Free with a sign-in.
The best 12 min of 38
No Priors Ep. 102 | With The Bot Company CEO Kyle Vogt
Kyle Vogt says home robots can return time spent on chores, and that the technology is ready for serious development.
No Priors · Sarah Guo & Elad Gil
- Kyle Vogt is cofounder of Twitch and Cruise, and CEO of The Bot Company.
The best minutes
- 10:03Home robots can return time lost to choresPro pick 5 min
- 21:13Robot design should match what the product can doPro pick 4 min
- 33:56Small teams can still take on ambitious company goals3 min
Every moment plays free. PRO marks a write-up that's in Pro.
Also worth watching
- 0:44Cruise shifted from retrofits to robotaxis as demand emerged3 min
- 15:29Home robots can launch with narrower functionality than robotaxis3 min
- 17:50Founders need to adapt beyond their original research3 min
- 26:43US robotics firms can compete with global manufacturing3 min
- 29:14Phased regulation can create a path to robotics scale5 min
Skip Tesla/Waymo and AV hardware, launch timing and China, the home-robot timeline, and Vogt's closing account of his drive.
Full analysis · 5 min read
Real Alpha · Nov 8
Kyle Vogt says home robots could reclaim time from a 24-hour day
- Vogt says a 24-hour day leaves little personal time after sleep and work.
- Learning methods and narrow tasks may help, but he gives no home-robot performance data.
- Builders should test a useful chore, then measure assistance, safety, and repeat use.
Free with an account
Read the rest of the analysis
7 sections, the 5 other voices it tests Vogt against, and what it means for you. Every episode we cut gets one.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Signed up before? The same buttons sign you in. The episode, its moments and the transcript stay free.
Chores make home robots' promise easy to grasp
Vogt's arithmetic is memorable. If people sleep eight hours and work eight to ten, he says, few hours remain that feel like their own.
He lists making the bed, doing dishes, folding laundry, and picking up toys after children. He calls those chores unskilled labor, saying they do not make us human.
For example, dishes and laundry make the promised benefit concrete enough to measure at home. But Vogt gives no estimate of hours returned, chore frequency, or willingness to pay.
“There's only 24 hours in a day.”
Our read
The benefit is intuitive, but the episode offers no estimate of time a robot could return. A first product case needs a chore people value enough to delegate.
Learning may help robots handle messy homes
The difficulty is not merely recognizing an object. Vogt describes homes as unstructured, with different layouts and objects, and routines that change.
He points to end-to-end learning, imitation learning, and reinforcement learning. He says robots could learn from teleoperated demonstrations, human data, or internet videos.
He also says people could speak to a robot instead of typing into an app or keyboard. As The Bot Company's CEO, he offers a founder's timing thesis, not deployment results. The episode gives no completion rate, intervention rate, or price.
Our read
So the technical rationale supports a build-now thesis, not a buy-now case. Teams still need proof of reliable task completion in real homes.
Home robots can start with narrower tasks
Vogt says a robotaxi has no useful product until it reaches very high safety performance. A home robot can start with constrained tasks, making a partial product possible sooner.
One approach he has seen is a fully teleoperated humanoid. He floated an illustrative price, hedged as "I don't know, like something crazy, like $50,000 and $1,000 a month."
At CES, he saw a little robot with a tiny hand. It could pick up a sock, he said. For instance, this example shows a bounded task, not broad household competence. These were other companies' examples, not products he identified as his own.
Our read
Yet a narrow launch only works if the task is worth paying for. The product also needs a plan for human assistance and clear limits.
Robot design can keep expectations in bounds
Vogt says a robot's appearance can raise expectations beyond its actual abilities. In his view, a humanlike design risks promising broad competence before the system can deliver it.
He recalled, without citing a survey, that like 75 or 80 percent were skeptical of self-driving cars before a ride. After one ride, he recalled the figure falling to like 20 or 30 percent.
