SPEAKER_00: when you're building a company like this and you're trying to get product market fit you have to find the place where your product can provide the most value in the short to midterm so you have customers but at the same time looking at the long term what's the biggest opportunity in a way you're being reinforcement learned as the founder here the same way the chess robot we're talking about in terms of reinforcement learning you could play the short game which is hey we got to get into factories and figure out how to move these dorito chips and batteries into the boxes without crushing them and quickly 24 hours a day 365 days a year but also hey winning the game could mean maybe losing some customers but building that general purpose robot that you could put a hundred of them into a factory and just say go find work is really an interesting concept

SPEAKER_03: this week in startups is brought to you by lemon.io need to speed up your product development without draining your budget hire vetted engineers from europe at lemon.io go to lemon.io slash twist to get 15 off for the first four weeks vanta compliance and security shouldn't be a deal breaker for startups to win new business vanta makes it easy for companies to get a sock to report fast twist listeners can get one thousand dollars off for a limited time at vanta.com twist and linkedin marketing to redeem a free 100 linkedin ad credit and launch your first campaign go to linkedin.com slash next

SPEAKER_04: unicorn all right one of the most interesting angles for artificial intelligence is how they might

SPEAKER_05: impact robots in the real world now of course robotics has been going on forever you've been seeing boston dynamics or cafe x making coffee or little tiny robots the rumba going around your house and vacuuming maybe you've seen the automation at an amazon factory incredible to watch and robots have largely lived inside of factories and they largely have been programmed by developers and there is no ai going on computer vision is a small area uh that's an exception we invested in a great company root ai that was picking berries uh and using some of the hands that have been made the hand technology robotic hand technology from mit to very carefully use computer vision to find the right strawberry to pick at the right time we've talked about it on this program over and over and over again but now that ai is starting to hit a tipping point as we've seen a lot of founders are focusing on hey can we get

SPEAKER_09: a robot to do reinforcement learning and we're going to hear all about that today from peter chen he's the ceo and co-founder of covariant peter welcome to the program thank you for having me it's great to

SPEAKER_11: be here all right um now you worked at open ai for uh for a year or two uh you went to berkeley

SPEAKER_05: you got your phd and you founded covariant back in 2017 raised a ton of money just did a 75 million dollar series c led by our friends over at index um let's talk a little bit you heard my preamble robots living in factories yeah robots not using ai being very being programmed to do very vertical specific tasks yeah you know one robot in the tesla factory is going to do something radically different than the next one exactly putting ai in front of these things um it's just not going to work in a lot of cases and it could cause a lot of damage because these robots are big powerful fast and they can break things including humans which tragically we see in these factories so what is your approach you got to this early what is your approach in terms of putting ai and robotics together to try to hit this future where my gosh could robots be learning uh and using ai to do new tasks

SPEAKER_22: in the actual real world yeah so um that is a really good preamble in terms of a history to

SPEAKER_24: robotics right so i would say robotics is not a new technology and not a new field like there are a lot of robots out there in the world in car manufacturing plants in electronics assembly lines there are robots in all of these different places and exactly like you said jason those robots are programmed and typically what they do is they do just the same motion again and again and then an automation line and assembly line is so costly because you need to perfectly engineer every step

SPEAKER_25: of the process so that a robot that is only doing repeatable motion again and again can succeed but you can imagine there are a lot of things in the world that just cannot be reduced to repeatable motion again in it and those are really everything that robotics has not been able to crack before including the strawberry picking examples that you mentioned including really all of the manipulation like

SPEAKER_24: things that require your hands in warehouses and logistics which is what we focus on when you think about those facilities like you're handling hundreds of thousands sometimes millions of different kinds of items that exist in an e-com warehouse there's no way you can reduce the order fulfillment of that many items to a perfectly repeatable mechanical process and those are all the places that we have not seen robots play a big role yet and that's really how i think about ai's role in robotics

SPEAKER_25: it's really not making those robots that are doing mechanical movement again and again better like you don't need ai there like you just don't need a program to do that but what ai can really do is take robots out of those perfectly structured environment where you're just doing the same

SPEAKER_24: thing again and again to a much bigger road where you really need to handle dynamic diverse circumstances that's changing every second every day every season and that really open up a couple or couple more models of many tiers of robotic applications that are possible covariant we are starting from warehouses and

SPEAKER_28: logistics but we really see the broader world as a fair game for this ai apply to robotics

SPEAKER_05: and they are really so going into the factories just just to because you mentioned that twice yeah is obviously um a great place to go because you have a high frequency of transactions as you mentioned number two you have a high variability the different sizes i ordered a bunch of straws you ordered a couch you know these are very different sizes a pack of batteries and uh you know a computer let's say a laptop uh and then on top of that um it is a semi-controlled environment so you're not building a robot that goes down the street and delivers a burrito and it's going to get kicked over so yeah while there's variability it's controlled variability it's variability on this you know a conveyor belt in

