Profitez du streaming de ce film sur «empresa de produção de drones» directement sur youtube.

The Robot Brains Podcast met en lumière le thème de « empresa de produção de drones ».

Retrouvez cette vidéo mise en ligne par The Robot Brains Podcast sur Youtube. autour de « empresa de produção de drones »:

Lorsque nous avons découvert cette vidéo récemment, elle suscitait de l’intérêt. Le décompte de Likes indiquait: 46.

La durée de 01:10:07 secondes et le titre Season 1 Ep.6 Keenan Wyrobek discusses how Zipline’s AI-drones are reshaping Africa sont à prendre en compte, ainsi que les informations de l’auteur et la description qui suit :« Dans l’épisode six du podcast Robot Brains, Pieter est rejoint par Keenan Wyrobek, co-fondateur et chef du produit et de l’ingénierie à la tyrolienne. Fondée en 2014, la tyrolienne est un service de livraison de drones à la demande en utilisant l’ingénierie de drones de pointe et l’IA pour livrer des produits médicaux urgents. Dans notre chat, Keenan parle de sa jeunesse déchirant les imprimantes 3D de son école, comment Bono d’U2 a rejoint le conseil d’administration de son entreprise et travaillant avec la NASA pour aider ses drones à voler à travers un temps extrême. Qu’est-ce que dans cet épisode: 0:00 Présentation de Keenan Wyrobek et de son entreprise Zipline 03:42 Keenan’s Education et « Why Classes Don’t Impatters » 05:54 Pourquoi il a décidé de fonder Zipline 11:55 un type très différent de drone 18:22 Getting U2 ONOS pour rejoindre le tableau de zipline de la Zipline 20:19 sur 100k Zips Worldwide 31:44 THE GLANT DIFFITE NASA 51:50 Intelligence artificielle et apprentissage en profondeur dans les drones 58:38 La prochaine génération de drones 1:00:31 aurons-nous bientôt des voitures volantes? 1:06:28 Conseils de Keenan pour les startups Liens: Twitter de Keenan: https://twitter.com/keenanwyrobek?s=20 Keenan’s LinkedIn: https://www.linkedin.com/in/keenanwyrobek/ SITE https://apple.co/3vrqblv Spotify: https://spoti.fi/3d0pewf amazon: https://amzn.to/2swdzfp google: https://bit.ly/3d52uju acast: https://bit.ly/3a0uqo et visitez notre site Web: https://bit.ly/3a0uqo et visitez notre site Web: https://www.therobotbrains.ai Hôte: Pieter Abbeel Producteurs exécutifs: Ricardo Reyes et Henry Tobias Jones Production audio: Kieron Matthew Banerji. Production vidéo: bo obradovic. Tous droits réservés | Le Robot Brains 2021 ».

youtube se distingue par sa capacité à réunir des créateurs et des spectateurs autour de contenus variés, le tout dans un environnement sécurisé et accessible.

Excellence en Vidéo Aérienne : La Production par Drone

Phases de développement et de finition

Les phases clés d’une œuvre audiovisuelle réalisée avec un drone

Chaque projet s’initie par une étude des besoins et la définition d’un concept bien établi. Anticiper les défis techniques implique un repérage des lieux et une planification adéquate du tournage. Des experts en pilotage s’occupent de la captation des images pour garantir des plans de qualité optimale. La dernière étape comprend le montage ainsi que la post-production pour un résultat optimal.

Outils de montage et d’édition

L’étalonnage et la correction des couleurs favorisent une esthétique visuelle unifiée. L’impact esthétique des vidéos est rehaussé par des transitions fluides et des effets visuels. En utilisant une bande sonore de qualité, avec musique et voix-off, on enrichit l’immersion et l’émotion du spectateur.

?Matériel et technologies en place

Les drones professionnels primordiaux

Les modèles comme l’Inspire 3, le DJI Mavic 3 et les drones FPV sont souvent choisis pour leur performance et leur capacité d’adaptation aux divers types de tournages. Les drones sont sélectionnés en fonction des besoins spécifiques du projet, garantissant ainsi la meilleure qualité d’image.

Le rôle essentiel des équipements supplémentaires

Les caméras et capteurs professionnels permettent d’obtenir une netteté inégalée. Pour une qualité audio optimale, il est essentiel d’utiliser des microphones spécialisés. Des logiciels avancés garantissent un montage précis et fluide en post-production.

Comment faire le bon choix d’agence de production drone ?

Éléments déterminants à scruter

L’expertise et le dossier d’une agence constituent des critères essentiels pour évaluer la qualité des services. Assurer la conformité aux règles de l’aviation permet d’atteindre une sécurité idéale pendant les prises de vue. La maîtrise des dernières technologies, couplée à une expertise technique, représente un atout majeur pour un rendu professionnel.

Quels avantages offre la collaboration avec des professionnels certifiés ?

Des télépilotes qualifiés garantissent la sécurité des vols tout en respectant les exigences légales. L’accompagnement sur mesure tout au long du projet permet de répondre aux souhaits du client et d’améliorer le rendu final.

Quels sont les atouts d’une boîte de production de drones ?

? La production audiovisuelle bénéficie grandement de l’utilisation des drones

Avec des prises de vue aériennes à couper le souffle, les drones redéfinissent la production vidéo pour un rendu cinématographique. En offrant la possibilité de captation en intérieur et en extérieur, ils apportent une flexibilité accrue. Ils fournissent une option économique pour les prises de vue par grue ou hélicoptère, contrairement aux méthodes traditionnelles.

Un effet direct sur la qualité des photographies aériennes

Les images prises en haute définition garantissent une qualité professionnelle grâce aux capteurs 4K et 8K. Les angles variés et la fluidité des séquences favorisent la création de vidéos engageantes, idéales pour le cinéma, la publicité et les films d’entreprise.

Transformez votre communication avec une vidéo aérienne !

En optant pour une agence de production de drones, il est possible d’obtenir des visuels spectaculaires et de promouvoir une entreprise. Un projet bien conçu attire l’intérêt du public et solidifie l’image de marque. Pour une communication renouvelée, faites appel à un spécialiste.

Ce partenariat marque un tournant dans le monde de la production audiovisuelle. SupraDrone et E-Media Production combinent leurs talents pour offrir des vidéos spectaculaires et immersives. Découvrez leur approche sur c e site.

Regardez la vidéo sur Youtube en cliquant sur ce lien :
la publication originale: Cliquer ici