That is a loose recollection about cars, not home-robot adoption data. His broader point is that firsthand use and trusted recommendations can beat specs and conventional marketing.
“the more human-like you make it, the higher the expectations I think the average person will have for that machine”
Our read
Still, a compelling demo is only useful if the delivered product matches it. A self-driving ride does not establish that people will trust a robot in their homes.
Vogt sees product experience offsetting cost pressure
Vogt says U.S. firms can use contract manufacturers abroad, though doing so takes travel and relationship-building. He sees pure mechanical engineering as hard to defend on price.
His proposed edge is software complexity and product experience, which he says are harder to copy. But he gives no target build cost, service cost, gross margin, or production volume for The Bot Company.
A later outside example comes from Bernt Børnich on Relentless in September 2026. Børnich described about four weeks between major CAD changes and a robot leaving the line. That is a separate production example, not evidence of Vogt's cost position.
Our read
Still, the comparison exposes the unanswered operating question of iteration speed and cost. Without unit costs, product differentiation remains a thesis, not a margin case.
Vogt says phased rules can enable home deployment
Vogt says home robots lack clear cybersecurity rules and a safety framework suited to household use. He flags cameras and microphones that may send private data somewhere users cannot identify.
He also says industrial-robot practices do not fit homes neatly, while consumer-product guidance leaves gaps. His preferred response is phased rules that open operations as systems meet maturity levels.
This is a practical case for predictable gates, but he names no regulator, test protocol, or liability standard. Max Tegmark said this on Into the Impossible With Brian Keating in October 2026. He called for safety proof before dangerous AI products are released.
That is not a phased robotics plan, so it is a contrast, not corroboration. And absent clear gates, teams cannot plan what evidence enables broader use.
Our read
A predictable path could help builders plan testing and deployment, while addressing risks for households. The episode does not specify the standards or enforcement Vogt wants.
Vogt expects coding tools to widen small teams
After Cruise, Vogt says he wants fewer management layers and a small team with ambitious goals. He says coding assistants let good engineers take on work once assigned to specialists.
His examples include iOS development and low-level Rust for motor drivers. The episode gives no headcount target or measured productivity gain. Physical testing still needs people, equipment, and time.
Later episodes offer comparisons, not proof for The Bot Company. Sam Altman said on AI & I in September 2026 that OpenAI is betting on small-team companies.
Dmitri Dolgov said on Minus One in October 2026 that AI speeds simple builds, while safety-critical systems resist commoditization. Gorkem Yurtseven said on Sourcery in November 2026 that fal's deepest reporting structure has two levels.
Now the practical test is validated task progress per team member, not code volume. Hardware and safety work may resist the same compression as ordinary coding.
Sam Altman · AI & I · Sep 30OpenAI is betting on a wave of small-team companies
Dmitri Dolgov · Minus One · Oct 8AI will speed simple builds, while safety-critical work stays hard
Gorkem Yurtseven · Sourcery · Nov 3fal keeps management flat with hands-on team leads
Our read
Coding tools can widen a team's reach, but robotics couples software iteration to hardware and safety validation. The board-level measure is validated task progress per team member, not code output.
What it means for you
- Choose the first chore. Test a narrow task such as picking up toys or a sock. Track setup time, successful completion, human intervention, and cleanup.
- Separate supervision costs. Vogt discussed remote help for robotaxis, not the cost of assisting household robots. Build scenarios for intervention frequency and labor cost before projecting unit economics.
- Measure the trust gap. Show prospective users a robot's form and ask what they expect it to do. Compare those expectations with its actual task limits.
- Define safety gates. Inventory cameras, microphones, data flows, and physical failure modes. Decide what evidence would justify expanding the robot's operating range.
What to watch
- Whether a first product can complete a narrow chore without frequent human intervention.
- How Vogt's low-cost goal holds up against hardware, service, and safety costs.
- Whether clear cybersecurity and safety requirements emerge for home robots.
Send this to the robotics product lead choosing a first household task.