SPEAKER_25: front of you let's say right exactly so like i think you can think about the evolution of autonomy robot autonomy to going from perfectly structured environment to semi-structured environment which is what we're handling in this type of warehouses distribution centers um industrial environments lots of variability but still semi-structured like you're not you're not going to have for example people kicking around or what self-driving cars can run into is like a turkey chasing a toddler on the street like really out of bounds um scenarios um and so it's kind of like somewhere in the middle and

SPEAKER_24: not fully to the open world but you still expose you to a lot of diversity and complexities um of the

SPEAKER_41: real world imagine this you got an idea for a tech startup you're going to change the world i know it but you got a problem you don't have any engineers engineers hard to come by they're very busy they got jobs backed up well you need to find great engineers you need to find them quickly and you need to reduce your burn rate right because you can't be spending like a drunken sailor you have a limited amount of resources as a startup now imagine there was a partner out there waiting to help you who had a thousand on-demand developers and they were vetted experienced results oriented and passionate about helping your startup grow and what if they charge competitive rates you know reasonable rates does this sound too good to be true well you need to head to lemon.io right now startups choose lemon.io because they only offer hand-picked developers with three or more years of experience with strong portfolios and if anything goes wrong lemon.io will replace your developer as soon as possible a bunch of launch founders have worked with 11.io they've had great experiences so here's the call to action super easy to learn more go to lemon.io twist to find your perfect developer or tech team in 48 hours or less and twist listeners get 15 off the first four weeks what a deal so stop burning money hire developers smarter and visit lemon.io twist easy peasy lemon

SPEAKER_43: squeezy at lemon.io twist in terms of where we're at in this process and where you're at with covariance

SPEAKER_05: um if i were to look at say um playing games yeah and so you have a very uh finite game chess and then you have an almost uh a finite game but a much larger uh base of possible outcomes go and then

SPEAKER_47: you have games that have a massive amount of human variability in them like say poker so we've watched

SPEAKER_05: as those things have fallen and then even deep mind taking the entire atari 2600 catalog we just had mustafa on the program uh one of the co-founders of deep mind where are we at in that timeline are you at like chess are you at go are you at you know random video games and do reinforcement learning if you had to baseline 2023 ai in robotics reinforcement learning where is it at yeah it's a really good

SPEAKER_22: question um and and the way that you're framing it um gave me a lot of feedback to my days of doing the reinforcement learning research at open ai and at berkeley i was training a lot of reinforcement learning agents like exactly also with atari suites of games so so it gave me a lot of flashback of

SPEAKER_25: memories but coming back to this question um this is a really interesting way to frame it and i would

SPEAKER_22: say where ai for robotics is at um it's really for my technology from an algorithm for my models

SPEAKER_25: from a compute power perspective we are at that atari moment and really beyond like i would say we are at

SPEAKER_53: even the starcraft dot oh that's a big chunk too right i mean the difference between a game like starcraft and you know a 2600 game like pong is like the difference between checkers and maybe poker

SPEAKER_56: or go right it's it's a that's a big leap it's exactly so from an algorithmic and from a modeling

SPEAKER_24: perspective we are there but what is missing is data so it's very very difficult to get robotics data

SPEAKER_25: diverse robotics data that can build this type of ai so let's even use goal as an example right if

SPEAKER_58: you think about alpha goal incredible achievement like that can be human champion um and go incredible

SPEAKER_24: breakthrough that the mind um has built a couple years ago um and and you have mentioned like why this was incredible this was incredible because because goal is a really complex game if you look at all the possible compositions of a goal board like there's some something like 10 to the 170 possible

SPEAKER_58: configuration of a goal game and that's more than the number of all particles in the observable universe

SPEAKER_59: and that's like a crazy pause on that the game go which just has two different stones

SPEAKER_61: exactly seemingly seems to be you know on on the surface value you would just look at go and be like

SPEAKER_05: checkers oh it's a simple peasant game it means nothing chess is much more sophisticated it's not because of the size of the playing board exactly you know and uh the number of that then i think it's the dynamism of what can happen when you have a multiple angle flip and you know three or four different root you know um rows change at the same time it's just so many possibilities of uh outcomes

SPEAKER_65: exactly so wild wild like wildly complex game and that was why like for a very long time leading

