.
20:28 Speaker 00 After we did the twenty engineering like.
20:32 Speaker 02 Yeah.
20:33 Speaker 00 Yeah, like, cause like apparently there is a mismatch, right? That's um in the camera frame, there's a human inside of the robot. And there are just a lot of things that we need to do to kind of convert a human data one to one to like as if it is robot data. And have the model not being able to tell the difference.
20:52 Speaker 02 Yeah, and that kind of relies on again, the whole, the full cycle iteration between hardware and software.
20:57 Speaker 00 What about RL?
20:59 Speaker 02 We see a lot of great promise for RL in locomotion. And, uh, you know, we think that will continue to be true for locomotion. Uh, so what we see really like RL is a method, uh, is very powerful. Uh, but it is much less sample efficient compared to imitation learning. And we see that to work great in environments where it's easy to simulate. Uh, for the case of locomotion, only you only need to worry about uh rigid body dynamics and rigid body contact between the robot and the ground. Uh, and, you know, because you engineer a robot, you know, everything. Uh, but for manipulation, it's hard, kind of hard for us to imagine. Like have this actually the same amount of diversity and the distribution of real object. In terms of matching both appearance, uh, it's it's uh appearance and physical properties.
21:44 Speaker 02 And we think that uh it's gonna be challenging compared to uh Glove data collection and Teddy Up.
21:49 Speaker 00 Yeah, I think it's really about which method can get us there faster. There might be different methods that will eventually get there. For example, like, you know, simulation world model, right? And like, it's almost a tautology to say that if I have a perfect world simulator. Anything can be done there, like as long as you can do it in the real world, you can do it in a simulation. And you can like, you know, cure cancer in a simulator, right? But what it turns out for robotics is that there's some some things are harder than others. And it really depends on the problem itself. So in the case of locomotion, as I mentioned, all we need to model. In a simulator are point contacts with a somewhat uh flat ground.
22:33 Speaker 01 Um hum, like feet.
22:34 Speaker 00 Yes, but sort of the behavior we want out of it is actually very difficult to model.
22:39 Speaker 01 Um.
22:39 Speaker 00 Like it's all these reactive behaviors that when you feel like your um your leg is hitting something, you should like retract and you know step again. Uh these are very, very hard to describe or try to learn from demonstrations directly. But in the case of manipulation, I think the difficulty is flipped. That it's a lot easier to capture the behavior itself and it's a lot harder to simulate the world.
23:06 Speaker 01 Um.
23:06 Speaker 00 For example, if you were to grasp a transparent cup with some orange juice in it. It's ridiculously hard to simulate how like your hand deforms around the cup. And how all those ripplings, how those like the color of the juice results in like the rendering and what the policy end up seeing. Simulating that is very expensive and difficult. But all we need to learn is just to like get your hand to be in front of the cup and then close with the appropriate amount of force. And that's actually very easy to learn. That's why like we see so many success of imitation learning in the case of robotics.
23:47 Speaker 00 Manipulation is because the behavior itself is actually not as hard as simulating the world. And that's why we see faster progress there.
23:58 Speaker 01 Is there anything that you have changed your point of view on um in data over the last year?
24:03 Speaker 02 Uh like one thing, I wouldn't say changed, but just uh data quality really matters. Uh like I think we know, I always knew data quality matters, but once you scale it up, like it it really matters. Uh and and then because you know, just like the diversity of behavior that you experience in the wild. Uh it's very hard to control and the hardware failures are hard to control. You need to constantly monitor them. Uh just you need to spend a lot of huge amount of engineering effort just to make sure that you know the data is clean.
24:32 Speaker 00 Yeah, and also building all those automatic processes, right? We have our own way of calibrating the glove before we ship it out. And we have this whole like software system to catch if a something is broken on the glove and we can detect it automatically. It's like the importance of uh data quality kind of translates to all these repeatable processes. And we don't need a human to be staring at the data to know that something is wrong.
24:55 Speaker 01 When you described the beta for next year, a lot of it sounded like, you know, we just want to understand behavior, like how people actually want to use it. We can make some design decisions for the actual product. What technical challenges do you still see?