#Season #Ep.6 #Keenan #Wyrobek #discusses #Ziplines #AIdrones #reshaping #Africa

Retranscription des paroles de la vidéo: [Music] today here with me is kenan weire kieran is the founder cto and product architect of zipline the world’s first drone service that delivers life-saving medicines to the most difficult to reach places on earth welcome kenan thank you great to be with you here peter it’s so good to uh see you again ken it’s been such a long time we used to spend a lot of time together back at stanford in our phd days and you’re working completely different things than i remember like the first time i heard of you and saw you was all of a sudden this robot existed at stanford that was on wheels and had two arms and had a head and it was just like wow this is so new and really what i’ve been hoping to work with forever and all of a sudden it exists awesome wow bringing back the good memories from the old days okay yeah that was quite the surprise um and of course um right now you’re in a completely different situation you’re you’re not not a pg student at stanford anymore um you’re you founded zipline and actually spent most of yesterday watching all the zipline videos that i could find and it’s absolutely amazing what you’ve been achieving there um but before we get to those achievements and what’s going on at zipline i’m curious uh where where did you grow up and how was the path to get to that moment when you founded zipline i grew up in california uh on a little lake uh in the in the uh yeah this awesome little lake like spent some of my time swimming sailing little rafts we would build like yeah just doing all kinds of fun stuff and how do you go from growing up you know enjoying life at a lake to to founding zipline i’ve always i’ve always been drawn to kind of like first of all like building stuff taking stuff apart like that’s just i don’t know that’s just what i did for fun all the way even when i was a little kid that’s just i was if there was if i get my hands on a screwdriver i’d take something apart and you know getting good trouble for it usually depending on what it was that was just like i don’t know there’s something about like just knowing how things worked and knowing what was inside and knowing sort of what was behind the scenes that was always just a big part of just the fun for me um and uh yeah i think the other thing that really as i yeah even in high school i really enjoyed just if there was a problem that i thought mattered like i had to be passionate about it but once i thought it mattered like you know you like that was just that was also just where the fun came from and i think even even in school i spent a lot more time working on things i was excited about than like you know homework or classwork and things like that uh and yeah i don’t know that that being passionate about things that mattered and working on them got me through school and high school and college and grad school really effectively and and i often joke with uh well i think even back when we were at stanford like i was spending more time building that robot than i wasn’t like doing my class work and like it was just always because i thought it mattered i thought it really mattered to try to build a kind of a common platform for robotics development i thought it just mattered more than like the classes i was taking that day so yeah i can tell you as a professor i think the hardest part of being a professor and advising phd students is to make them realize that the classes don’t matter anymore and sounds like you completely mastered the classes don’t matter think pretty early on i mastered it too early and i like there were times i had to like you know kind of get out of that so i’d actually pass my classes now when you say you love taking things apart i can’t help but wonder if there were any any days where you took something apart and uh you got really stressed out that you couldn’t put it back together and somebody would be really uh upset about it one of my favorite memories was a printer and it was this was back in the days and laser printers really expensive and it broke uh and of course like you know there were no manuals and i i think i was pretty sure as to what broke there’s a little thing you could see down in the middle of printer that just mechanically broke i thought how hard could this be and i still remember like i had parts strewn from that printer around my parents house this was in high school you know each one labeled there were probably like 300 parts i had to take apart just to get at that one part i saw and uh that stressed me out pretty good because i you know by the time i got it all back together i was 100 sure but it actually did work and it actually actually worked at the end of the day um it didn’t work quite as well as you know brand new but it’s printed which it wasn’t doing before i took it apart so that’s an awesome story i don’t i never managed to take things apart that thoroughly or when i did i don’t think i ever managed to put them back together so that’s that’s pretty amazing another great story that comes to mind is like my first like near-death experience from rapid unscheduled disassembly of something i built and that was in high school or i wanted to break the newtonian physics record for how far you could throw up a uh a golf ball uh and so i built this catapult thing with you know a big old winch to tension everything down it did break the record but then i wanted to break that record and the process of tensioning this thing to picture a structure build out of two by fours right so you know maybe 20 feet tall with the arm up high so that they’ll have two by fours but anyway so tensioning this thing i mean firing it with the over tension it blew up and it blew up to the point where there was like splinters of two by fours stuck in the tree i was staying next to like like full-on they just blew off and stuck in the tree and i remember thinking to myself like okay if that hit me i’d probably be dead yeah that was that was a pretty formative experience too you’ve been taking things apart building things ever ever since you’re boring it smells like but then at some point you decided zipline is is the company you want to build what is zipline and and why why did you want to start zipline sure sure so so zipline is all about access to medical supplies and it’s uh you know at our technology core we’re a drone delivery company uh but really that’s not why we exist and this is something that i did for the first time professionally at zipline which was didn’t touch the technology until we actually went out in the world and found customers and were convinced that they really wanted this thing um and that’s really that’s why i answer that question that way because a lot of people assume we’re an ai company or we’re a drone company or a robotics company or where it’s and don’t get me wrong we are at a certain level um but we we exist today because of our customers and i say i don’t mean that in some way weird like we exist to serve them like literally we would have failed as a company if we didn’t find those customers first because what we wanted to do initially was not useful like we would not have made a successful business if we hadn’t brought customers in at the beginning to really guide us what we’re doing so tell me about that journey the company started because of the customers well when did you run into the people you’re helping like what what’s the story there yeah yeah so so after uh so i spent about seven years on something you you know about well ross willow garage had a wonderful had a wonderful journey there uh but at a certain point uh once once once the open source robotics foundation was off to the races you know i was done with that um and really wanted to find a next thing to do um and uh and to be clear not done with it because i wasn’t having fun but done with it because it was in better hands than i i was providing as you know i think well well so um and and this was actually this is actually sort of a bit of a soul-searching moment because i i knew i liked to work on things that i think mattered and i thought ross mattered a lot i thought pr2 mattered a lot um and and that’s why like i got so into it had so much fun doing it um and along the way i worked on lots of things that scaled from a business perspective better than they did but didn’t matter very much and and i just i knew i wanted to find something that did both something that i thought would matter that would get me out of bed and excited uh and something that we could solve at scale you know through a viable business that could you know they could really fund itself and and and really make that work um yeah and so how do we find customers so this was this is actually two years of my life doing random things it was actually a funny period of my life because because of willow garage a lot of people would come to me saying hey can you help me make a robotics company um and i did something which i had never done myself i told them this advice i’d gotten and never followed which was basically go find customers when you have customers come back to me and let’s talk right and as you can imagine almost none of them came back to me but a couple did and it was really cool to sort of see how those companies evolved and basically i just took that time to find something that i knew would check those two boxes met a bunch of potential co-founders looked at all kinds of things at one point with uh with actual keller my co-founder here at zipline we actually walked away from a massive nre check from from a potential path that was very different than zipline to be cleared didn’t involve flying robots or anything like that but along the way uh it was actually our family members um my wife’s an epidemiologist um keller has family in the public health space as well and they kept talking to us about hey you know you know back back when we started zip line uh amazon was talking about drone delivery right so this wasn’t like some crazy new idea at the technology sort of like what can you do with drones level um but uh our fam our family kept nudging us of like hey you’re looking for things we keep hearing about these international vaccine campaigns we keep here that fail on logistics we keep hearing about these international efforts to you know eradicate various diseases or even like treat basic things like diarrhea in certain parts of the world that just fail on logistics uh go look at that um yeah and we did it was awesome we spent a bunch of time in central america and in africa getting another problem and um yeah i’ll be honest i’m a pretty skeptical person so i spent time like i was assuming okay we’re gonna go meet these customers and we’re gonna learn a thousand reasons why we can’t actually solve the problem uh and anyway the opposite happened and so the rest is history and we went for it and just because not everybody will know ahead of time when you say the rest of history you went for it what did you go for what did you build yeah sure so um so the basically if you think about this from a nuts and bolts like what our customer cares about we do fulfillment and delivery on demand of medical supplies right and by the way we started with just doing blood right and blood is this really special medical supply right it has a very short shelf life it’s rare everywhere in the world you have less blood than you need and because that’s such a short shelf life as few as seven days for certain blood products and there’s so many blood types you need to get it to the right person at the right time if you guesstimate where to send it most of it will expire before it’s used um and so what we built the first thing we built was a drone delivery system that enabled us to centralize uh blood storage in in basically one location and then fly very long range uh uh with these uh basically small aircraft small airplanes picture a small fixed-wing drone that flies out delivers the blood comes back and gets the next order and does the next order of blood and on demand right so basically when they know what type of blood they need and how much blood they need they would they request that blood and that’s what we built initially where many versions later and now we deliver vaccines and other medical supplies at much longer range and bigger scale but that’s what we built now one of the crazy things i saw um when when watching the videos on the technology side right is well most people when they think about drones definitely when i think about drones i i think of you know quadcopter type drones that i might use to take some videos when i’m going for a hike and and you know maybe other people do it for skiing and so forth um the things you you can just buy off the shelf in a little box and it it makes videos of you um but actually a zipline drone looks very very different can you describe what what it physically looks like sure so our drones are they look like small airplanes um about 10 foot wingspan and uh they deliver a box about the size of a cake box to kind of kind of make it physical um what do they look like they have red wings and a white body and they’re launched off of a catapult um and they land in this crazy contraption for for landing uh but all of that is done it’s a fixed wing versus a quadcopter for range which is what our customers care about uh if we launch with a catapult to make the plane as simple as possible for cost which again was what our customers care about and landing same thing right in order to have something that can fly really far really efficiently through all kinds of storms you basically don’t want to put any complexity on the drone itself you don’t have to so you don’t want landing gear you don’t want any fancy system on the drone for landing so we have a contraption on the ground for landing um and uh yeah that’s sort of the nuts and bolts of it if you will and let’s say as an engineer i was absolutely fascinated by by the way the drone takes off and lands is any chance you’re able to maybe pull up a video and talk us through what’s what’s going on when a zipline drone takes off and lands because it’s not what i think most people think of when they think of their drones they have played with or when they sit in an airplane that flies them it’s quite different so at our distribution centers we have inventory on site that’s really important because we deliver you know in a matter of minutes from an order coming in you saw a quick version of how the order’s packed uh then put into the body of one of our drones we call them zips i might call them zips accidentally i always forget to call them drones um and uh that goes in the launcher you put a wing on uh on that on that drone then a battery pack goes in there’s a little pre-flight process and then it gets launched and this launcher is a lot of fun this launcher has a motor and a super capacitor bank that takes that drone you saw there to zero to 110 kilometers an hour and a quarter second wow and it’s it’s about as much power as flooring and entry model tesla to launch that that fast it just flies right by and it drops the package uh during that flyby this is slow motion now just to be clear i just want slow-mo on you so um the the package has a little paper parachute on it uh this is actually a very old this is a this is a one our very first version so this delivery box is about the size of a shoe box our new drones are about the size of a cake box but it’s basically the same idea there’s actually a lot of technology behind the scenes to do that well uh to to basically measure and estimate the wind speed and direction and compensate so that package ends up where the customer wants it not in the tree um but yeah that’s what a delivery looks like then the the the drone flies back of course automatically and lands in this and this system uh we kind of call this an upside down aircraft carrier and uh the poles snap up and grab the plane out of the air and they’ll do it even in crazy stormy conditions now you can see that was slow-mo there so you can see it i’ll let this loop again so you can see it uh so am i seeing this correct canon that the plane is just flying like normal but then it gets snatched by a cable that somehow brings it to a stop almost instantaneously exactly so the plane just flies by it’s actually not trying to land it’s trying to just do a very safe flyby and then there’s that line between those two poles and those pulls will snap up right at the right moment uh to basically throw the line between those poles into the into the tail hook of the plane uh you can see it right amazing now why not just have a regular landing like when we take a flight into a regular plane there is you know the plane just lands and why not set it up that way sure so it’s a really really really a couple reasons one um if you’re looking at so one of the things our customers care the most about is reliability right they want us to be able to do this day and night storms all day long and if you look at what determines the reliability of an aircraft you know even passenger aircraft one of the hardest things to deal with is what they call a hard landing so is you know it’s a some of us experience them right your plane’s coming in and hits the runway a little harder than you’d like right that the the mechanical stresses that come from that in your plane are very very hard to predict and model um it’s very hard to build a small aircraft especially even a large aircraft will have a very predictable life in the face of hard landings so this is designed from the ground up to not have any hard landings uh and we’ve shown this out past five nines of reliability that zip will either do a flyby and come around again which sometimes happens uh or or or have this very repeatably gentle load case right into the tail we call the tail boom that thing at the end of the plane with the hook on it um that’s one reason uh cost is another keeping the planes very low cost is important we’re doing you know hundreds we’re up to hundreds of flights per day per distribution center so we prefer to have some costs on the ground and make every drone less expensive um so we can so the economics work out for our customers uh and uh yeah and then that storm piece is really important as well um you know just like it’s it basically makes our operators they don’t have to worry about what happens right because a small aircraft