SPEAKER_25: artificial intelligence researchers didn't think we would crack go uh in any time soon and so it was a big shock like when deep reinforcement learning was able to crack uh the game of goal and but if you look behind the hood like one key thing that really power that is the amount of data like if you look behind the goal winning um deep neural nets it's trained on more than a hundred years of goal playing experience i mean we can go into the details of like it starts from self-play expert play and all of those kind of good stuff about the data but just just pause for a moment and think about the sheer amount of data that is trained on yeah when you're playing against alpha go you're effectively

SPEAKER_72: playing against a goal player that has done nothing in her or his life for a hundred years just playing

SPEAKER_05: right their life is a hundred years of playing a hundred games simultaneously who knows i mean it depends on how many h-100s i guess you have exactly racked but it's playing such an amazing number of games uh and figuring out outcomes and it doesn't even need to be trained so this is a good place to pause given your background if you were going to explain reinforcement learning in the case of go or starcraft or yeah you know playing pong to somebody who had never heard of artificial intelligence you just want to understand how reinforcement learning works on a very basic level yeah what are the you know three or four key concepts and terms of art in reinforcement learning yeah so the most two most

SPEAKER_25: fundamental concepts that you need to understand for reinforcement learning one is you learn from

SPEAKER_22: doing different actions so if you only do the same action again and again there's no reinforcement

SPEAKER_24: learning because you have no contrast right and then the second concept that you need to understand is um there needs to be a reward function like so once you do action one it leads to some outcome one you do action two it leads to some outcome two there needs to be a reward function that can rate which

SPEAKER_25: one is better once you can have an agent that do different actions and the actions lead to different outcomes that can be rated by a reward function then you can start doing reinforcement learning and at a high level it's really simple it's about the agent exploring the world by taking different actions and that lead to different outcomes and that outcome is rated by reward function and then the

SPEAKER_24: reinforcement learning algorithm just look at what are the actions that tend to be better and you start that learning loop on there there are a lot of technicalities on how to make that work and how to

SPEAKER_58: make that run efficiently and how to make it work well together with a big neural net um like for

SPEAKER_25: example how do you turn gpt4 into chat gpt4 like there's lots of craft and details that's needed in making that happen but really at a high level it's as simple as that it's taking different actions

SPEAKER_24: and figuring out which one is better and try to do that more often that's really the core basis of

SPEAKER_84: reinforcement learning okay so reinforcement learning the uh to reflect it back to you

SPEAKER_05: the behavior you have to have a an a b c choice right so you have to have a behavior choice in the case of you know chess it would be moving one of the pieces according to the rules and there's only a certain number of pieces that can move in the opening move and then what is winning what what do you want to reinforce what do you want to tell it good and good in chess is having a piece taken or not having a piece taken are those the two basic components there yeah those would be

SPEAKER_24: like a good incremental reward function and then your ultimate reward is whether you have won the game or not right you can imagine losing all of the pieces but if you won the ultimate game

SPEAKER_05: that's still a good outcome well and this is like a great point because if you look at somebody like magnus or some of the top chess players i watch clips of them i don't know if you've ever watched clips of them on like tick tock or youtube they they they now like make little short clips of the best endings one of magnus's like incredible gifts is he sacrifices massive amounts of material he'll sacrifice a pawn and then he'll sacrifice a rook and you're like oh my god he's dead but those sacrifices lead to a series of moves that boom checkmate and so it could be that getting a material advantage um is the wrong uh training it's like it's like that's the short term thing that's right to do take the pawn take the rook so the other player playing magnus thinks they're doing the right thing but they haven't thought as far ahead as magnus who is now you know mate and two when

David Friedberg: you or mate and one when you when you make that error to take the rook yeah and you are referring

SPEAKER_25: to a technical concept um here in reinforcement learning that is called like how do you optimize for a very delayed reward like you're optimizing for something that has a long-term dependencies like you make a move now and maybe you lose a couple steps but you ultimately win the game and a big challenge in reinforcement learning is how do you figure out that delay outcome and how do you figure out that long-term dependencies um that you have um i want to i want to take a step back and coming

SPEAKER_24: back to the data question on on robotics and on on your earlier question of where we are in the ai for robotics evolution and i make this comment that from an algorithm and model perspective we have what we have like but we don't have data we don't have the equivalent of goals data of 100 plus years

SPEAKER_95: of diverse playing or chat gpt ingesting reddit and twitter exactly and open crawl or google indexing

SPEAKER_24: the web and putting it into bard exactly we don't have the red equivalent we don't have the github equivalent um in robotics and that is the key limiting factor hold on a second though does not i i gotta

SPEAKER_05: think bezos who's a genius would have cameras all over you know the conveyor belts in the factories would you know a couple of cameras watching humans do this be the potential you know uh data source or is it not trained enough so a robot and an ai watching a human pack boxes would that be if you had a million hours of that enough for you to send a robot in there so it's a really good um question