Watch the minute where Vogt frames chores against a 24-hour day
Written by Real Alpha from the episode's transcript and the other moments it links, then checked against the transcript. Quotes are the speaker's exact words.
Minute by minute
A Pro write-up, open to you here. Try Pro free for 7 days
A Pro write-up. The minute by minute is in Real Alpha Pro; the episode, its moments and the transcript stay free.
Read it with Pro, 7 days free- 0:35Cruise began with a lean approach
Vogt says he started Cruise in 2013 with a retrofit system that could make a regular car drive, rather than trying to build a complete self-driving car from the outset. After pitching about 120 investors, he shifted toward robotaxis as Uber and Lyft’s growth created demand for a way to reduce driver costs.
- 3:08Vogt backs Tesla’s business model
Vogt says Tesla made money while other self-driving efforts spent billions, and describes camera-heavy and sensor-heavy systems as different routes toward low-cost autonomy. He argues that newer models can extract depth from camera images, while automotive compute and reliable connectivity remain constraints; he sees commodity sensors and potentially satellite connectivity as part of the path forward.
- 8:09Remote assistance can work economically
Vogt says that, based on what he has seen, Chinese autonomous vehicle deployments still appear to use substantial teleoperation. He argues that even a 1:4 ratio of remote operators to vehicles would cut labor costs by 75%, and that moving to 20:1 or 50:1 would add only a few more points of margin.
- 10:03Home robots can reclaim time
Vogt says he chose home robotics because it combines impact, hard technical problems and products he wants to build. He sees chores such as dishes, laundry and picking up toys as work robots could take over. He says homes are unstructured and change constantly, but techniques such as imitation learning and teleoperation can help robots learn tasks and respond to natural-language instructions.
- 14:59Home robots can launch sooner
Vogt says a driverless taxi needs very high safety performance before it can become a product, while home robots can launch sooner with narrower functions or constraints. He describes two routes: expensive, fully teleoperated robots, or simpler products that perform small useful tasks, such as picking up a sock.
- 17:50Strong teams can outlast a bubble
Vogt expects a funding bubble and shakeout in robotics, but says the strongest companies can keep going through a downturn. He warns that some founders lack product focus or are too committed to commercializing one research approach. He points to Cruise, Aurora and Zoox as startups from the self-driving wave that endured in different ways.
- 20:45Robot design shapes expectations
Vogt says a robot’s appearance can make people expect human-like abilities it may not have, so its form should match what it can do. He says skepticism about self-driving cars fell from roughly 75–80% before a ride to 20–30% afterward, and argues that trying a product and hearing from trusted users is more persuasive than marketing. He says the home-robot timeline could range from one to 20 years because of unknowns.
- 26:43US hardware firms can compete
Vogt says US companies can use contract manufacturers and existing supply bases abroad, though doing so takes travel and relationship-building. He argues that companies doing only mechanical engineering may struggle on cost, but complex technology and product experience are harder to copy. He says Chinese self-driving companies remain years behind the leading US companies, based on his experience.
- 29:14Regulation can enable robotics
Vogt argues that clear rules and liability protections could help autonomous vehicle companies operate, and says home robots also need attention to safety and cybersecurity. He proposes phased regulatory paths that let companies expand as they demonstrate maturity, rather than facing either no oversight or a barrier to innovation.
- 33:56Vogt favors smaller teams
Vogt says GM’s priorities shifted away from Cruise, and he does not want to sell another company. He says layered management created bureaucracy and communication gaps, so he plans to keep The Bot Company small and hire exceptional people. He argues that coding assistants and LLMs let a few engineers take on work that once required specialists.
- 36:47Vogt follows ideas to completion
Vogt says that when an idea feels timely, he struggles to focus on anything else until he has pursued it. He finds motivation in taking on difficult problems, making progress each day and building something complicated over time.
Published Feb 20, 2025.
Transcript · 38 min read
No Priors, hosted by Sarah Guo and Elad Gil, with Kyle Vogt.