like this on a runway the way we land in a storm literally can just get blown off the runway oh wow and having a system like this like eliminates those kinds of challenges that’s amazing then and it’s also super compact as a consequence it seems like you need almost no space to to set this up is that right absolutely one of the reasons that that recovery system we call it that the landing system is so tall so we can be right next to buildings wherever wherever it makes sense uh whatever is a good central location right next to our supply that’s where we want to set up our distribution center one of the uh things i noticed as i was you know researching the latest happening at zipline is also that you actually have bono involved i mean youtube’s bono um what is bono doing at zipline oh he’s fantastic so he’s he’s a member of our board board of directors uh fun fact it’s the only corporate board he’s ever been on uh and he uh you know zipline is solving like again this bono has committed his career to solving some really important public health challenges uh and zipline is uniquely uh positioned to solve what we think it was like the unsexy part of that which is like you know logistics get the supplies where they need it when they’re needed in a way that well basically has never been solved at a really global scale before um and yeah so his role is fantastic he helps us get connections with with the right people to to get operating the right places um and obviously his credibility and his um uh you know he he’s just he’s done so many things in public health over the last 20 years he helps us see around corners that it was really important for the company i agree i mean it’s awesome to have somebody like that on board but how do you even get them more like do you just reach out and you say hey i love your music and and your other work also and can we talk i mean he must be hard to reach i mean i think it’s like it’s like anything in a startup right it’s all about relationships right um it’s all about relationships and then over the years we’ve built relationships with people he respects and done things that he you know that the guy on his radar screen and and that’s how we connected yeah wow so he he was kind of aware of what you were doing uh before you talk to him then yeah it’s it’s it’s like what we do this logistics and public health is there’s so few people focus on innovation there that if you care a lot about public health you’ve heard about zipline for sure how many deliveries have been done with sips we’re well over 100 000 deliveries now as i get it that’s that’s counting a drone flight as as one delivered because i also saw something about millions of vaccines being delivered for covid so we’ve done millions of vaccines pre-covered we will hopefully do a million vaccines of vaccine deliveries of covet here just a matter of months we’ve already done tens of thousands of covered vaccine deliveries just actually in the last week we just got our in ghana we just started delivering cobia vaccines uh which again like we got our training wheels delivering well non-covered vaccines we learned how you know cold chain challenges the chain of custody challenges we were really good at that now uh and then when they called us up and said hey can you do covet now like we were ready uh and now we’re doing that at scale which i am personally really excited about because you know he in the states you know my parents i mean it was just getting them vaccinated was just insane right the trying to figure out the schedules and the like but in ghana that’s not what’s happening we’re sending directly to health clinics your doctor calls you up and says hey i have a vaccine come on in like that’s how it should be right like that’s how it should be and someday we’ll be at that level here in the u.s too uh but we’re just so excited to to it just it’s like so obvious right like if you can deliver right to the doctor especially the doctor who knows you personally it’s just it’s just such a better health experience and just easier and yeah just it just works yeah it’s gonna be amazing i mean there’s some so much logistics to be i mean in the us all you hear about for covet is the logistics right now for the vaccines and you’re effectively solving the core of that in africa which is so exciting yeah by the way one little fun geeky piece of it but the big challenge with logistics with trucks is usually making big deliveries which means you’re where you deliver to they have to have good refrigeration or freezers right but if you’re doing on-demand delivery so like in ghana we’ll typically deliver enough for a couple days maybe even a week of supply uh the vaccine we’re delivering there can you be stored in just a normal fridge for that time frame it’s no challenge right and so it basically reduces by literally we’re talking about you know thousands of delivery sites so it reduces by a thousand the number of well freezers you have to maintain and stuff like that which it just makes things easier right it’s just way more flexible that’s amazing now when when you said you started out with solving a problem like finding the problem you really want to solve and you started in africa because that’s where you saw um that you could really go go help by by setting up a zipline there but africa is gigantic i mean it’s it’s not it’s not just just a country right it’s it’s a whole continent there’s so many countries can you say a bit more about i mean you can’t just start in africa you go it’s got to be a more precise plan i imagine yeah oh i mean the early days it was you know it was it was it was start-up precise right so in the very early days you know we started basically networking and meeting with these different governments uh some in central america some in africa uh we learned very quickly that customers like this require a lot of attention and you know back then we were just you know literally five people so we couldn’t you know give them all the attention we needed so we we got started to get smarter about okay which of these countries we’re talking to should we start with we narrowed that down to three um and and then we and then because we’re a startup and we were trying to get a government contract we we knew we had to keep at least three right you gotta you gotta manage your risks which turned out to be really uh uh really important because up until maybe three months before we launched with rwanda our first customer uh it was we were sure it was gonna be a different country like up until close but that contracting process basically got stuck in the mud and had to kind of reset uh and rwanda said hey we want to be first uh and and that contract was going really well and we ended up launching in rwanda um but yeah it’s again it comes back to the relationships that networking and yeah as you can imagine though we were layering a lot of sort of technical data onto it basically trying to figure out you know which which where we where were we going to be most successful for your first customer if it’s not successful as a startup you’re in trouble we were looking at um we’re also you know looking at like integration challenges you know we make planes right we had to rwanda’s very high altitude country we had to make sure we could fly at those high altitudes all kinds of interesting things like that had to be sorted out to kind of make sure we would have a good successful launch customer you said three countries rhonda is one of them which are the other two that you started off with so we start out with rwanda tanzania and costa rica got it wow yeah now that’s a bit spread out i mean did you physically go visit and spend time at all three places or yes absolutely absolutely and our head of business development spent a lot more time our head of business development was never home in early in the early years uh yeah yeah now that was that was very challenging um yeah but it was challenging but it also wasn’t that challenging i think a lot of people think of africa especially as like a difficult place to travel in and things like that i think that was probably true 20 years ago but it’s it’s easy now i mean you can get to africa and you know it’s pretty much like getting to anywhere in asia it’s nice and easy and um it’s the world is nice and small it’s way smaller than you would think uh yeah but you know it’s also startup so like you know not sleeping for the first few years is part of what you should expect to do so i i can attest to that um now one thing that struck me is that when you say you start for example in rwanda it’s not just that you start in rhonda as i understand it you actually build a relationship with the ministry of health of rwanda so all the way at the top can you see a bit more about that any country we work with it’s there’s a whole bunch of stakeholders that are involved right so there’s the obvious ones like the regulators air space regulators uh and as well as the met the the the health industry are basically the equivalent of like the fda style regulator uh then there’s the actual public health systems that we’re working with um and then you layer on top of that the political layers right because nothing big happens in a country without political basically support right no no no one in a public health system would take a huge risk without political support what was so cool about rwanda specifically is when we went in to meet with uh the minister of health um you know she like well we’re actually worried at first because she’s starting to text on her phone we’re like what’s this all about uh we lost her right but she in walked two economists uh and like from the very beginning we worked with those economists to literally put together a basic health impact analysis as included in economic impact analysis um and like that was their step zero based on really good data that they had and they put that case together and then once once uh the minister of health had that then she went to the president and said hey we have to do this right and even later in the process when the department of defense had to sign off on sort of american drones coming you know having her in the room saying you got to figure this out and here’s why right here’s why this is so big for us that was absolutely huge and yeah and then from there that’s that then you start working through the contracting process figuring out what you’re actually going to do and how you’re going to charge and and all that kind of stuff and in most of the countries you operate uh like rwanda and ghana you know you’re talking about political approvals involved right so you have to you have to you have to pitch this and you know there’s a whole process usually votings are involved it’s it’s pretty wild to get through through the other end yeah but there you are and and what i also noticed i don’t know that’s part of how it all played out but when i watched the videos it it seems you get a lot of uh you hire a lot of people locally and it doesn’t look like you bring in a large team you you set up shop and you start finding people locally is that right absolutely we we actually have a place so first of all all over zipline rwanda and all of ziplin rwanda is run and staffed and led all by locals rwandans and ghanaians um and in rwanda we have a place at our first distribution center we have a thing we call zipline academy and so even our operators that are now operating here in the u.s we send them out there to get trained uh and then they come back here to operate took off so well that that’s where the training happens now even if you’re yeah i mean it’s where you want to get trained right because it’s the highest volume operation uh it’s also i mean it is just disciplined and elite it’s really cool to see it’s the kind of thing where like i there was this great there’s this transition when i very first days when i was out there right i could help out and like pinch in and things like that but now when i go back and visit i’m not allowed to touch anything i’m not certified up to the standard that they operate at wow that’s quite the story locked out of your own uh operations it’s it’s it’s yeah it’s it’s it’s amazing it’s awesome now you started in rwanda how many countries are you operating in right now yeah so today we operate in rwanda uh we operate in ghana and we operate here in the us so also in the us uh i think i think that’s less well-known can you say a little bit about what what’s happening in the us yeah so our operation in the u.s started back in may and it was actually accelerated by covid um and uh yeah and to be clear in the us we’re like we’re like three years behind where we are in africa in terms of scale which is uh we’ll catch up don’t worry but we’re working very closely with the us regulator to get to do that uh we started doing uh we started doing the very basics um there’s actually there’s actually some poetic uh balance here so as i mentioned when we started in rwanda we started with blood right because blood is rare it needs to be delivered on demand right where you need it uh when we started north carolina in may during covid uh we started delivering ppe because it was rare it needed to be delivered exactly where and when it was needed um and uh so it’s a we we have we have a known story for how we start a new country uh just unfortunately in the us that meant delivering masks and things like that to healthcare workers wow that’s a great story too though one thing i also noticed for the us i don’t know um if that’s the same kind of things you’re delivering was something about walmart mentioned in one of the articles yeah so we’re launching with walmart later this year um we have a handful of other other other operations getting built in the us right now which is uh which is really exciting uh walmart’s one of them uh walmart will be delivering what they call their health and wellness products initially so this is basically what you’d expect like a cvs pharmacy um and uh yeah it’s it’s it’s it’s a it also represents a really exciting shift for zipline which we’ve been going through in africa just starting as well which is so in africa we started with hospitals and then did health clinics but now we’re moving to patient home delivery and with walmart same thing so walmart already does on-demand delivery you know doordash style we call it you know people in cars and that works for you know five kilometers or so maybe eight kilometers uh but we can go you know 80 kilometers uh and so basically we’re enabling that kind of on-demand pharmacy delivery to be for folks who are in just way more remote locations wow now as you’ve done so many deliveries at this point um i’m curious are there are there any stories that send stand out to you of you know certain deliveries where it just felt i mean i imagine many but some must really stand out that feel really life-changing as you do this blood deliveries vaccine deliveries anything you could tell us about your personal favorites sure i have one geeky story to tell and one really really kind of emotional story to tell the geeky story was fun so our test sites are here in california and we’ve you know we’ve been flying all day every day at these test sites long before we started in africa to work you know work out all the issues system uh but it’s really important that our test sites that the drones fly you know where you can see them so they kind of do circles and stay near stay where you are right and so if a drone at a test site ever were to like fly over the horizon like that’s a bad thing like that something went wrong i still remember the very first time in rwanda our very first uh deliveries to our very first hospital the first time i saw a drone just like take off and just disappear like my heart just skipped a beat and like i was like something is wrong this is not right uh and then of course you know it was a pretty short delivery but like 10 minutes later it comes back doesn’t have the blood anymore and it was like wow that just happened holy smokes yeah and uh so that was that was that was one of my more formative like and traumatizing geeky level experiences um you have to really trust it beyond what you can see exactly just we just went it just went um and uh one other fun fun anecdote about that flight this was a very beginning and we didn’t want to fly in rain because we want to kind of ease into it so we looked at all the weather data sources and said it won’t rain so but that that plane come but came back soaking wet so it’s like oh there must be some rain over the horizon we didn’t know about anyway um wow that was a fun moment i was like uh but yeah i think one of the more personal ones was meeting a woman uh this was lady that same that first year this woman named alice and she we actually ended up delivering to her twice believe it or not so she we delivered to her while she was uh having postpartum hemorrhage uh while delivering blood uh sorry delivering a baby she went to postpartum hemorrhage and we delivered blood for that case uh and then six months later she got malaria and one of the one of the uh sort of things that happened to malaria can’t happen is you develop anemia which also requires blood and so she like we delivered to the same woman twice and so she had to come check us she had to see what this was about and so she actually came by and visited and that was that was a pretty surreal moment to meet somebody who like you know who heard it like in her words like she’s like yeah i heard about this crazy thing called you know i heard my was what’s a drone and she really wanted to see what it was so she literally just found out where we were and came by and knocked on the gate and then checked it out that was that was that was pretty that was pretty special yeah that must be amazing and when you say one step removed you’re really 50 to 100 kilometers removed right so this this woman came a long way probably to to meet you yep she did yeah wow now these are great stories i’m curious about stories that or maybe things where you really struggled and things didn’t really go according to plan um any that any that are your favorites there yeah oh struggles so so i met my wife uh skiing uh she’s costa rican uh she’s a doctor and an epidemiologist anyway skiing so my sister my sister we both know my sister through berkeley and uh and my wife has a epidemiology phd from berkeley anyway so she on this ski weekend