25:11 Speaker 02 So to me, I think there's like two kind. The number one is really figuring out the training recipe at scale. We, as a field, just entered the realm of scaling, and we just got the amount of data that we need. I think now is a perfect time to start to research and to figure out what exact training recipe we need to actually get robust behaviors. And I think we're in a unique position because like the amount of data and the entire pipeline we built around data. The second point, I think, is just really hardware is hard. We're still pushing the envelope, performance envelope of hardware. It's not really clear actually what is needed, what is needed, what is necessary for the hardware to be reliable. Because like, whenever the mechanical team build a hardware, the learning team will try harder to push it against the boundary,
25:56 Speaker 02 and then it'll break at some point. But I think what's interesting in this company is that everybody's under the same roof. So immediately after something breaks, it goes straight back into mechanical design, and we have another iteration, like I said, for the hand parts very quickly. Hardware is hard, but it is important. And I think, you know, it's a hard but right thing to do. And I think we, as a field, shouldn't avoid doing the hard things just because they're hard.
26:22 Speaker 00 Yeah, I want to echo Chen's point about like first the research. I think when there is data scarcity, it is really easy to come up with like cute fancy research ideas that doesn't end up scaling very well. And this is why, like, when we build a company, we actually focus on the infrastructure and a scalable data pipeline and operations before we start to like really dive into research, which we only started to do like three months ago. I think we really want to avoid doing research that doesn't scale. We want to focus on things that contribute to the final product. The second point is like, I think robotics is so intrinsically a...
27:00 Speaker 00 Like a system that right now, we don't like, there's not a existing general purpose home robot out there. And we don't really know what the interface of different system is, like what is even good. And in that case, if you're working with a partner, it's actually really hard for them to understand your standard of good, because your standard of good is changing all the time.
27:24 Speaker 01 Uh huh.
27:25 Speaker 00 This is why we are like building everything in-house in a more full stack approach. It would build our own data collection device that is co-designed with a robot. We build our own like operation team to be like, how can we most efficiently get the most high quality data out? And of course, our own AI training team, um, that's make the best use of these data. I think these are the things that are really not easy. It makes the company a lot harder like to build that right now you suddenly need like so many teams and they need to orchestrate together, but we believe it is the right thing to do.
27:59 Speaker 01 Okay, I'm gonna ask you a few questions that are uncomfortable guesses now, but when will people be able to buy robots commercially for the home?
28:06 Speaker 00 Like this is something we're really excited about because we have so many of prototype robots in our office and really want to get it out there. So the next step of our plan is to have a beta group program 2026. And what it means is that for people who sign up that we selected, they will have a real robot in their home and it will start doing chores for you. And uh, it's a, it's going to be a really interesting learning lesson for us because we will see like how human interact with the robots. We'll see like what kind of things people just really want the robot to do. I think this will be before we actually ship it to the masses, because we just have an incredibly high standard of what we are willing to ship as a, for a consumer experience standpoint.
28:51 Speaker 00 We want the robot to be highly reliable, want it to be capable, want it to be cheap. I think it really depends on the results of the beta program that will decide when is a good time to ship it. Is it 2027? Is it 2028? And but all of those are possible.
29:06 Speaker 01 But it's not a decade away.
29:08 Speaker 00 No, it's definitely not a decade away.
29:10 Speaker 01 How much do you think it could cost?
29:11 Speaker 00 Right now, the prototype robots we have in-house, I think the cost ranges from like $6,000 to something like $20,000. And this is actually pretty interesting that the big difference here is not like, oh, we find a better actuator. They're using the same actuators that are like very low cost. But actually, it's the cladding of the robots. When you're trying to make them at low scale, it's just really expensive. Like the claddings are like a few thousand dollars to make. But this is the type of things that as we scale up, it becomes like dirt cheap. Because instead of like doing CNC, instead of hand painting them, it'll become injection molding. What we see is that as we get the scale to a few thousand units,
29:55 Speaker 00 we can drastically reduce the material cost, likely under 10K. And what it implies is that when we sell the robots, the price will be somewhere around it.