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
Hi, listeners, and welcome back to KnowPriors. Today, we're joined by Kyle Vogt, a serial entrepreneur who has helped build some of the most influential tech companies. He co-founded Twitch, shaping live streaming, Cruise, the autonomous vehicle company acquired by GM for a billion dollars, and now Kyle has launched the Bot Company, a startup focused on building consumer robots. Kyle, welcome to KnowPriors. Awesome. Let's get going. Obviously, you've done a variety of different things over time, everything from co-founding Twitch, you started Cruise, you're now working on a new startup. Can you tell us a little bit more about your Cruise experience?
Because I think that whole era was incredibly formative for everything that's happening today in AI. And we'd just love to get your perspective on why start Cruise when you did, how that all evolved, and then how that's informing what you're doing now. Sure. Yeah, we can go back to the beginning. This is 2013.
“And so suddenly there was like a strong market pull for self-driving technology, whereas before I had been seen as just kind of a cool sci-fi thing.”
Read the whole transcript
do you think about the different approaches that people are taking today? So there's Tesla on one side with a very specific approach, kind of moving everything to things that are a bit more camera eccentric, but training on sort of a richer set of sensors and approaches. There's sort of the Waymo approach, which is much heavier on the hardware side in terms of what's actually on the vehicle. Both seem to be doing very interesting things. One is robo-taxis, one is kind of still building mainly cars. How do you think about the different approaches, both from a business model perspective, but also from a technology perspective? To be fair, Elon nailed it from a business model perspective. He's been making billions of dollars of profit while developing self-driving cars, whereas everyone else has been burning billions of dollars to try to get to basically the same point. I think in the end, when you start with custom vehicles with lots of sensors,
really expensive, and you make it work in a constrained environment, and Tesla starts with unconstrained environment, low-cost sensors, but it doesn't quite work without a driver, they're all trying to get to the same spot in the end, which is low-cost, works everywhere, you know, commodity sensors. Different paths to get there. I think Elon won that hands down. What do you think of the criticism that you can't get there? You can't get to full self-driving? from like mostly self-driving. That statement is almost certainly wrong given a long enough time span. Again, going back to what Elon's approach was, he doesn't have to like finish by a certain date or run out of money.
He's like making money along the way. And so, you know, I think the only risk is that, you know, customers get fed up and sort of rage quit his program, but they're getting something they like along the way, Tesla full self-driving. So I think that's the right approach. But in 2013, for sure, in 2015, even 2018, it really wasn't viable to have a full driverless car that just use cameras and low-cost sensors. It just wasn't. The technology was not there. I think now, if you take a fresh look at where we are today with large language models and that, you know, models and other things that that sort of class of technology applied to the classical challenges of perception for autonomous driving even motion planning for autonomous driving completely changed the game in terms of the magnitude of compute that you need the expense of that um and i think now from cameras as long as you have sufficient redundancy and low light sensitivity and you know some robustness there uh you can extract from a single camera image not even stereo you can get beautiful depth data really accurate and those models are getting better every day and so you if you're making a bet on the right technical approach in 2025, it does not involve a bunch
of expensive LIDARs or exotic sensors. It involves the most commodity, a sort of, you know, high volume, readily available sensors you can get, and probably just several more of them than you would find on a typical driver assistance system. So I think that's the path from here on out. Is there anything else you think is lacking from a technology perspective, either in terms of hardware or just scaling models or, as you know, better than anyone, everybody started moving towards end-to-end deep learning over the last, you know, year or two. And that's really made a big difference. But is it just scaling that up or is something else lacking?