we go skiing and my wife never skied before so anyway i basically spent the weekend teaching her to ski and one of the things when you’re learning to ski you always tell people is you look where you want to go don’t look where you don’t want to go right stare at the tree you’re way more likely hit the tree if you stare at the place that doesn’t have the tree and you’ll go there and so she made fun of me because after willow garage i had this and i was looking for my next thing i wanted to do i had this list of things that should not be in a startup and they included things i had everything on the list you don’t want to start up i didn’t want regular regulation i didn’t want government customers i didn’t want hardware i didn’t want safety critical i didn’t want safety critical hardware and satrio software and so i had this list going and when i was telling what after the first trips looking at zipline uh looking at what became zipline i was like i’m really thinking about this and she started making fun of me she’s like you were staring at that list too much because everything on the list of don’t put in startup is in this potential thing uh and i think it’s uh anyway so it always makes me laugh because a lot of the really challenging things about zipline have been in in that world right selling the government’s just really hard to do the regulation stuff just really because like we work in a regulated environment without regulations which is like the worst right there aren’t drone regulations that work for ziplines so we’re always every country we go to we’re working to basically create the regulations that enable zipline and that’s um that’s been an adventure and then of course all the safety critical aspects of this has been really wild all of this sounds uh like it has many many challenge many layers of challenges maybe being an engineer in in training um i’m really curious about the technical side um and some some challenges that you might have run into there because i mean you don’t just build a drone like that i mean you you gotta somehow decide what drone you even build how you’re going to build it um say a bit more about that how does this become what it is now this is fun because when we started zip line i actually thought i was going to buy drones and then maybe modify them for package delivery and that’s how we get going and so the first kind of like uh let’s call it rough lesson was talking to all the drone manufacturers who could who could potentially build drones to do what we wanted to do and basically learning that none of them worked for what we wanted to do i still remember the most the closest we got to that was a drone manufacturer who wanted to charge us 200 000 a drone with a 200 flight warranty if we didn’t fly in the rain and it was like it was it was just like so impossibly not you know useless for what we wanted to try to do uh that was like oh wow we actually have to start in order to our first customer we have to build a drone ourself um i remember that was one of those like wake up call moments of like okay there’s more work here than we thought can you contrast the 200 000 with the rough cost of building a drone yourself sure sure like the the i mean our drones are now it’s a little hard to like describe the drone itself but think of our drones as like low very low tens of thousands right so and they also last a very long time and they fly through rain and all the time and they’re very easy to maintain and they have a lot of other things that we could not get off the market if you will um yeah um one really interesting technical challenge like i think the oftentimes like we like if if you’ve been trained by anybody right in engineering they teach you to think about risk right so you’re kind of thinking you’re like making your list so like what are all the things they’re going to get me and one of those but what really burns you is the things you don’t really think are going to be big risks right and so one of the advantages we have is we have a plane and a plane flies really efficiently and that’s you know if you stand 100 feet back from our plane it looks a lot like all the other planes right in terms of got a wing and a body and a tail and you know you’re good to go um and so you know one of the things we wanted to do was fly through weather and so we went to the experts in the industry uh places like nasa and said hey like how do we make sure this thing can fly through weather and they gave us their models they’ve been developed over many years of how to simulate and how to basically set your requirements for flying through weather um and we did that okay followed all that follow their their recommended design practices their requirements and so on uh we went to rwanda we started flying and it didn’t work right we have a parachute landing system on our drones uh we were parachute landing way too often way more often than that that analysis and modeling suggested we could and then we turned around and really started thinking about the problem went back to those experts and we’re like okay how did you build these models and so on and it turns out we were actually doing something that basically nobody does nobody flies near the ground in storms um it’s just not done and now in our head even i was like yeah but like you know uh search and rescue helicopters they must do it so get on the phone literally talk to the team in the california coast guard talk to the team on zermatt right the biggest mountain is the swiss alps like their rescue team and both of them said the same thing oh we don’t fly in weather anymore that was 20 years ago too many people died we stopped we stopped we like we wait till the weather clears right and so we’re sitting here going like okay we’re flying in this really interesting uh domain that nobody flies in and anyway so we had to go learn about this domain develop our own models and actually pretty much completely redesign our drone to actually be able to survive in the crazy sort of updrafts and downdrafts and sort of stormy conditions that happen and this we call them energy concentrators as you have winds flying over hills and mountains these crazy basically energy sources uh we had to redesign the plane the control system pretty much everything to actually do this reliably like we did today and we anyway so those are two steps first i didn’t even think i’d make the drone and then i thought well at least nasa knows how to fly in this in this kind of environment so i listened to them and both of those things turned out to be learning experiences we are dropping new interviews every week so subscribe to the robot brains on whichever platform you listen to your podcasts ai of course is powering many many systems these days and i’m curious uh when today as it you know flies off and comes back what kind of ais involved in that process today yeah so today in operations the the ai in operations is mostly around weather um now there’s no coincidence that our biggest reliability challenge today in operations is also weather and so that’s what we put that’s where we put our big guns in terms of what we really solve problems around in our development process at our test sites and things it’s around aircraft detection and then soon it’s going to be around things like ground obstacle detection and delivery site selection and getting that package right where you want it on your doorstep one of the things that happened kind of in during i mean zipline already existed but kind of started happening last five years really is ai has really broken through and got a lot of new capabilities that weren’t possible before and image recognition speech recognition and so forth right and all powered under the hood with with deep learning and i’m very curious if you know deep learning plays a role at zipline now or what what what do you see kind of the future of deep learning in the context of zipline absolutely absolutely and this this comes back so so ai in general’s role at zipline is growing so in the past it’s only been offline and that’s been very intentional because i was dreading having to explain you know ai basically to regulators uh in the early days because it was already hard enough to explain the more way more sort of deterministic stuff we were doing um but this the ai has a big role in the future of flying for us uh and it’s really connected to that that march we’re going down starting with hospitals health clinics and outpatient home delivery right because with hospitals from one distribution center you know maybe 10 to 20 hospitals we can reach depends where we are in the world health clinics you know now we’re talking about thousand health clinics but now they’re in patient home delivery we’re talking about hundreds of thousands of delivery sites from one distribution center and so a lot of the things we did uh that make economic sense and scale for those those those uh for health clinics and hospitals don’t make sense anymore and really aren’t going to create the customer experience we want for this next level of operation so uh and it really it comes down to things like well detecting and avoiding other aircraft uh detecting and avoiding uh ground obstacles you know cell towers and power lines and trees and things like that uh and then uh the last place also related to perception that we’re working hard on is really around the delivery experience and being able to get dealing dealing with that you know the people want their deliveries in their backyards uh from drones and so being able to actually sort of navigate if you will and find a good delivery place in their backyard that’s going to work and be safe this is a big part of where we’re going with it um and so uh i can i’ll talk a little more about that in a second the other place that we’ve we’re using deep learning machine learning is actually weather forecasting we have our own weather forecasting we call it now casting because we care about like basically what’s right in front of the drones and we feed data in from satellites and from our drones to guests uh there’s basically a very particular type of weather event we can’t fly through uh it’s called a micro burst and so we really wanna oh what’s microburst so micro versus what happens right at the beginning of a thunderstorm when that thunderstorm head forms um you end up with this huge updraft and and then outside of a downdraft and i say huge updraft i’m talking like we we have recorded data at like 40 50 meters per second of the the vertical component of the wind on the middle and on the outside it’s you know coming down um and uh well let’s just say that’s more than our drones can handle then that if we often not every time often if we end up in those we end up parachuting the drone um and so being able to forecast those has been something that that we actually end up rolling ourselves because this is another one of those things that we care a lot about uh it’s a very one of these near ground effect uh things uh and it turns out the the rest of the weather industry isn’t working on it so uh uh the the brilliance of ai today one really good engineer uh has has made a huge impact on this for zipline already i’m kind of curious because when we think about ai if we talk about this specific one for a moment sure it’s like the typical model is that you you collect data you have observations and then the ai often a neural network has to turn those observations coming from satellites radar and so forth turn that into a prediction yep but the way it’s it’s trained is by also having access to ground truth predictions that you have as examples and so it seems almost like a chicken and egg problem because you gotta predict those you gotta find those microbursts so you can collect your data so you can say a little bit about that sure so there’s basically three ways that we there’s three sources of ground truth for this problem that we use one source of ground truth truths are our historical logs from every flight we’ve ever done uh places where we’ve flown into microburst or flow near them uh because if you’re right in the center it’s it’s one thing but if you’re near them you can still see the sense them that’s one source uh the other source is is historical satellite imagery right so you know you have the satellite you have the satellite imagery and you can actually these microbursts have very unique uh they look very unique visually in these in this imagery and so you can see when it happens and then you try to train a model on basically what came before uh you try to predict based on what came before when it’s about to happen uh and then there’s another thing we’ve been using which is basically um so computational fluid dynamics right lets you sort of build like a sort of a 3d model uh of how the wind moves over surfaces and so this is historically done over like your arrow your airplane wing uh but we’re also doing this over terrain so we actually have 3d models of countries and that lets us build very detailed models that helps us basically fill in the gaps between those other two sources i mentioned right so if we’re flying if we have data from a parachute landing where we flew into one of these uh we will go back we take the data from our plane flying in we take the weather data from that you know from that moment of time that whatever we can get our hands on we run that through these sort of macro uh models of computational fluid dynamics models to predict what was actually happening basically ahead of where we parachute landed right the thing we didn’t quite get to see right uh and then that helps us fill in some of the gaps so this is fascinating because i mean this is this is definitely not a mainstream ai problem that you would find in the research literature this but it is exactly the same kind of methodology as you would use for the kind of more commonplace vision and speech problems it seems exactly exactly and it’s also one of those fun sort of necessity problems right you know when we launched in rwanda and ghana this was not our biggest challenge for reliability right but over time we’ve gotten to the point where it is uh and now we’ve been tackling it um and uh and we like you said we have because we have this historical data it helps sort of jump start that chicken and egg cycle that we can use that to really uh well sort of mind that for any information that we can use to train these models on that note of data is are there other things that are on your mind for the future where it’s okay thanks to flying these drones we have very interesting data that can allow us to build more advanced systems that can do even a better job or something new yeah yeah i know absolutely and i the so you well already with weather generally it’s something we’ve been thinking a lot about because our drones sort of represent these flying uh weather stations that that are very good uh we care a lot about them being precise so our planes fly well uh and and and they and they also capture data that’s very clean so if you think about the weather challenge in general weather forecasting in general um you know it’s it’s all about where your data comes from and if you’re collecting data on the ground that’s like the that’s like the worst data ever because it’s a wind especially if just it’s all affected by what’s around you right so a weather station next to a tree is not as good as a plane nice in the nice sort of free stream of the air um and we’d love to sort of use that to help with weather forecasting in general is that what you’re asking are you you’re going more for like internal yeah i was thinking just generally kind of things that because you have the data new ideas that have come to your mind as you know opportunities you you could leverage the data for yeah yeah so i’ll be honest most of the most my efforts of how to use our existing data are really about how to get past the next hurdle for the company exactly exactly uh so some of the things we’ve looked at which are starting to bear fruit are essentially using so we have logs from every flight we’ve ever done ever and we and we’ve invested quite a bit in the data systems that make uh using those logs really um easy uh and of course like a lot of what those logs get used for is like intentional right so let’s say we have a problem and we use those logs to study data to help us root cause a problem another example of intentional of course is a design problem we want to understand what are the loads that a wing sees right and so we can look through all of our flight data to get a nice really good statistical picture of what loads of wings sees but of course there’s there’s definitely information in those logs that we don’t know to go look for but there’s a fantastic amount of data and so we’ve actually we’ve been starting to look at basically sort of anomaly detectors to help us flag like what is weird right um which can now be a predictor sort of a near-miss if you will signal uh that hey we’re going to have a problem coming and we treat this as a near-miss now we dig into it we may be able to basically head off a reliability challenge that we don’t understand yet that’s sort of order magnitude beyond the reliability we are today if that makes sense right it seems like you could do preventative maintenance in really nice ways at the scale that you’re starting to operate exactly exactly we’re really excited about that too yeah going back to to what you mentioned earlier the other main application that you’re already seeing now for for deep learning and ai in general uh on the image recognition side and you said something about landing or destinations can you say a bit about that sure absolutely so there’s a number of things we’re working on so um let me sort of paint a little bit of a picture so the first thing we’ve been working on using ai and perception is around aircraft detection uh and we actually we have our first submission into the fa uh in november for that which by the way fun fact as far as we know as far as the fa people we’re working with know this is the first time the fa has ever reviewed uh basically a trained model or an ai based ai functionality for use in the national airspace uh so this this will be a this is a this is a this is i’m saying that for two reasons because it’s not approved yet to be clear but it is approved uh it’s a big milestone for the fa as well because they’re they’re basically uh sort of