SPEAKER_25: like this topic like has an academic name like this is called third person imitation learning like so this is like you're seeing a third if from a third person view someone else doing it can you learn from that and i would say like the best state of the art is you could learn something from it but it's never as good as if this is coming from your own actions how you have tried it and whether you can actually learn from that yourself and the reality is actually like even for amazon like they actually release a couple data sets on the items that exist in their warehouses and actually the data set is much smaller than what we have um collected even here at covariant from across the diverse set of customers um that we have a big chunk of that is it's not just about the data it's also getting about like it's getting exactly the right format of data and getting the right type of data right like if you think about the modern movements in large language models a lot of the secret sauce is in the type of data that you curate and this is not something that you can just oh let me try to crawl more of the internet right the equivalent of that would be try to put more cameras in these warehouses and just look at conveyors right but are you actually capturing the useful moment the most meaningful data and to understand what kind of data you need to capture actually require really deep understanding of what you actually need the robots um to solve so can we collect um useful data yes for sure like we can already start collecting that today but what we have found is that really to teach

SPEAKER_24: robots to operate very autonomously you need extremely high quality data and if you need very high quality data they need to be collected in a very yes purposeful um way i would think the hardware

SPEAKER_05: hardware uh that you use is distinctly different than a human hand now they might have modeled it after

SPEAKER_00: the human hand but it's going to move in a different way you can move faster it can move in ways that would give us carpal tunnel syndrome that they would you know the hr department would say you know do not bend over like this it's going to cause you know carpal tunnel it's going to cause back problems a robot does not have those right the robot can just move in any direction you could you know uh you know hyperextend your elbow and dislocate your elbow to move a package with a robot so the data you collect

SPEAKER_09: has to take that into account is the flexibility of the robot which which can twist and turn in any

SPEAKER_00: any given way where's the let's let's talk about where that robotic the robotic hands are robotic arms because you can buy a robotic arm now exactly that's capable of doing you know 24 hours a day 365 days a year with very little downtime yeah that can lift hundreds of pounds for how much now what is the entry-level robotic arm that you might see in the cafe x coffee machine or you might use in a small

SPEAKER_04: warehouse uh to move you know a 10 pound package or something what do those go for now industrial robots

SPEAKER_58: are incredibly um robust and mature technology and so they they're really good like they last for a

SPEAKER_24: long time like they can work 24 7 and with proper maintenance like those robots can go for 10 years so it's actually really incredible um technology that's been built up and tooled by the automotive industry um these type of robots depends on size and payload it typically goes somewhere from 25k to 50k

SPEAKER_25: so which is not a very significant cost like if you really think about like the length of time

SPEAKER_00: super industrial if you if you were to compare it to a human arm a human arm in a factory would cost you 50 an hour in total compensation and if it was working you know 24 hours a day that's 1200 a day 365 days a year we're talking about a half million dollars per year over 10 years five million dollars one of these arms could do that for 50k exactly like so the robot arm itself

SPEAKER_117: incredibly cost effective um technology if you're a sas or services company that stores customer

SPEAKER_119: data in the cloud then you need to be uh sock to compliant you knew that from a third party and you need that third party to close big deals and if you want to get compliant easier and faster you need to use vanta v-a-n-t-a vanta makes it so easy for you to get and renew your sock 2 on average van customers are sock 2 compliant in just two to four weeks prepare that to three to five months without vanta and vanta can save you hundreds of hours of manual work and up to 85 percent of compliance costs this is a total no-brainer and vanta does more than just sock 2 compliance they also automate up to 90 compliance for gdpr hipaa and more you can't afford to lose out our major customers we all know that listen it's a hard year last year was hard you can't lose those major customers because you don't have your compliance dialed in just work with vanta get your compliance automated and tight and tight is right lock down those big deals here's the best part vanta is going to give you a thousand dollars off that's ten hundies get one thousand dollars off at vanta.com twist that's vanta.com twist

SPEAKER_00: for a thousand dollars off your sock two so we have a video speaking of robotic arms maybe a good point to stop here uh you can take a sip of your coffee there and then show us this video uh and

David Friedberg: then we can continue the discussion maybe because people uh you brought some show and tell if you're listening and you're not watching on youtube just go over to youtube type this weekend startups covariant

SPEAKER_24: and you'll find it real quick so what we're seeing here is the robots that's picking up a really diverse set of items in a really chaotic pile of items so what the robots need to do is not just repeating the same motion again and again but it really needs to understand what's in front of it in 3d what are the different objects what are the different ways to approach the item what's the best way to pick it up and really manipulate it and transfer it successfully and for those of you who is watching the video you can really see a diverse set of items ranging from items from pharmaceuticals to cpg um to candies food grocery items and so you can really see how these items they come in different orientations in the world uh and they also each one can come in in a different position but also different deformation like if it's a bag like there's not the same bag that would always appear in