End approach will be an end-to-end type model. I mean, it's sort of hard to put it in a bucket of end-to-end or smaller models because there's such a spectrum in between and everything I've seen is a mix and match of various technologies. If I look at the limiting factors, at least on the hardware side, at least previously, it's been hard to get high performance compute in an automotive, you know, high temperature range, safety critical environment. And so, you know, crews made custom chips, I'm sure Waymo makes custom chips, and sort of piecing together things from the supply chain to solve that is a little challenging. And so there is room for more high performance compute automotive silicon. And I've seen some things happening in that space. That's one. And then I'd say the other piece, you know, most robo taxi deployments I've seen rely on some form of remote assistance. And so there's a question of like, how you get reliable connectivity to a vehicle from anywhere using multiple cell networks, like what Cruise has done and Waymo, I'm sure others, works, provided you have cell phone coverage. I think the missing piece may be Starlink or something similar where you can have always on connectivity between Starlink and maybe
a cell phone and some other fallback. I think that really opens up the opportunity in the number of places you can deploy robotaxis, whereas before it's sort of an open question, what you do if you're driving down Highway 1 in California and there's no cell coverage there, like, should you still have an av on that road given you know if there's an issue or you know a customer needs some help you like literally can't get in touch with them in retrospect is there was there like a right year to start a self-driving car company You were so ahead of the ball on this. It takes a long time to sort of spin up automotive pipeline and everything.
So probably like, you know, circa 2020 or so would be the right time to get started. And then, you know, around now, I think you'd be having like a combination of hardware and software that's mature enough. If you have a nimble enough engineering team that's able to adopt new technologies when they pop up and quickly pull them into the pipeline, you're actually well positioned. and even if you started a while ago and your tech stack was based on some older technologies, if you have all the infrastructure in place for validation and testing and training models and deploying on public roads with test drivers, I think you can go a lot faster, even if you have to like sort of rip out and change some of your tech stack to adapt with the times.
One thing I've heard two opposing viewpoints on is the autonomous vehicle market in China. And one point of view is, well, it's not that real and it's mainly tele-op and it's a little bit more sizzle than steak. And then the other opposing view is, well, actually they've advanced dramatically really rapidly. There's fewer safety constraints so you can do more, try more, et cetera. And that models and approaches that are at least a parity with the leading contenders in the Western world. Which of those two views do you subscribe to or how do you think that market will evolve? From what I've seen so far, and I don't have a lot of inside information just from what I've seen videos online and other thing, it does still seem like there's a lot of teleoperation.
Like I think even someone like Tesla may start off with like a one-to-one ratio of remote operators to vehicles. Cruise and Waymo probably started off pretty close to that. And then I think over time moved to a smaller ratio. And so I think in the interim to like get the deployment numbers up to get more experience to accelerate data collection, people are brute forcing it, which means there's probably a lot of remote operation. I actually think that's fine because it doesn't take much. Once you get to like, you know, 50% remote assistance or even 25%, at 25% remote assistance, you've already reduced the labor costs, you know, 75%. And so you're almost already at the diminishing returns point. And it sounds on one hand, kind of crazy to say like, oh, if there's a, you know, 1000 AVs out there, then maybe 250 people monitoring them. But that's actually not crazy from a cost from a unit economic standpoint, it actually makes a ton of sense. And a ratio of one to four is like, trivial, I think, with today's technology.
Over time, you could get to 20 to 1 or 50 to 1, but, you know, you're just talking single-digit points of margin at that point. The real benefit is just going, you know, anywhere less than one full human in a car makes the economics of this really good and I think, you know, puts you on a pathway towards, you know, better safety for the vehicles because they're primarily driven by a robot that has great reflexes. can avoid situations, but then also lower cost to consumers over time. You did amazing work on Cruise, and then you decided to start another company, which I
“There's only 24 hours in a day. And if you're sleeping for eight hours, working for eight to 10, there's precious few hours left”
and then what things are just completely different or new? Because people would often talk about driving environments as similarly chaotic and messy. And, you know, the canonical example is always like the kid chasing the ball across the road suddenly or things like that. Is it even more difficult in the home? Is it less? Is it more structured? I'm just a little bit curious about
“for a driverless robo-taxi, you basically have no product until it achieves superhuman safety performance.”