learning how to think about this kind of performance-based uh way of looking at these problems um which is uh which it’s a big lift uh it’s it’s i mean to be clear it’s a big lift for them and i know that because it’s a big lift for us even for us to get the confidence of like hey we are confident in the safety performance of the system in order to then submit it to them just getting that done internally was was uh uh it required people better at math than i am well you got some really great people apparently um now when i think about this i mean you build a vision system to recognize other aircraft um what does that look like i mean how do you build a system to detect other aircraft yeah how do you build a system detect other aircraft so uh the first thing i’ll mention is the thing we submitted to the faa is not what you think it is it’s not vision and i can’t tell you what it is quite yet but it’s very cool and it works incredibly well so i’m very excited about it but that’s that’s enough teasing that idea we actually do have we actually i do have vision capabilities in development right now and those are focused around what we call the delivery experience so this is finding the spot you know at your house or your backyard where we can deliver right so we we want to see what we want to find the trees so we don’t put it delivering to the trees is a bad customer experience uh we want to figure out like where’s a good you know we don’t want to deliver in your pool uh there’s a lot of things like that that they want to really understand well um and so a lot of that is uh you know it’s it’s gathering data from just operations using our sort of existing methods of of getting operating uh it’s it’s it’s also gathering data sort of by hook and by cook right like we put we’ll put a camera payload on a fishing pole or basically a sort of extendo pole and send it home with people to kind of build out data sets and and things like that to help make sure we’re getting uh um diversity in our data sets uh and yeah and then it’s just a lot of really understanding the problem uh and then being really clever about how do we sort of reduce the dimensionality of the problem how do we make this problem tractable right how do we how do we stop thinking about it is like a brute force like we’re just going to solve the whole thing to like how do we how do we make this something we can really get our head around uh train a network to solve in a way that that actually works for the customer and and uh and makes sense too that’s one of the last pieces i think really important when you’re dealing with people you know if it doesn’t make sense uh to anybody but the ai team like it’s not it’s hard people have to look at they have to be like i want that or i understand that i gotta get what’s happening and that has to come together in a reasonable way now one thing i’m very curious about a lot of the challenges of putting ai to work in in the real world relate to kind of the long tail always new things happen is that is that the case are you trying to address that or you know does a customer have to satisfy specific requirements ahead of time that you specify before you know it it’s it’s a place you can deliver it’s very much about the long tail and this is part of the reason we pushed really hard to find a use case and a customer that we could get operating without requiring ai first because what that means is you can kind of think of our ai-based capabilities as essentially adding one of two things improving our safety case right or basically improving our efficiency uh or in some cases enabling experiences we couldn’t have done otherwise um but when you think about a safety case for example like so safety cases sort of this in aerospace is this end to end you know safe sort of statistical argument about how safe are you um and uh and and there’s a lot of ways to solve the problems in that chain right and some you know some of them aren’t very efficient or aren’t very elegant but you do them anyway because you want you got to be safe at all times but because we have that baseline today we’re able to layer in these capabilities in ways that enable us to scale while maintaining safety for example which is a lot of what we’re doing um but without requiring perfection at the beginning right as long as now it’s important that we know how imperfect it is because otherwise you can’t stand behind the safety case but if you know how imperfect it is you can you can you can launch it you can operate on it you can learn about it um and then you can improve it over time which improves the safety case improves your ability to scale uh and that’s sort of you know we’re kind of in that dream situation where you know a lot of people describe it similar like tesla like tesla makes plenty of money selling cars and they can use that to then sort of basically improve their cars if you will and make them more and more valuable by adding a base capabilities which is a great way to enter the market versus having to have an ai capability that just like nails your product on the first go which is hard to do you talk about it so concrete about you know deliveries and so forth and i’m kind of curious um could i personally expect a drone delivery from zipline as just as a consumer just buying something online is that something that’s on the near horizon here in let’s say california i mean it all depends on your definition of your horizon but we’ll get we’ll get it done we will get that done for sure it will definitely initially be health related um and it might be as simple as like you have a telehealth visit and like while the telehealth is happening on your phone like the doctor sends you something directly to your home like that’s what it’ll be initially uh to homes here in the u.s um and you know some things like walmart if you’re in more remote areas uh and then we’re gonna go from there and keep like th there’s this awesome thing happening in the u.s around healthcare at home accelerated by covid and the reason i can say it’s awesome is actually africa is way ahead of us europe’s way ahead of us and i’ve got to experience myself a few times in my life and just how cool it is to have really good health care at home whenever you can uh and here in the states it’s going to happen too and we want to be a really big part of accelerating it because it’s just it’s just so much better it’s so much better yeah well i look forward to that um i i definitely know about the discrepancies between healthcare and us and europe so a lot of gaps there now when a lot of people think about drones i think the most natural thing they still think about today is dji drones because those are the the toys that you buy and you have fun with and it’s a consumer product right yeah and so i’m kind of curious how do you see the the broader kind of landscape um you know what’s happening with with drones beyond what’s happening at zipline and what are some exciting things that are happening i mean the most exciting thing that’s happening with drones at zipline and elsewhere is really around autonomy right it’s it’s really around these systems being able to take care of a lot of the basics of like don’t hit things uh whether it’s whether it’s a drone like what uh you know skydio makes which is sort of a dji that basically has a lot of autonomy in it so you could basically it’s hard hard to crash it almost impossible it just flies itself um and those kind of capabilities for uh for all manner of things the the the skydio has this amazing product out for insurance adjusters where basically you walk up to a house you push a button and it gives you like a you know this like you know super high resolution and imagery of the roof for that you know in a matter of seconds you don’t fly it it just flies itself and so there’s really cool stuff happening from companies like that around for industries like that so take here in california right pg e is like you know has this huge problem with uh starting fires right and you know last year they’re like look we have i forgot the number was you know i think it’s millions of of kilometers of power lines or maybe hundreds of thousands just a fantastically large number way more than you can expect with helicopters which is what they used to do and they came to the drone industry and said hey can we do this at scale right and we’re at the cusp of saying yes right and that really comes back to things like autonomy you know being really good at sort of the range and those kind of capabilities and and of course once we unlock these you know then the awesome next step that’s going to happen i’m very confident in our lifetime is the flying car step right we’re going to start we’re going to start commuting and flying drones and that’s going to be awesome too yeah i mean i think that that for a lot of people will probably the most impact biggest change in in their day-to-day life if they if they fly to work instead of drive to work though a lot of people working from home of course now but um and just think about commuting in 3d instead of on on the roads it’s going to be so fun it’s going to be so great and so quick it’s going to change everything the flying cars are going to change the economies in really big ways can you say a bit more well just think of the basics of just home prices right like you know but you know if you’re in the san francisco bay area for example right if you’re sort of near the big companies your home prices is twice as much as you are if you’re 20 30 miles away but when i say 20 30 miles away i mean like as the drone flies right and like uh you can buy a house for literally half the cost uh and that that drone flight will take you all of like 10 minutes uh but driving through the highway system can take you an hour and a half so people will pay for that really expensive house instead of instead of living over the side of the hills or whatever it might be what i’m personally very curious about but related to the applications you mentioned and curious if you’ve seen anything is um fire prevention and fire fighting any thoughts on that yeah definitely i mean the fire fighting is really exciting because just the tanker style operations being able to do that without people in those in those planes is going to be uh it’s i mean it’s just it’s an incredibly dangerous thing to have people flying these tankers into fires um going in and doing inspection for a spot like looking for hot spots also really powerful to do with drones uh this is one thing that i this is one thing that happens i think there’s a company boeing has a company that does this for firefighters at scale here in california it’s not very well advertised uh but they will fly these drones that are designed for military surveillance into the fires to look for hot spots um and it’s uh it really helps the firefighters know what’s happening in the fire and fire fight the fire much faster and more safely now seems like those are things we’ve all imagined are going to be part of our future and maybe the big question at least on my mind is you know when you talk about these applications what kind of timelines do you think because i think we’ve all seen the jetsons with flying cars and we know 100 years from now sure but what do you think that’s a great question i i drone delivery to homes in the u.s you know it’s it’s it’s going to start this year i i think i’m almost 100 sure of it wow this would be a more remote locations um i think the more the more dense locations you know three four years feels very practical to me um that’s where the technology is and i think i’m reasonably confident that’s where the regulators will be here in the us um i think flying cars like flying cars have a longer road to hoe but i if that took 20 years i’d be surprised i’d be surprised if it took that long i think it’s more like the 10-year time frame for a number of reasons um i think there’s there’s a chance 20 years of time frame for like where the with economics because make it mainstream and 10 years is sort of a more of like a premium service um but uh yeah there’s there’s not that much technologically in the way anymore of flying cars batteries today are good enough and they’re just getting better uh learning how to fly in this kind of weather is something that people like us are like really learning how to master um the safety and the safety for the safety systems around it like it’s it’s in a lot of ways it’s it’s you know more tractable from an autonomy perspective than you know well an autonomous car on the road right for a lot of reasons well just to be clear when you said flying is something people like us are mastering you mean is it going to be automated or are you saying that people are going to be actively flying no i mean well the very early flying car stuff that people will be flying those but that’s i i think that’s more from a regulatory sort of kind of walk crawl run than anything else um and uh when i say people are mastering i mean companies like zipline are getting really good at flying in this kind of weather and this kind of in this sort of low altitude like how to fly reliably um and i think that what in the long run you know autonomy is going to win there there’s there’s there’s as my pilot friend is always saying you know when you look at it when you look at the cockpit in his little in his little two-speeder most of the buttons there he has to remind himself not to push because if he does like you know bad things will happen right uh and it turns out that like uh you know this this is something that zipline has gotten very good at is the the time like what we think we call it the low-level autonomy this is like we call it sort of even at that health and status of like hey you have a problem with this part of your hardware what are you gonna do about it like we have software systems that are very good at doing the right thing about that and they’re not they’re good at it they do it in a fraction of a second they make the right decision a fraction of a second and and they’re they’re you know they’re auditable they have all the features you want uh that the regulators want but don’t know they need yet um and to do this at scale and so um i think that’s gonna be a big part of it is is uh well it’s basically you know you have to learn this it’s all a learning process uh and and i don’t see anything technologically sort of fundamentally blocking flying cars it’s just it’s just it’s basically already here um and once the batteries get a little bit better it’s going to be here in a cheap way too that’s very exciting yeah much looking forward to that you do now another thing i’m really curious about you i mean we talked a lot about zipline but you also did another company before willow garage and i believe between willow garage and zipline you were part of a few other startups and some kind of curious zooming out um i mean there’s a lot of people out there that you know did love to start a company and and it generally seems a great way to have a lot of impact in the world by starting a company versus joining an existing company you have some advice that uh you can share from your experiences my high level advice on startups is pretty simple like do it because you have to right like if someone has to talk you into doing it don’t do it you won’t enjoy it because it’s hard but you and you will enjoy the hard if it’s something that you really have to do i mean you have to do it like you just like you can’t control yourself right you’re just really excited to go do that that’s the first thing um and the second thing is is you know customers first right and literally especially for your technology company so many technology companies make the excuse of like ah there’s too much technology to do here i need to make some demos or whatever before finding the customer i have come to believe that like that is never true like it doesn’t matter how complex your technology is if you find customers first and then develop the product based on the technology with those customers like the everything gets easier when i say everything i literally mean you know the investors i mean most investors want to invest in customers right if you have customers investment comes hiring great people like most you want to inspire someone to join your crazy adventure and you can talk about an actual customer with an actual need that you really understand that will inspire people to join you um product decisions technical decisions what you can cut and what you can’t right if that’s if those debates in your company are anchored in actual customers you can literally go visit and get to know those decisions become easy like literally all the hard parts of a startup become much much much easier if it’s all guided by those customers um and yeah i find if you do those two things it’s almost hard to screw up if you’re really listening to your customers don’t get me wrong as a geek it’s easy to act think you know talk convince yourself if you’re listening to your customer or not but if you really listen to them and are willing to like you know let them drag you a bit uh i don’t mean like go ask them like what do you want but go really understand their problem and like let that process of understanding their problem you know guide you off of your initial course like you you’ll be successful it’s it’s like success is almost guaranteed if you really listen to the customer keenan it’s been absolutely wonderful to have you on it’s been far too long since we got to catch up and i’m absolutely stunned by the tremendous impact zipline has had saving lives hard to imagine a better story of robotics being taken out of research labs into the real world thank you peter this has been absolutely fantastic it’s so cool to catch up and well anyway it’s also just been so fun to follow both your lab work and your startup work and uh it’s very inspiring and uh well literally educational we read your papers well i’m sure you managed to educate a lot of people listening in today but now if anybody listening in just like me would like to keep learning even more from kenan i highly recommend checking out the company’s website flyzipline.com and also highly recommend following keenan on twitter where he tends to share ziplines latest innovations and impact that’s being achieved on twitter that’s add kitten wyroback and the same goes for linkedin thanks again keenan cool thank you you .