30:07 Speaker 01 Okay, so you fast forward two, three years out. If you look like five years and beyond, the home robots are ubiquitous. Like, what does life look like? How does it change for your average person?
30:18 Speaker 00 This is a different answer for everyone. For me, like, I just really hate dishes. Like, in my sink, there's always like four or five dishes that are like somewhat dirty out there that kind of stinks a little bit. And after a long day of work, it really doesn't feel good to come and like see a home like that. So I think the world we'll live in is
30:40 Speaker 01 It's gonna be cleaner.
30:41 Speaker 00 It's gonna be cleaner. And I was just thinking about it as like the marginal cost of labor in homes goes to zero.
30:47 Speaker 01 The last thing I want to make sure we do is like talk about demos, right? There's a lot of robotics launch videos today. It's been years since you saw an optimist serving drinks at a bar. Why are those not available? And what is actually hard?
31:03 Speaker 00 I I think the way I will put it is make zero assumptions, no prior.
31:08 Speaker 01 Okay.
31:08 Speaker 00 As in, if you see a robot handing one drink to one person, first, ask the question of, is that autonomous or is that teleoperated? So this is the first thing. So we should look at the tweet and see what they say about it. And and then is that, does it show giving another slightly different colored cup to the same person or not? If they didn't show it, it means that a robot can literally only pick up that single cup and give it to that same person. When we look at demos, we tend to put our human instinct into it. They're like, oh, if you can hand a cup to that person, it must be able to hand a different cup to another person. Maybe you can also do our dishes, maybe you can do laundry. There are a lot of like visual thinking that we can have about it, which is what's great about robotics, that there are a lot of imaginations.
31:53 Speaker 00 But I think when we look at demos, only index on things that is strong. And that's likely the full scope of that task.
32:04 Speaker 02 I think another aspect is, at least me as a review researcher, I uh I appreciate the number of interactions that happens uh in in a demo. So usually, the more interactions you have, like every interaction, there's a chance of failure. So the the longer the sequence is, the harder it actually is. So that's something we really emphasize here. Uh and and that's actually somewhat uniquely easy for us because the glove way of data collection is so intuitive after people.
32:31 Speaker 00 Yeah, it's really about like generalization and reliability.
32:34 Speaker 01 So can you explain the demos that you guys are showing?
32:36 Speaker 00 Yeah, of course. So we're sharing like basically three categories of demos. The first one, as you saw, is we have this whole like messy table. And what the robot does is to clean up the whole table and, you know, dump the food into the food waste bin and load the dishes into the dishwasher and then operate the dishwasher. What makes this? Demo really hard is that it is a mix of really fine-grained manipulation with these super long horizon, full-range tasks, as in like need to go up and it also need to go down very much. It's a mobile manipulation task. Exactly. The reason that we can show this is just how nimble and easy for us to collect these data sets,
33:18 Speaker 00 to make Horizon Dexter demo possible. And it's also about the forces as well. So you might have seen like, we're trying to pick up two wine glasses with one hand and I struggle with this, but yeah. It's actually really hard. And because it's like transparent objects, we need to also load it very precisely into the, into the dishwasher. A lot of it is about how much force you apply. Uh, because if you're trying to grasp two in one hand, if you squeeze a little bit harder, you're going to break one of the glass. And when you load it into a dishwasher, if you're pushing it in the wrong direction and it hits something, it's going to shatter. We, we did shatter a ton of glasses when we were like experimenting with it. So these are tasks are like really high stake that
34:00 Speaker 00 it's not just about recovering from mistakes, but about not making those mistakes in the first place. And this is what's generally the case in a lot of the home tasks, that you're just not allowed to make any mistakes. And then we get into the generalization demos, which we basically show our robot, we book like six AMB and Bs, and we get it there, zero shots. And see if can do like part of the task. So two tasks we use, one is I go around table and collect all the utensils into the caddy. The other is to grasp a plate and then load it into the dishwasher. What makes these, these demos very interesting is that we don't need any data when we enter that home.