“which to me is like a huge red flag, because that means you are, you're saying, I'm not going to be flexible on how I solve the problem.”
of what happened in the industry. I mean, you had a great point in terms of Zoox and others also being part of the things that either had an exit or worked in different ways over time or continue going to be built against. One other thing that a lot of people do in the context of robotics, and you can see this maybe even being accentuated more in the context of a home robot, is they ascribe personality or they kind of project personhood onto these machines. Is that something that you think is worth leaning into? Is it something that's worth avoiding, like the anthropomization, I can never say that word, dehumanization of these devices? How do you think about that as somebody who's actually building things that will be in the home with
“The more transformative and the more science fiction a technology feels, the higher the skepticism will be for that kind of thing.”
five, 10, 20 years until everyone who can afford them expects robots in the house like they expect home appliances. Like what changes that timeline between the five to 20 year span or whatever it ends up being? Well, I'm in my office. So it's basically when I get off this podcast and go back to work. OK, so we should let you go is what you're saying. Yeah, no, no, not at all. So, no, I think to me, and I felt this way in 2013 when I started Cruise, it seems like the basic building blocks are there.
I can point to all the challenges for a low-cost home robot, and low-cost is important to me because I want a lot of people to have this. I can point to all the technical challenges that, at least today, that I think we're going to face. And I can point to a technology we've either built, you know, the last year, or some research that came out, or like a chip that's coming in the pipeline, whatever it is. I can see, you know, from where I sit today, the path to put these things together and assemble a great product and a great product experience. And so I think it comes down to like execution, how quickly and how well those things are put together. And then the big question, which you're always facing something like this is what are the unknown unknowns? Like, what can I sit here today and just simply not see because we haven't put enough robots in homes and haven't tried this out. And I think there could be like something, some new discovery that happens. Maybe it turns out that, you know, people are not happy with a home robot unless it does x or whenever a robot does this other thing it you know it makes people never want to use it again um and so we're kind of early in our own journey to figure out what those things are as
far as i've seen there's no one else really doing this and so it's a big the cloud of uncertainty is is there's a lot of it in front of us and that's that's why i can't uh be more specific on the timeline you know it could be like it could be like one year or it could be like 20 years um and so you know the best way to figure that out is just charge forward and and try to discover it as quickly as possible. All of those are really exciting as timelines. We talked about Chinese
“and then maybe a phased release process to go from like a working prototype to, you know, being regulated, but being able to operate at scale.”
versus you know over regulation versus none at all is having these like phased approaches What's the number one thing that you do differently given the cruise journey besides like, I guess, have more fun.
Don't sell it to GM.
“I'm going to keep the company small to build like a company that has grand ambitions”
You've taken on three high-risk, complex companies back-to-back with little downtime in between. You've run seven marathons on seven continents in three and a half days. Where does your drive and stamina come from? Is there a supplement we should all be taking? Is there like some Bryan Johnson style? Like we should inject ourselves with young blood. Like what's the deal? I haven't tried that. if you do that, let me know how it goes. I have a problem, which is that once I get an idea in my head, it just, it just like burns a hole in my brain. I cannot, I cannot sleep. I cannot do anything. I can't focus until I like, you know, see this idea through. And this, for whatever reason, it doesn't happen. It's not like, you know, 20 ideas a day. It's like, I'll get this idea that, Hey, the stars aligned, this thing should happen. Now is the time I need to go for it. And then once I'm on that track, I just cannot let go until it's done. And so I think I just latch onto these problems and have this like sense of delayed gratification where I want to work on
it for a long time and get the result. But that's also really satisfying to me, the notion of like putting in a ton of effort. Building something really complicated and hard, sort of taking on these really difficult challenges and making a little bit of progress each day, that is motivating to me. And so, you know, you could say starting these companies in high risk and things is difficult and it's hard, and it is, but I enjoy every day. And so I can't imagine doing anything else. That's awesome. I think that's what makes Silicon Valley so great. So thank you so much for joining us today. Thanks for having me.
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.