Déroulement de la vidéo:

0.56 [Music]
0.56 today
0.56 here with me is kenan weire kieran is
0.56 the founder
0.56 cto and product architect of zipline
0.56 the world&;s first drone service that
0.56 delivers life-saving medicines to the
0.56 most difficult to reach places on earth
0.56 welcome kenan thank you great to be with
0.56 you
0.56 here peter it&;s so good to uh
0.56 see you again ken it&;s been such a long
0.56 time we used to spend a lot of time
0.56 together
0.56 back at stanford in our phd days and
0.56 you&;re working completely different
0.56 things than i remember
0.56 like the first time i heard of you and
0.56 saw you was
0.56 all of a sudden this robot existed at
0.56 stanford
0.56 that was on wheels and had two arms and
0.56 had a head and it was just like wow
0.56 this is so new and really what i&;ve been
0.56 hoping to work with forever and all of a
0.56 sudden it exists
0.56 awesome wow bringing back the good
0.56 memories from the old days okay
0.56 yeah that was quite the surprise um
0.56 and of course um right now you&;re in a
0.56 completely different situation you&;re
0.56 you&;re
0.56 not not a pg student at stanford anymore
0.56 um you&;re you founded zipline
0.56 and actually spent most of yesterday
0.56 watching
0.56 all the zipline videos that i could find
0.56 and it&;s absolutely amazing
0.56 what you&;ve been achieving there um but
0.56 before we get to
0.56 those achievements and what&;s going on
0.56 at zipline i&;m curious
0.56 uh where where did you grow up and how
0.56 was the path to get to that moment when
0.56 you
0.56 founded zipline i grew up in california
0.56 uh on a little lake uh in the in the uh
0.56 yeah this awesome little lake like spent
0.56 some of my time swimming
0.56 sailing little rafts we would build like
0.56 yeah
0.56 just doing all kinds of fun stuff and
0.56 how do you go from
0.56 growing up you know enjoying life at a
0.56 lake to
0.56 to founding zipline i&;ve always i&;ve
0.56 always been drawn to kind of like
0.56 first of all like building stuff taking
0.56 stuff apart like
0.56 that&;s just i don&;t know that&;s just
0.56 what i did for fun all the way even when
0.56 i was a little kid that&;s just i was
0.56 if there was if i get my hands on a
0.56 screwdriver i&;d take something apart
0.56 and you know getting good trouble for it
0.56 usually depending on what it was
0.56 that was just like i don&;t know there&;s
0.56 something about like just knowing how
0.56 things worked and knowing what was
0.56 inside and knowing sort of
0.56 what was behind the scenes that was
0.56 always just a big part of
0.56 just the fun for me um and uh
0.56 yeah i think the other thing that really
0.56 as i
0.56 yeah even in high school i really
0.56 enjoyed just if there was a problem that
0.56 i thought mattered
0.56 like i had to be passionate about it but
0.56 once i thought it mattered like you know
0.56 you
0.56 like that was just that was also just
0.56 where the fun came from
0.56 and i think even even in school i spent
0.56 a lot more time working on things i was
0.56 excited about than like
0.56 you know homework or classwork and
0.56 things like that
0.56 uh and yeah i don&;t know that that being
0.56 passionate about things that mattered
0.56 and working on them
0.56 got me through school and high school
0.56 and college and grad school really
0.56 effectively and
0.56 and i often joke with uh well i think
0.56 even back when we were at stanford
0.56 like i was spending more time building
0.56 that robot than i wasn&;t like doing my
0.56 class work and like
0.56 it was just always because i thought it
0.56 mattered i thought it really mattered to
0.56 try to build a
0.56 kind of a common platform for robotics
0.56 development i thought it just mattered
0.56 more than like
0.56 the classes i was taking that day so
0.56 yeah
0.56 i can tell you as a professor i think
0.56 the hardest part
0.56 of being a professor and advising phd
0.56 students is to make them realize that
0.56 the classes don&;t matter anymore and
0.56 sounds like you completely mastered the
0.56 classes don&;t matter think pretty early
0.56 on
0.56 i mastered it too early and i like there
0.56 were times i had to like you know
0.56 kind of get out of that so i&;d actually
0.56 pass my classes
0.56 now when you say you love taking things
0.56 apart i can&;t help but wonder if there
0.56 were any any days where you took
0.56 something apart
0.56 and uh you got really stressed out that
0.56 you couldn&;t put it back together and
0.56 somebody would be really uh upset about
0.56 it
0.56 one of my favorite memories was a
0.56 printer and it was this was back in the
0.56 days and laser printers really expensive
0.56 and it broke uh and of course like you
0.56 know there were no manuals and i
0.56 i think i was pretty sure as to what
0.56 broke there&;s a little thing you could
0.56 see down in the middle of printer that
0.56 just mechanically broke
0.56 i thought how hard could this be and i
0.56 still remember like
0.56 i had parts strewn from that printer
0.56 around my parents house this was in high
0.56 school
0.56 you know each one labeled there were
0.56 probably like 300 parts i had to take
0.56 apart just to get at that one part i saw
0.56 and uh that stressed me out pretty good
0.56 because i you know by the time i got it
0.56 all back together i was 100 sure but it
0.56 actually did work
0.56 and it actually actually worked at the
0.56 end of the day um it
0.56 didn&;t work quite as well as you know
0.56 brand new but
0.56 it&;s printed which it wasn&;t doing
0.56 before i took it apart
0.56 so that&;s an awesome story i don&;t i
0.56 never managed to take things apart that
0.56 thoroughly
0.56 or when i did i don&;t think i ever
0.56 managed to put them back together so
0.56 that&;s that&;s pretty amazing
0.56 another great story that comes to mind
0.56 is like my first like near-death
0.56 experience from
0.56 rapid unscheduled disassembly of
0.56 something i built
0.56 and that was in high school or i wanted
0.56 to break the newtonian physics
0.56 record for how far you could throw up a
0.56 uh
0.56 a golf ball uh and so i built this
0.56 catapult
0.56 thing with you know a big old winch to
0.56 tension everything down
0.56 it did break the record but then i
0.56 wanted to break that record and the
0.56 process of tensioning this thing
0.56 to picture a structure build out of two
0.56 by fours right so
0.56 you know maybe 20 feet tall with the arm
0.56 up high
0.56 so that they&;ll have two by fours but
0.56 anyway so tensioning this thing i mean
0.56 firing it with the over
0.56 tension it blew up and it blew up to the
0.56 point where there was like
0.56 splinters of two by fours stuck in the
0.56 tree i was staying next to like like
0.56 full-on they just blew off and stuck in
0.56 the tree
0.56 and i remember thinking to myself like
0.56 okay if that hit me
0.56 i&;d probably be dead yeah that was that
0.56 was a pretty formative experience too
0.56 you&;ve been taking things apart building
0.56 things ever ever since
0.56 you&;re boring it smells like but then at
0.56 some point
0.56 you decided zipline is is the company
0.56 you want to build what is zipline
0.56 and and why why did you want to start
0.56 zipline
0.56 sure sure so so zipline is all about
0.56 access to medical supplies
0.56 and it&;s uh you know at our technology
0.56 core we&;re a drone delivery company
0.56 uh but really that&;s not why we exist
0.56 and this is something that
0.56 i did for the first time professionally
0.56 at zipline which was
0.56 didn&;t touch the technology until we
0.56 actually went out in the world and found
0.56 customers
0.56 and were convinced that they really
0.56 wanted this thing um
0.56 and that&;s really that&;s why i answer
0.56 that question that way because a lot of
0.56 people assume we&;re an ai company or
0.56 we&;re a drone company or a robotics
0.56 company or where it&;s
0.56 and don&;t get me wrong we are at a
0.56 certain level
0.56 um but we we exist today because of our
0.56 customers
0.56 and i say i don&;t mean that in some way
0.56 weird like we exist to serve them
0.56 like literally we would have failed as a
0.56 company if we didn&;t find those
0.56 customers first
0.56 because what we wanted to do initially
0.56 was not useful
0.56 like we would not have made a successful
0.56 business if we hadn&;t brought customers
0.56 in
0.56 at the beginning to really guide us what
0.56 we&;re doing
0.56 so tell me about that journey the
0.56 company started because of the customers
0.56 well
0.56 when did you run into the people you&;re
0.56 helping like
0.56 what what&;s the story there yeah yeah so
0.56 so
0.56 after uh so i spent about seven years on
0.56 something you you know about well
0.56 ross willow garage
0.56 had a wonderful had a wonderful journey
0.56 there uh but at a certain point
0.56 uh once once once the open source
0.56 robotics foundation was off to the races
0.56 you know i was done with that um and
0.56 really wanted to find
0.56 a next thing to do um and uh and to be
0.56 clear not done with it because i wasn&;t
0.56 having fun but done with it because it
0.56 was in better hands than i i was
0.56 providing
0.56 as you know i think well well so um
0.56 and and this was actually this is
0.56 actually sort of a bit of a
0.56 soul-searching moment because i
0.56 i knew i liked to work on things that i
0.56 think mattered and i thought ross
0.56 mattered a lot i thought pr2 mattered a
0.56 lot
0.56 um and and that&;s why like i got so into
0.56 it had so much fun
0.56 doing it um and along the way i worked
0.56 on lots of things that
0.56 scaled from a business perspective
0.56 better than they did
0.56 but didn&;t matter very much and and i
0.56 just i knew i wanted to find something
0.56 that did both something that i thought
0.56 would matter that would get me out of
0.56 bed and excited
0.56 uh and something that we could solve at
0.56 scale you know through a viable business
0.56 that could you know
0.56 they could really fund itself and and
0.56 and really make that work
0.56 um yeah and so how do we find customers
0.56 so
0.56 this was this is actually two years of
0.56 my life doing random things it was
0.56 actually a funny
0.56 period of my life because because of
0.56 willow garage a lot of people would come
0.56 to me saying hey
0.56 can you help me make a robotics company
0.56 um and i did something which i had never
0.56 done myself i told them this advice i&;d
0.56 gotten and never followed which was
0.56 basically go find customers when you
0.56 have customers come back to me and let&;s
0.56 talk
0.56 right and as you can imagine almost none
0.56 of them came back to me
0.56 but a couple did and it was really cool
0.56 to sort of see how those
0.56 companies evolved and basically
0.56 i just took that time to find something
0.56 that i knew would check those two boxes
0.56 met a bunch of potential co-founders
0.56 looked at all kinds of things at one
0.56 point
0.56 with uh with actual keller my co-founder
0.56 here at zipline
0.56 we actually walked away from a massive
0.56 nre check from
0.56 from a potential path that was very
0.56 different than zipline to be cleared
0.56 didn&;t involve flying robots or anything
0.56 like that
0.56 but along the way uh it was actually our
0.56 family members
0.56 um my wife&;s an epidemiologist um keller
0.56 has family in the public health space as
0.56 well and they kept talking to us about
0.56 hey
0.56 you know you know back back when we
0.56 started zip line uh amazon was talking
0.56 about drone delivery right so this
0.56 wasn&;t like some crazy new idea at the
0.56 technology sort of like what can you do
0.56 with drones level
0.56 um but uh our fam our family kept
0.56 nudging us of like hey you&;re looking
0.56 for things
0.56 we keep hearing about these
0.56 international vaccine campaigns
0.56 we keep here that fail on logistics we
0.56 keep hearing about these international
0.56 efforts to
0.56 you know eradicate various diseases or
0.56 even like treat basic things like
0.56 diarrhea
0.56 in certain parts of the world that just
0.56 fail on logistics uh go look at that
0.56 um yeah and we did it was awesome we
0.56 spent a bunch of time in central america
0.56 and in africa getting another problem
0.56 and um
0.56 yeah i&;ll be honest i&;m a pretty
0.56 skeptical person so i spent time like i
0.56 was assuming okay we&;re gonna go meet
0.56 these customers and we&;re gonna learn a
0.56 thousand reasons why we can&;t actually
0.56 solve the problem
0.56 uh and anyway the opposite happened and
0.56 so the rest is history
0.56 and we went for it and just because not
0.56 everybody
0.56 will know ahead of time when you say the
0.56 rest of history you went for it what did
0.56 you go for
0.56 what did you build yeah sure so um
0.56 so the basically if you think about this
0.56 from a nuts and bolts like what our
0.56 customer cares about
0.56 we do fulfillment and delivery on demand
0.56 of medical supplies
0.56 right and by the way we started with
0.56 just doing blood right and blood is this
0.56 really
0.56 special medical supply right it has a
0.56 very short shelf life
0.56 it&;s rare everywhere in the world you
0.56 have less blood than you need
0.56 and because that&;s such a short shelf
0.56 life as few as seven days for certain
0.56 blood products
0.56 and there&;s so many blood types you need
0.56 to get it to the right person at the
0.56 right time
0.56 if you guesstimate where to send it most
0.56 of it will expire before it&;s used
0.56 um and so what we built the first thing
0.56 we built was a drone delivery system
0.56 that enabled us to centralize
0.56 uh blood storage in in basically one
0.56 location
0.56 and then fly very long range uh uh with
0.56 these uh basically small aircraft small
0.56 airplanes
0.56 picture a small fixed-wing drone that
0.56 flies out delivers the blood comes back
0.56 and gets the next order and does the
0.56 next order of blood
0.56 and on demand right so basically when
0.56 they know what type of blood they need
0.56 and how much blood they need they would
0.56 they
0.56 request that blood and that&;s what we
0.56 built initially
0.56 where many versions later and now we
0.56 deliver vaccines and other medical
0.56 supplies
0.56 at much longer range and bigger scale
0.56 but that&;s what we built now one of the
0.56 crazy things i saw
0.56 um when when watching the videos
0.56 on the technology side right is
0.56 well most people when they think about
0.56 drones definitely when i think about
0.56 drones
0.56 i i think of you know quadcopter type
0.56 drones that i might use to take some