SPEAKER_36: the same um if it's a bag of doritos which i think i see bags of doritos and bags of gummy bags

SPEAKER_00: a gummy bears there yeah if you throw 50 bags of gummy bear that 50 bags of doritos or gummy bags into a tray um they're going to land in all different ways it's going to look very different to the robot but after training like you're saying it's going to know this is a bag of doritos it has a certain texture you don't want to crush it if you hit it too hard it's going to crack some doritos um whereas the gummy bears maybe you could you could hit those a little harder so what is that arm called it's not a hand type arm it's not a pincher it looks like it's got like two

SPEAKER_05: digits that are kind of like suction cups what am i what are we looking at there in terms of these

SPEAKER_25: are these are these are vacuum based uh hands so like um jason like i said like you can very nicely

SPEAKER_24: articulate the difference between the arm like which is like the white body that we're seeing here this is a robot arm that's manufactured by abb one of the world's top four robot manufacturers and then there is the hand like which is actually the part that gets in contact with the physical world like

SPEAKER_58: so the hands here it's actually a very simple mechanism it's like so um there's vacuum that gets generated and it's just kind of like one of the vacuum at your home like you can suck things um into it and then you have two tubes of vacuums or two cups um that the robot is individually actuating there so you can choose to use one cup or both cups together depends on what is trying to pick up

SPEAKER_04: or i suppose a different percentage on each one uh exactly and and so what we have found is that um

SPEAKER_24: suction or vacuum has turned out to be a fairly general hand technology like you can use this fairly extensively in the warehouse setting like it's not going to solve everything in the world like

SPEAKER_25: just it's not as dexterous as a human hand but it's actually fairly universal what we have found to be very important though is you cannot have the same robot hand uh everywhere like if you're handling much larger items you can imagine you need bigger suction tubes you need this is picking up a kettlebell

SPEAKER_00: a 50-pound kettlebell is not getting picked up by the suction arm probably exactly but it's easily picked up by a pincher you know or whatever you call uh you know clasping like a finger based

SPEAKER_24: um um gripper right so yeah and what what this actually points out is something that's very interesting in um ai for robotics is the ai for robots needs the ability to adapt to different

SPEAKER_58: kinds of physical hardware like it really needs the ability to um not just handle one kind of physical

SPEAKER_24: body but actually multiple kinds of physical bodies like because we haven't been able to build a hand or a robot arm and robot body that is as universal as human body what that means is that

SPEAKER_25: for different use cases for different customers you need different hardware and now your ai needs to

SPEAKER_58: have that ability to adapt to different physical embodiment um um that it has um and and this is

SPEAKER_25: actually a pretty getting at a pretty interesting idea right then how do you actually build an ai that can learn across multiple different physical bodies and how can you build an ai that can learn across

SPEAKER_24: multiple scenarios and different kinds of item sets that you're building and that really sits at the core of what we're building which is what we call covariant brain a foundation model for robots and we say it's a foundation model for robots because similar to chat gpt like that is learning across translation tasks coding tasks like all of these different language tasks related together

SPEAKER_25: the foundation model that we build also learn across multiple different robot tasks with different

SPEAKER_24: robot hardwares in different customer scenarios in different verticals together and we do that because that is necessary to solve the data problem for ai for robotics that we mentioned earlier like

SPEAKER_25: imagine like if every time i need to come up with a new robot hardware platform i need to collect a bespoke set of data sets for that hardware and if i need to collect a bespoke set of data set for one customer you're never going to build up to that alpha goal moment of a hundred plus years of

SPEAKER_24: experience like so really the only way to bootstrap this foundation model for robotics is to collect all of them together and you have to be one ai that can learn across all these different tasks together

SPEAKER_30: which begs an interesting question someone like amazon would see this information this data this

SPEAKER_00: learning as a proprietary advantage over target and walmart whereas target walmart or some other vendors who maybe were far behind let's say target was way behind walmart um and amazon in terms of automating their factory i'm just making this up they might very much want to contribute their data in order to get your solution so how do you think about the go-to market strategy as a founder where some

SPEAKER_05: people might say i don't want to give you that data or how do you get that data and then how do you

SPEAKER_58: negotiate that with your customers yeah so we work with um customers that grow with big e-com customers that are um innovation forward like they know automation with ai power robots is the way to go in