34:40 Speaker 00 It's pure generalization. And this is as close to getting like a real product as you can get. Because when someone buy, buy our home robot, we really don't want them to like collect a huge data set themselves and just to like unbox it. Also, in addition to the generalization, those two tasks are also really precise. We're using the exact silver bears uh in the home and Need like basically like a few millimeter of precision, we have to grasp it properly. Those forks are also hard to perceive because they're reflective, like the lights looks weird on it. We have have a transparent table home. I think that the table looks like nothing and the robot still like reacts very well to it. And again, the reason that we can do it is because we have all these like more than five hundred people.
35:24 Speaker 00 And we've seen so many glass tables in that data set, so the robot is able to do it. I think the last bit of the task that we did is kind of pushing what's possible in terms of dexterity. The two tasks we chose, one is an espresso, operating espresso machine. The other is like folding socks.
35:44 Speaker 01 Uh huh.
35:45 Speaker 00 What makes these hard is that they require very fine-grained force. That is hard to get if you're doing with teleoperation. Because these days, there's not a good teleoperation system. That can let you feel how much force the robot is feeling. So basically when you're teleoperating, your hand is numb. And sometimes you're applying like a huge amount of force on the robot, but you don't know it. And that can result in like very low data quality that robots also doing in that aggressive way that we really want to avoid for our system. Uh, the sock is a very good example that when you're trying to fold it, your two fingers can touch. And that forms a, what we call like a force closure, you have a closed loop uh for the force.
36:30 Speaker 00 And if your robot is stiff, you can apply infinite amount of force at it and doesn't look like anything. Uh, but for us, because we're using the glove to collect the data, uh, the human who is collecting it can just naturally feel it is very intuitive. I think we're the first to do the sock folding. And using end to end to do like espresso machine uh out of the whole industry.
36:53 Speaker 01 One of the things that you will also need to scale as you guys, um, you know, scale up the company is the team, um. What are you hiring for? What are you looking for?
37:03 Speaker 02 One thing I'm really looking forward is
37:06 Speaker 01 Chang speaks up.
37:08 Speaker 02 Yeah, so it's full stack roboticists and people who aspire to become full stack roboticists. Really, what you learn in this company is just that robotics is such a multidisciplinary field. You need to know a little mechanical, a little electrical, a little code, a little data to actually fully optimize the system. And we have a couple of examples of training, you know, just full stack software engineers to become roboticists, training I'm engineers to become roboticists. And so if you want to learn about robotics, you want to learn the whole thing and not just to be boxing into your small, you know, little cubicle, let us know.
37:45 Speaker 01 And you told me that you didn't write code until you got to college or something.
37:50 Speaker 02 Yeah, I was super enthusiastic about robotics. I was mostly doing like a mechanical and like for design before that. And then I realized, okay, the bottleneck is actually how the robot will move. And there's like, there's something called like programming. And then the more I get into it, like the deeper it gets. And then toward the end of college, I realized, okay, there's a thing called machine learning. And you figure out how to train models. I think this thing just goes on and on. I think it's very natural for me to gradually expand my skill set because I'm always looking forward to build a robot.
38:21 Speaker 01 Well, I hope you discover the next field because you're no longer doing dishes.
38:25 Speaker 00 It's a very fun place to work. Whatever you can imagine about robotics and consumer products and machine learning, you can find it here because we're just fundamentally such a full stack company. We're not just about the software. We're not just about the hardware, but we're about the whole experience, the whole product and making sure that product is general and like scalable in the future.
38:47 Speaker 01 Awesome. Congratulations.
38:49 Speaker 00 It's really exciting.
38:58 Speaker 01 Find us on Twitter at NoPriorsPod. Subscribe to our YouTube channel if you want to see our faces. Follow the show on Apple Podcasts, Spotify, or wherever you listen. That way you get a new episode every week. And sign up for emails or find transcripts for every episode at no-priors.com.
Previous How AI Will Accelerate Breakthroughs in Biotechnology with Benchling CEO Sajith Wickramasekara Next Scaling Legal AI and Building Next-Generation Law Firms with Harvey Co-Founder and President Gabe Pereyra Page updated October 12, 2026 · transcript and summary are machine-generated