0.56 videos when i&;m going for a hike and and
0.56 you know maybe other people do it for
0.56 skiing and so forth um the things you
0.56 you can just buy off the shelf in a
0.56 little box and it it makes videos of you
0.56 um but actually a zipline drone looks
0.56 very very different can you describe
0.56 what what it physically looks like
0.56 sure so our drones are they look like
0.56 small airplanes
0.56 um about 10 foot wingspan and
0.56 uh they deliver a box about the size of
0.56 a cake box
0.56 to kind of kind of make it physical um
0.56 what do they look like they have red
0.56 wings and a
0.56 white body and they&;re launched off of a
0.56 catapult
0.56 um and they land in this crazy
0.56 contraption
0.56 for for landing uh but all of that is
0.56 done
0.56 it&;s a fixed wing versus a quadcopter
0.56 for range which is what our customers
0.56 care about
0.56 uh if we launch with a catapult to make
0.56 the plane as simple as possible for cost
0.56 which again was
0.56 what our customers care about and
0.56 landing same thing right in order to
0.56 have something that can fly really far
0.56 really efficiently through all kinds of
0.56 storms
0.56 you basically don&;t want to put any
0.56 complexity on the drone itself
0.56 you don&;t have to so you don&;t want
0.56 landing gear you don&;t want any fancy
0.56 system on the drone for landing so we
0.56 have a contraption on the ground for
0.56 landing
0.56 um and uh yeah that&;s sort of the nuts
0.56 and bolts of it if you will
0.56 and let&;s say as an engineer i was
0.56 absolutely fascinated by by
0.56 the way the drone takes off and lands is
0.56 any chance you&;re able to maybe
0.56 pull up a video and talk us through
0.56 what&;s what&;s going on when a zipline
0.56 drone
0.56 takes off and lands because it&;s not
0.56 what i think most people think of when
0.56 they think of
0.56 their drones they have played with or
0.56 when they sit in an airplane that flies
0.56 them it&;s quite different
0.56 so at our distribution centers we have
0.56 inventory on site that&;s really
0.56 important because we deliver you know
0.56 in a matter of minutes from an order
0.56 coming in you saw a quick version of how
0.56 the order&;s packed
0.56 uh then put into the body of one of our
0.56 drones we call them zips
0.56 i might call them zips accidentally i
0.56 always forget to call them drones um
0.56 and uh that goes in the launcher you put
0.56 a wing on
0.56 uh on that on that drone then a battery
0.56 pack goes in there&;s a little pre-flight
0.56 process
0.56 and then it gets launched and this
0.56 launcher is a lot of fun this launcher
0.56 has a motor and a super capacitor bank
0.56 that takes that drone you saw there to
0.56 zero to 110 kilometers an hour and a
0.56 quarter second
0.56 wow and it&;s it&;s about as much power as
0.56 flooring and entry model tesla
0.56 to launch that that fast it just flies
0.56 right by
0.56 and it drops the package uh during that
0.56 flyby this is slow motion now just to be
0.56 clear
0.56 i just want slow-mo on you so um the the
0.56 package has a little paper parachute on
0.56 it
0.56 uh this is actually a very old this is a
0.56 this is a one our very first version so
0.56 this
0.56 delivery box is about the size of a shoe
0.56 box our new drones are about the size of
0.56 a cake box but it&;s basically the same
0.56 idea
0.56 there&;s actually a lot of technology
0.56 behind the scenes to do that well
0.56 uh to to basically measure and estimate
0.56 the wind speed and direction and
0.56 compensate so that package ends up where
0.56 the customer wants it not in the tree
0.56 um but yeah that&;s what a delivery looks
0.56 like then the
0.56 the the drone flies back of course
0.56 automatically and lands in this
0.56 and this system uh we kind of call this
0.56 an upside down
0.56 aircraft carrier and
0.56 uh the poles snap up and grab the plane
0.56 out of the air and they&;ll do it even in
0.56 crazy stormy conditions
0.56 now you can see that was slow-mo there
0.56 so you can see it i&;ll let this loop
0.56 again so you can see it uh so am i
0.56 seeing this correct canon that the plane
0.56 is just flying like normal
0.56 but then it gets snatched by
0.56 a cable that somehow
0.56 brings it to a stop almost
0.56 instantaneously exactly
0.56 so the plane just flies by it&;s actually
0.56 not trying to land it&;s trying to just
0.56 do a very safe flyby
0.56 and then there&;s that line between those
0.56 two poles and those pulls will snap
0.56 up right at the right moment uh to
0.56 basically throw the line between those
0.56 poles into the
0.56 into the tail hook of the plane uh you
0.56 can see it right amazing
0.56 now why not just have a regular landing
0.56 like when we
0.56 take a flight into a regular plane there
0.56 is you know the plane just lands and
0.56 why not set it up that way sure so it&;s
0.56 a really
0.56 really really a couple reasons one um
0.56 if you&;re looking at so one of the
0.56 things our customers care the most about
0.56 is reliability right they want us to be
0.56 able to do this
0.56 day and night storms all day long
0.56 and if you look at what determines the
0.56 reliability of an aircraft
0.56 you know even passenger aircraft one of
0.56 the hardest things to deal with is what
0.56 they call a hard landing so
0.56 is you know it&;s a some of us experience
0.56 them right your plane&;s coming in and
0.56 hits the runway a little harder than
0.56 you&;d like
0.56 right that the the mechanical stresses
0.56 that come from that
0.56 in your plane are very very hard to
0.56 predict and model
0.56 um it&;s very hard to build a small
0.56 aircraft especially even a large
0.56 aircraft
0.56 will have a very predictable life in the
0.56 face of hard landings
0.56 so this is designed from the ground up
0.56 to not have any hard
0.56 landings uh and we&;ve shown this out
0.56 past five nines of reliability
0.56 that zip will either do a flyby and come
0.56 around again which sometimes happens
0.56 uh or or or have this very repeatably
0.56 gentle load case right into the tail
0.56 we call the tail boom that thing at the
0.56 end of the plane with the hook on it
0.56 um that&;s one reason uh cost is another
0.56 keeping the planes very low cost is
0.56 important we&;re doing
0.56 you know hundreds we&;re up to hundreds
0.56 of flights per day per distribution
0.56 center so
0.56 we prefer to have some costs on the
0.56 ground and make every drone less
0.56 expensive
0.56 um so we can so the economics work out
0.56 for our customers
0.56 uh and uh yeah and then that storm piece
0.56 is really important as well
0.56 um you know just like it&;s it basically
0.56 makes our operators they don&;t have to
0.56 worry about what happens
0.56 right because a small aircraft like this
0.56 on a runway the way we land
0.56 in a storm literally can just get blown
0.56 off the runway
0.56 oh wow and having a system like this
0.56 like
0.56 eliminates those kinds of challenges
0.56 that&;s amazing then and it&;s also super
0.56 compact as a consequence it seems like
0.56 you need almost no space to to set this
0.56 up is that right
0.56 absolutely one of the reasons that that
0.56 recovery system we call it that
0.56 the landing system is so tall so we can
0.56 be right next to buildings
0.56 wherever wherever it makes sense uh
0.56 whatever is a good central location
0.56 right next to our supply that&;s where we
0.56 want to set up our distribution center
0.56 one of the uh things i noticed as i was
0.56 you know
0.56 researching the latest happening at
0.56 zipline
0.56 is also that you actually have
0.56 bono involved i mean youtube&;s bono
0.56 um what is bono doing at zipline oh he&;s
0.56 fantastic
0.56 so he&;s he&;s a member of our board board
0.56 of directors uh
0.56 fun fact it&;s the only corporate board
0.56 he&;s ever been on
0.56 uh and he uh you know
0.56 zipline is solving like again this bono
0.56 has committed
0.56 his career to solving some really
0.56 important public health challenges
0.56 uh and zipline is uniquely uh positioned
0.56 to solve
0.56 what we think it was like the unsexy
0.56 part of that which is like
0.56 you know logistics get the supplies
0.56 where they need it when they&;re needed
0.56 in a way that well basically has never
0.56 been solved at a really global scale
0.56 before
0.56 um and yeah so his role is fantastic he
0.56 helps us get connections with with the
0.56 right people to to get operating the
0.56 right places
0.56 um and obviously his credibility and his
0.56 um uh you know he he&;s just he&;s done so
0.56 many things in public health over the
0.56 last 20 years
0.56 he helps us see around corners that it
0.56 was really important for the company
0.56 i agree i mean it&;s awesome to have
0.56 somebody like that
0.56 on board but how do you even get them
0.56 more like do you just reach out
0.56 and you say hey i love your music and
0.56 and your other work also and
0.56 can we talk i mean he must be hard to
0.56 reach i mean i think it&;s like it&;s like
0.56 anything
0.56 in a startup right it&;s all about
0.56 relationships
0.56 right um it&;s all about relationships
0.56 and then over the years we&;ve built
0.56 relationships with people he respects
0.56 and done things that he you know
0.56 that the guy on his radar screen and and
0.56 that&;s how we connected
0.56 yeah wow so he he was kind of aware of
0.56 what you were doing
0.56 uh before you talk to him then yeah it&;s
0.56 it&;s it&;s like what we do this logistics
0.56 and public health is
0.56 there&;s so few people focus on
0.56 innovation there that if you care a lot
0.56 about public health you&;ve heard about
0.56 zipline for sure
0.56 how many deliveries have been done with
0.56 sips
0.56 we&;re well over 100 000 deliveries now
0.56 as i get it that&;s
0.56 that&;s counting a drone flight as as one
0.56 delivered because i also saw something
0.56 about
0.56 millions of vaccines
0.56 being delivered for covid so we&;ve done
0.56 millions of vaccines
0.56 pre-covered we will hopefully do a
0.56 million vaccines
0.56 of vaccine deliveries of covet here just
0.56 a matter of months we&;ve already done
0.56 tens of thousands of covered vaccine
0.56 deliveries just actually in the last
0.56 week
0.56 we just got our in ghana we just started
0.56 delivering cobia vaccines
0.56 uh which again like we got our training
0.56 wheels delivering well non-covered
0.56 vaccines we learned how you know
0.56 cold chain challenges the chain of
0.56 custody challenges we were really good
0.56 at that now
0.56 uh and then when they called us up and
0.56 said hey can you do covet now like we
0.56 were ready uh and now we&;re doing that
0.56 at scale which
0.56 i am personally really excited about
0.56 because you know he
0.56 in the states you know my parents
0.56 i mean it was just getting them
0.56 vaccinated was just insane
0.56 right the trying to figure out the
0.56 schedules and the
0.56 like but in ghana that&;s not what&;s
0.56 happening we&;re sending directly to
0.56 health clinics your doctor calls you up
0.56 and says hey i have a vaccine come on in
0.56 like that&;s how it should be right like
0.56 that&;s how it should be and someday
0.56 we&;ll be at that level here in the u.s
0.56 too
0.56 uh but we&;re just so excited to to
0.56 it just it&;s like so obvious right like
0.56 if you can deliver right to the doctor
0.56 especially the doctor who knows you
0.56 personally it&;s just it&;s just such a
0.56 better
0.56 health experience and just easier and
0.56 yeah just
0.56 it just works yeah it&;s gonna be amazing
0.56 i mean
0.56 there&;s some so much logistics to be i
0.56 mean in the us all you hear about for
0.56 covet is the logistics right now for the
0.56 vaccines and you&;re
0.56 effectively solving the core of that in
0.56 africa which is so exciting
0.56 yeah by the way one little fun geeky
0.56 piece of it but the big
0.56 challenge with logistics with trucks is
0.56 usually making big deliveries
0.56 which means you&;re where you deliver to
0.56 they have to have good refrigeration or
0.56 freezers
0.56 right but if you&;re doing on-demand
0.56 delivery so like in ghana we&;ll
0.56 typically deliver
0.56 enough for a couple days maybe even a
0.56 week of supply uh
0.56 the vaccine we&;re delivering there can
0.56 you be stored in just a normal fridge
0.56 for that time frame it&;s
0.56 no challenge right and so it basically
0.56 reduces by literally we&;re talking about
0.56 you know thousands of delivery sites so
0.56 it reduces by a thousand
0.56 the number of well freezers you have to
0.56 maintain and stuff like that which
0.56 it just makes things easier right it&;s
0.56 just way more flexible
0.56 that&;s amazing now when when you said
0.56 you started out
0.56 with solving a problem like finding the
0.56 problem you really want to solve
0.56 and you started in africa because that&;s
0.56 where you saw
0.56 um that you could really go go help by
0.56 by
0.56 setting up a zipline there but africa
0.56 is gigantic i mean it&;s it&;s not it&;s
0.56 not
0.56 just just a country right it&;s it&;s a
0.56 whole continent there&;s so many
0.56 countries
0.56 can you say a bit more about i mean you
0.56 can&;t just start in africa you go
0.56 it&;s got to be a more precise plan i
0.56 imagine yeah oh i mean the early days it
0.56 was you know
0.56 it was it was it was start-up precise
0.56 right so in the very early days
0.56 you know we started basically networking
0.56 and meeting with these different
0.56 governments
0.56 uh some in central america some in
0.56 africa uh we
0.56 learned very quickly that customers like
0.56 this require a lot of attention
0.56 and you know back then we were just you
0.56 know literally five people so we
0.56 couldn&;t you know give them all the
0.56 attention we needed so we we got started
0.56 to get smarter about okay
0.56 which of these countries we&;re talking
0.56 to should we start with we narrowed that
0.56 down to three
0.56 um and and then we and then because
0.56 we&;re a startup and we were trying to
0.56 get a government contract we we knew we
0.56 had to keep at least three right
0.56 you gotta you gotta manage your risks
0.56 which turned out to be really
0.56 uh uh really important because
0.56 up until maybe three months before we
0.56 launched with rwanda our first customer
0.56 uh it was we were sure it was gonna be a
0.56 different country
0.56 like up until close but that contracting
0.56 process
0.56 basically got stuck in the mud and had
0.56 to kind of reset uh
0.56 and rwanda said hey we want to be first
0.56 uh and and that contract was going