SPEAKER_24: the future and they also look at themselves and say um there's no way i can build that competencies internally like in order to compete with amazon in order to keep up with the innovation i have to work with a startup that takes a partnership approach and can really bring that technology to them so one of this example is um one of the recent customer that we have announced in europe is with a customer called auto group otto group um they're actually the second biggest e-com company um in europe behind amazon and so they exactly compete against amazon so they they are the big e-commerce concom but they also own i think the american more familiar brand would be crate and barrel so they own crate and barrel um and and why do they partner with covariant like they partner with covariant exactly because they see ai robotics as inevitable and they see that using that capability as one foundation model one platform to transform multiple places of their supply chain network as crucial but at the same time they cannot build that themselves right so really the best move is to partner with covariant and

SPEAKER_58: really bring that technology um to their network and yes like they're contributing data um to the platform but they're also benefiting significantly from yeah gift again and and and because we have already

SPEAKER_24: built up such a broad data sets over robots across multiple continents they the robots that's deployed at their site um can also achieve much better performance they want before you even contribute any data to it like you can already start benefiting from it and this is a large part of what makes the current state am so powerful right because like even before you fine-tune for example a llama on your own data sets it's already quite useful right because it has already learned so much about the world that even for your own business problem hasn't seen all the proprietary data yet it can already

SPEAKER_58: perform well out of the gate and by working with us like we give them the ability to um ingest their

SPEAKER_93: data onto this large data sets that also make the ai more powerful on this specific um use cases got it

SPEAKER_152: when you're selling to b2b buyers you really want to get your pitch in front of the decision maker the

SPEAKER_155: person who gets to sign the check because these upper level execs they're the ones who make the purchasing decision everybody can have an opinion on the team of course it's 2023 but there's always somebody where the buck stops and that buck stops on their desk and doesn't get into your bank account these high level folks are hard to find they're hard to target on social media platforms but linkedin is the social network for business and they have 930 million members ready to do business with you and that includes the 180 million senior level decision makers plus don't tell anybody there's also 10 million c level executives there that's a ton purchasing power linkedin ads is built specifically for b2b marketers no other platform in the world can offer these eyeballs and you can target them obviously by their location the size of their organization their vertical and their title when you think about business i want you to just think about linkedin linkedin equals business business equals linkedin it's that simple folks when you present them with an opportunity they will of course be in the mindset to receive that because they're not posting pictures of their food from Italy on vacation make b2b marketing everything it can be and get a hundred dollar credit towards your next campaign by going to linkedin.com slash next unicorn to claim your credit that's linkedin.com slash next unicorn terms and conditions apply because linkedin is so generous to the this weekend startups audience

SPEAKER_30: what do you think of the projects like tesla's i think it's called optimus um you know building an

David Friedberg: actual um you know full-on robot that looks like a human and walks around like a human obviously these arms vertical you know they're they're vertically integrated they've been out there for now what is it 30 40 years of these arms you know being at scale i guess and uh making a huge difference beating humans every day um but you know some folks are going to take the approach like elon's doing uh this figure i guess is the other one uh of hey here's here's a robot that is a human it's going to

SPEAKER_00: reinforcement learn uh and uh walk around your factory or walk around your house and put dishes in the

SPEAKER_09: dishwasher or pick up and clean up uh after the dog if it you know spills its uh food over or worse yeah

SPEAKER_57: so i'm very glad that someone is working on humanoid robot like this is going to be a very

SPEAKER_25: key enabling platform to really open up to a wide set of robotics use cases so if you think about

SPEAKER_24: like these industrial robot arms they are really good but what's the limitation of them the limitation of them they are largely fixed stationary robots meaning you have to bring work to it and you have to

SPEAKER_25: constantly feed work um to it and those type of robotic applications only make sense if you're running a

SPEAKER_24: two to three shift operations and the robot constantly can be busy and that's the type of use cases that

SPEAKER_25: we are solving for our customers like these are heavy industrial environments that there's work constantly happening and that's where you have this super positive um business case um it's a very big

SPEAKER_24: market right like we can easily sell billions of dollars of arr worth of um robots um to this type of logistics setups that run two three shifts of operation but it's not everything right a lot of things that um in a not as intense industrial setting like maybe you would only do your dishwasher twice a day at most right and that's twice a week who knows yeah yeah or twice a week right so does it make sense to have a dedicated robot arm that's fixed around your dishwasher to only do that

David Friedberg: unless you're running a cafeteria and i looked at uh there's a dish bot i think it's called which

SPEAKER_00: was using magnetics to pick up the things but you had to use the same dishes so if you had a cafeteria and use the same size bowls the same type cups and they're made of like plastic not china you could actually use it but yeah so with this you know tesla optimus or you know the other ones in the market

SPEAKER_22: they can go find work they can go to your backyard so they don't work exactly they don't need to work in one fixed setup like that only high volume industrial environment can can afford like it's

SPEAKER_24: like commercial kitchen like it's another type of like industrialish environment so this is going to be a very important platform like basically it's going to open up robotics to even more use cases where it's starting to go into automating things that aren't frequent that don't happen frequently so very very important technology building um that needs to happen i personally don't really have a forecast of when this will land like because this is a very tall challenge right you are trying to do a lot of things that's not very high level of very high value each one in isolation like but you need to be able to do a lot of these so that puts a lot of burden on the generality of the hardware platform and the cost of it because each single one of this is not going to be very high value

SPEAKER_123: um that also means like if your humanoid robot costs a million dollars that's not you can't use it in

David Friedberg: your house unless you're you know got a lot of dispensable income exactly uh but you could use it uh if you're a military application you know or it's a a newfangled firefighter that can go into

SPEAKER_00: a burning you know uh pet store and take the pets out of the pet store without getting burned like there are going to be some applications where you're willing to pay a million bucks and actually we have

David Friedberg: bomb robots now they just don't look like humans they look like you know um little rc cars right they drive them around such an interesting point you make because when you're building a

SPEAKER_00: company like this and you're trying to get product market fit you have to find the place where your product can provide the most value in the short to midterm so you have customers but that but at the same time looking at the long term what's the biggest opportunity in a way you're being reinforcement learned as the founder here the same way the chess robot we're talking about in terms of reinforcement learning you could play the short game which is hey we got to get into factories and figure out how to move these dorito chips and batteries into the boxes without crushing them and quickly 24 hours a day 365 days a year but also hey winning the game could mean you know uh maybe losing some customers but building that general purpose robot that you could put a hundred of them into a factory and just say go find work is really an interesting concept it is it is very interesting and

SPEAKER_57: and we believe the key way there is um keep building a general ai right because that that is the thing

SPEAKER_24: that is going to transcend whatever use cases that you're looking at today and whatever um platforms uh hardware platforms um that you're building it on top of because a generalized understanding of the physical world and how to interact with it is independent of the use cases independent of the use

SPEAKER_25: case frequency and independent of the hardware platform so like from a covariant perspective like

SPEAKER_24: i wish um the tesla optimus like hardware platform exists today because that can allow us to put our ai

SPEAKER_65: our foundation model on it to solve a lot more problems right and i think the biggest problem in

SPEAKER_05: hardware like what is the big uh is it the actuators is it creating the pulley systems i know there's many

SPEAKER_148: different types of pulley systems what seem to be the hang-ups there from a humanoid robot perspective

SPEAKER_13: yeah or from just generally generally and then we could go to humanoid but i think generally what

SPEAKER_05: what is it that the robots can't do yet is there some blocker that everybody's going oh you know like storage used to be or bandwidth used to be in the internet if there was an equivalent in robotics is it those like actuators or the the pulley systems that you know create the strength where the arm can move is it the the tips of the fingers to know this is a ripe strawberry versus this is a firm piece of corn yeah how do you how do you think about that so um the answer to that actually is very different if

SPEAKER_25: you think about it as a general purpose humanoid robot versus like kind of a more classical robotic automations um perspective and i'm going to say a little bit what's the difference the difference is that um from a more classical um robotic automations um the key thing is can you customize very quickly because like for any single physical problem that you want to solve you typically can come up with some clever mechanisms that make the problem easier because the problem is not every single thing that

SPEAKER_24: a human body needs to do so the hardware challenge there is not so much of any single individual one of those but it's for every new problem you need to customize your hardware design a little bit like

SPEAKER_25: and it's your speed of customization that sits at the core of this the humanoid robot is interesting the humanoid robot i would say it's as much as a product problem like as much as a hardware problem as a product problem like we cannot build a humanoid that is as good as human today but then what is the first humanoid product that you should build like i would say it's as big a problem as the specific hardware challenges um that's there like like you said maybe we should build a humanoid robot that focuses on bomb removal use cases first right but then like once you articulate that product problem yeah you can find a way to engineer for it like what's very difficult is engineer something that is

SPEAKER_24: as good as human that's too general and too vague too general all right problem problem um i'm to tackle

SPEAKER_30: well i like the approach that you've taken which is we know the total addressable market for e-commerce

SPEAKER_00: and moving packages around is high volume high transaction and high value right so it's got exactly a lot of the if you would put the circles high volume yeah high transaction high variability maybe or complexity maybe a better word and then there's money involved so exactly it might be small transaction sizes maybe it's 40 on average but there's like a million of them a day coming out of this factory exactly that means you got 40 million dollars worth of product going out a day you know and whatever that is per year so yeah it's extraordinary how much value is it when you studied the tam of other markets obviously factories were one but they're well you know factories are well oiled machines i don't

SPEAKER_148: think they apply as much to what you're doing because there's not variability so this is actually

SPEAKER_25: that like um let me make two comments like one comment coming back to your multiple circles and then

SPEAKER_22: the complexity and the variability part is actually a super important part for building a general ai for robotics because the ai that we build in warehouses and distribution centers can see really virtually any

SPEAKER_24: objects that exist in the world that gives us an extremely good training ground to build ai that can

SPEAKER_28: work um elsewhere like so i mean given this is a technology podcast like i like to like make sure

SPEAKER_25: um and then the second point about um what you were saying um and like if you think about like the type of use cases that we are deploying into like they are you can really think of them as um starting ground to

SPEAKER_24: build the future of robotics again because the key insight here is that the ai that understands the physical world and interacts with the physical world is almost independent of use cases and independent of hardware platform and so by finding this high volume use cases in one industry like it gives us the ability to start this flywheel to start building the ai and start solving the data problem for robotics um that other people just can't have access to because they don't have that real data they don't have

SPEAKER_181: that real world robotics interactions uh amazing uh this is just extraordinary you're in year five or

David Friedberg: six of your journey and uh it looks like the the the understanding of ai and the importance of in the world has caught up to your vision uh so that's nice i guess a lot of investors suddenly that said no to

SPEAKER_00: you for the first five years are now banging on your door saying how do i get an allocation talk maybe a little bit on a practical basis about being a founder when people think that's too hard that's not good that's going to be a money burning pit to everybody saying oh my god that's the future it's here now i need to go make up for this mistake and not backing your company five

David Friedberg: years ago you must have a lot of boomerang investors yeah yeah like it's very interesting

SPEAKER_24: because when we started the company the term foundation model didn't even exist like so when we started telling people what we did was we are building one ai that can learn across multiple tasks learn across multiple robots and then people would be like i don't see why a specific ai isn't better like why shouldn't you just train an ai that is on one specific use case for one specific customer like no one would make that claim anymore today like because people have seen how gpt4 is better at at translation than google translate even though google translate is also a deep learning powered ai based translation system that is just a more task specific one right so it turns out that gpt4 by learning to do a lot of other different language tasks that are not translation gave it better understanding of the world in terms of semantics history memes grammar that actually make it much better at translation than is ai that is specifically built for translation so it definitely has been tremendously helpful in the last um half year or so that um the success of our friends at open ai and far back or core here are the places that have done have really so some

SPEAKER_53: of those vcs came back some of those vcs uh emailed you back after turning you down i take it

SPEAKER_22: definitely like i would say that's that's definitely a huge increase in interest just because

David Friedberg: that's the greatest it's not the greatest feeling ever as a as an as a founder is you get rejections all day long man when you can't you must have met with a hundred investors you know before this ai

SPEAKER_144: boom and gotten 97 noses am i approximately correct 90 no's 95 no's we are somewhat fortunate that like we

SPEAKER_24: have really um big supporters um internally and so we actually had not had the need to fundraise very extensively outside i guess you're open ai mike wopi um from index has been a big believer in

SPEAKER_25: this type of general model um very early on and so we got lucky there but yes like even and that's true not not even just with investors like that's true with customers right like because yeah like

SPEAKER_24: this boomerang customers this whole thesis of general ai being a better approach than specific ai like also customers didn't use to believe that right but now they really cannot refuel that anymore because the whole world is moving into that direction like no one wants to train specific ai model on their own

SPEAKER_25: specific data sets that's going to get there that's less performance than a more general ai platform

SPEAKER_24: anymore so um that movement has been extremely both validating to the approach that we have taken for the last couple years um but it also helps us tremendously yeah customer side from the capital

SPEAKER_30: market and all these different aspects absolutely fantastic to have you on the program uh peter chen

SPEAKER_09: covariant ai you can follow him on twitter slash x he's not super active peter xi chen x i peter x i c h e n covariant ai um what's your domain name so i can send people because i know you're hiring oh covariant.ai yeah so if you're hiring what's that what's the what's the hardest thing to hire for right now what do you need which i'd like to help fill some positions for you to thank you for

SPEAKER_24: your time here on the program yeah we're always looking for great engineering talent great ai talent like if you're interested in solving hot problems that have not been solved elsewhere like which is

SPEAKER_00: what we're doing every day like we would love to hear from you covariant.ai slash careers the hr

SPEAKER_216: department's gonna uh thank me for uh sending up the careers page so yes great having you on the

SPEAKER_00: program continued success and i would love to check in with you in about a year and see how you're doing

SPEAKER_04: on this incredible journey and we will see all of you next time on this week in startups bye bye