0.56 really well and we ended up launching in
0.56 rwanda
0.56 um but yeah it&;s again it comes back to
0.56 the relationships that networking and
0.56 yeah as you can imagine though
0.56 we were layering a lot of sort of
0.56 technical data onto it basically trying
0.56 to figure out you know which
0.56 which where we where were we going to be
0.56 most successful
0.56 for your first customer if it&;s not
0.56 successful as a startup you&;re in
0.56 trouble
0.56 we were looking at um we&;re also you
0.56 know looking at like integration
0.56 challenges
0.56 you know we make planes right we had to
0.56 rwanda&;s very high altitude country we
0.56 had to make sure we could fly at those
0.56 high altitudes
0.56 all kinds of interesting things like
0.56 that had to be sorted out to kind of
0.56 make sure we would have a good
0.56 successful
0.56 launch customer you said three countries
0.56 rhonda is one of them which are the
0.56 other two that
0.56 you started off with so we start out
0.56 with rwanda tanzania and costa rica
0.56 got it wow yeah now that&;s a bit spread
0.56 out i mean
0.56 did you physically go visit and spend
0.56 time at all three places or yes
0.56 absolutely
0.56 absolutely and our head of business
0.56 development spent a lot more time
0.56 our head of business development was
0.56 never home in early in the early years
0.56 uh yeah yeah now that was that was very
0.56 challenging
0.56 um yeah but it was challenging but it
0.56 also wasn&;t that challenging i think
0.56 a lot of people think of africa
0.56 especially as like a difficult place to
0.56 travel in and things like that
0.56 i think that was probably true 20 years
0.56 ago but it&;s it&;s easy now i mean you
0.56 can get to africa and you know
0.56 it&;s pretty much like getting to
0.56 anywhere in asia it&;s
0.56 nice and easy and um it&;s the world is
0.56 nice and small it&;s way smaller than you
0.56 would think
0.56 uh yeah but you know it&;s also startup
0.56 so like you know not sleeping for the
0.56 first few years is part of what you
0.56 should expect to do so
0.56 i i can attest to that um
0.56 now one thing that struck me is that
0.56 when you say you start for example in
0.56 rwanda it&;s not just
0.56 that you start in rhonda as i understand
0.56 it you actually
0.56 build a relationship with the ministry
0.56 of
0.56 health of rwanda so all the way at the
0.56 top can you see a bit more about that
0.56 any country we work with it&;s there&;s a
0.56 whole bunch of stakeholders that are
0.56 involved right so there&;s the obvious
0.56 ones like the regulators air space
0.56 regulators
0.56 uh and as well as the met the the the
0.56 health industry are basically the
0.56 equivalent of like the fda style
0.56 regulator uh
0.56 then there&;s the actual public health
0.56 systems that we&;re working with um
0.56 and then you layer on top of that the
0.56 political layers right because
0.56 nothing big happens in a country without
0.56 political basically support
0.56 right no no no one in a public health
0.56 system would take a huge risk without
0.56 political support
0.56 what was so cool about rwanda
0.56 specifically is when we went in to meet
0.56 with uh
0.56 the minister of health um you know she
0.56 like
0.56 well we&;re actually worried at first
0.56 because she&;s starting to text on her
0.56 phone we&;re like what&;s this all about
0.56 uh we lost her right but she in walked
0.56 two economists
0.56 uh and like from the very beginning we
0.56 worked with those economists to
0.56 literally put together a
0.56 basic health impact analysis as included
0.56 in economic impact analysis
0.56 um and like that was their step zero
0.56 based on really good data that they had
0.56 and they put that case together and then
0.56 once once uh the minister of health had
0.56 that then she went to the president and
0.56 said
0.56 hey we have to do this right and even
0.56 later in the process
0.56 when the department of defense had to
0.56 sign off on sort of american drones
0.56 coming
0.56 you know having her in the room saying
0.56 you got to figure this out and here&;s
0.56 why
0.56 right here&;s why this is so big for us
0.56 that was absolutely huge
0.56 and yeah and then from there that&;s that
0.56 then you start working through the
0.56 contracting process figuring out what
0.56 you&;re actually going to do and how
0.56 you&;re going to charge and
0.56 and all that kind of stuff and in most
0.56 of the countries you operate
0.56 uh like rwanda and ghana you know you&;re
0.56 talking about political approvals
0.56 involved right so you have to
0.56 you have to you have to pitch this and
0.56 you know there&;s a whole process
0.56 usually votings are involved it&;s it&;s
0.56 pretty wild to get through
0.56 through the other end yeah but there you
0.56 are and and
0.56 what i also noticed i don&;t know that&;s
0.56 part of how it all played out but
0.56 when i watched the videos it it seems
0.56 you get a lot of uh
0.56 you hire a lot of people locally and it
0.56 doesn&;t look like you bring in a large
0.56 team you
0.56 you set up shop and you start finding
0.56 people locally
0.56 is that right absolutely we we actually
0.56 have a place
0.56 so first of all all over zipline rwanda
0.56 and all of ziplin rwanda is run
0.56 and staffed and led all by locals
0.56 rwandans and ghanaians
0.56 um and in rwanda we have a place at our
0.56 first distribution center
0.56 we have a thing we call zipline academy
0.56 and so even our operators that are now
0.56 operating here in the u.s
0.56 we send them out there to get trained uh
0.56 and then they come back here to operate
0.56 took off so well that that&;s where the
0.56 training happens now even if you&;re yeah
0.56 i mean it&;s where you want to get
0.56 trained right because it&;s the highest
0.56 volume operation
0.56 uh it&;s also i mean it is just
0.56 disciplined and elite
0.56 it&;s really cool to see it&;s the kind of
0.56 thing where like i there was this great
0.56 there&;s this transition when i
0.56 very first days when i was out there
0.56 right i could help out and like pinch in
0.56 and things like that
0.56 but now when i go back and visit i&;m not
0.56 allowed to touch anything
0.56 i&;m not certified up to the standard
0.56 that they operate at
0.56 wow that&;s quite the story
0.56 locked out of your own uh operations
0.56 it&;s it&;s it&;s
0.56 yeah it&;s it&;s it&;s amazing it&;s awesome
0.56 now
0.56 you started in rwanda how many countries
0.56 are you operating in right now
0.56 yeah so today we operate in rwanda uh we
0.56 operate in ghana and we operate here
0.56 in the us so also in the us uh
0.56 i think i think that&;s less well-known
0.56 can you say a little bit about
0.56 what what&;s happening in the us yeah so
0.56 our operation in the u.s started back in
0.56 may and it was actually accelerated by
0.56 covid
0.56 um and uh yeah and to be clear in the us
0.56 we&;re like
0.56 we&;re like three years behind where we
0.56 are in africa in terms of scale
0.56 which is uh we&;ll catch up don&;t worry
0.56 but we&;re working very closely with the
0.56 us regulator to get to do that
0.56 uh we started doing uh we started doing
0.56 the very basics um
0.56 there&;s actually there&;s actually some
0.56 poetic uh
0.56 balance here so as i mentioned when we
0.56 started in rwanda we started with blood
0.56 right
0.56 because blood is rare it needs to be
0.56 delivered on demand right where you need
0.56 it
0.56 uh when we started north carolina in may
0.56 during covid
0.56 uh we started delivering ppe because it
0.56 was rare
0.56 it needed to be delivered exactly where
0.56 and when it was needed
0.56 um and uh so it&;s a we we have we have a
0.56 known story for how we start a new
0.56 country
0.56 uh just unfortunately in the us that
0.56 meant delivering masks
0.56 and things like that to healthcare
0.56 workers wow
0.56 that&;s a great story too though one
0.56 thing i also noticed for the us i don&;t
0.56 know
0.56 um if that&;s the same kind of things
0.56 you&;re delivering
0.56 was something about walmart mentioned in
0.56 one of the articles yeah so
0.56 we&;re launching with walmart later this
0.56 year um we have a handful of other
0.56 other other operations getting built in
0.56 the us right now which is uh which is
0.56 really exciting
0.56 uh walmart&;s one of them uh walmart will
0.56 be delivering what they call their
0.56 health and wellness products initially
0.56 so this is basically what you&;d expect
0.56 like a cvs pharmacy
0.56 um and uh yeah it&;s it&;s it&;s
0.56 it&;s a it also represents a really
0.56 exciting shift for zipline which we&;ve
0.56 been going through
0.56 in africa just starting as well which is
0.56 so in africa we started with hospitals
0.56 and then did
0.56 health clinics but now we&;re moving to
0.56 patient home delivery
0.56 and with walmart same thing so walmart
0.56 already does on-demand delivery you know
0.56 doordash style we call it you know
0.56 people in cars and that works for you
0.56 know
0.56 five kilometers or so maybe eight
0.56 kilometers uh but we can go
0.56 you know 80 kilometers uh and so
0.56 basically we&;re enabling that kind of
0.56 on-demand pharmacy delivery to be for
0.56 folks who are in just way more remote
0.56 locations
0.56 wow now as you&;ve done so many
0.56 deliveries at this point
0.56 um i&;m curious are there are there any
0.56 stories that send stand out to you of
0.56 you know
0.56 certain deliveries where it just felt i
0.56 mean
0.56 i imagine many but some must really
0.56 stand out that feel really life-changing
0.56 as you do this
0.56 blood deliveries vaccine deliveries
0.56 anything
0.56 you could tell us about your personal
0.56 favorites sure i have one geeky story to
0.56 tell and one really really kind of
0.56 emotional story to tell
0.56 the geeky story was fun so our test
0.56 sites are here in california
0.56 and we&;ve you know we&;ve been flying all
0.56 day every day at these test sites long
0.56 before we started in africa to work
0.56 you know work out all the issues system
0.56 uh but it&;s really important that our
0.56 test sites that the drones fly you know
0.56 where you can see them so they kind of
0.56 do circles and stay near stay where you
0.56 are
0.56 right and so if a drone at a test site
0.56 ever were to like fly over the horizon
0.56 like that&;s a bad thing
0.56 like that something went wrong i still
0.56 remember the very first time
0.56 in rwanda our very first uh
0.56 deliveries to our very first hospital
0.56 the first time i saw a drone just like
0.56 take off
0.56 and just disappear like my heart just
0.56 skipped a beat and like i was like
0.56 something is wrong
0.56 this is not right uh and then of course
0.56 you know it was a pretty short delivery
0.56 but like 10 minutes later it comes back
0.56 doesn&;t have the blood anymore and it
0.56 was like wow that just happened holy
0.56 smokes
0.56 yeah and uh so that was that was that
0.56 was one of my more formative like
0.56 and traumatizing geeky level experiences
0.56 um
0.56 you have to really trust it beyond what
0.56 you can see exactly
0.56 just we just went it just went um
0.56 and uh one other fun fun anecdote about
0.56 that flight
0.56 this was a very beginning and we didn&;t
0.56 want to fly in rain because we want to
0.56 kind of ease
0.56 into it so we looked at all the weather
0.56 data sources and said it won&;t rain
0.56 so but that that plane come but came
0.56 back soaking wet
0.56 so it&;s like oh there must be some rain
0.56 over the horizon we didn&;t know about
0.56 anyway um wow that was a fun moment i
0.56 was like
0.56 uh but yeah i think one of the more
0.56 personal ones was meeting a woman
0.56 uh this was lady that same that first
0.56 year this woman named alice and she
0.56 we actually ended up delivering to her
0.56 twice believe it or not so she
0.56 we delivered to her while she was uh
0.56 having postpartum hemorrhage
0.56 uh while delivering blood uh sorry
0.56 delivering a baby she
0.56 went to postpartum hemorrhage and we
0.56 delivered blood for that case uh
0.56 and then six months later she got
0.56 malaria and one of the
0.56 one of the uh sort of things that
0.56 happened to malaria can&;t happen is you
0.56 develop anemia which also requires blood
0.56 and so she like we delivered to the same
0.56 woman twice
0.56 and so she had to come check us she had
0.56 to see what this was about
0.56 and so she actually came by and visited
0.56 and that was that was a pretty surreal
0.56 moment to meet somebody
0.56 who like you know who heard it like in
0.56 her words like she&;s like yeah i heard
0.56 about this crazy thing called
0.56 you know i heard my was what&;s a drone
0.56 and she really wanted to see what it was
0.56 so she literally just found out where we
0.56 were and
0.56 came by and knocked on the gate and then
0.56 checked it out that was that was that
0.56 was pretty that was pretty special
0.56 yeah that must be amazing and when you
0.56 say one step removed you&;re really
0.56 50 to 100 kilometers removed
0.56 right so this this woman came a long way
0.56 probably to
0.56 to meet you yep she did yeah wow
0.56 now these are
0.56 great stories i&;m curious about stories
0.56 that or maybe
0.56 things where you really struggled and
0.56 things didn&;t really go according to
0.56 plan
0.56 um any that any that are your favorites
0.56 there
0.56 yeah oh struggles so so i met my wife
0.56 uh skiing uh she&;s costa rican
0.56 uh she&;s a doctor and an epidemiologist
0.56 anyway skiing so my sister
0.56 my sister we both know my sister through
0.56 berkeley and
0.56 uh and my wife has a epidemiology phd
0.56 from berkeley anyway so she
0.56 on this ski weekend we go skiing and my
0.56 wife never skied before so anyway
0.56 i basically spent the weekend teaching
0.56 her to ski and one of the things when
0.56 you&;re learning to ski you always tell
0.56 people is
0.56 you look where you want to go don&;t look
0.56 where you don&;t want to go right stare
0.56 at the tree
0.56 you&;re way more likely hit the tree if
0.56 you stare at the place that doesn&;t have
0.56 the tree and you&;ll go there
0.56 and so she made fun of me because after
0.56 willow garage i had this and i was
0.56 looking for my next thing i wanted to do
0.56 i had this list of things that should
0.56 not be in a startup
0.56 and they included things i had
0.56 everything on the list you don&;t want to
0.56 start up i didn&;t want regular
0.56 regulation i didn&;t want government
0.56 customers i didn&;t want hardware i
0.56 didn&;t want safety critical
0.56 i didn&;t want safety critical hardware
0.56 and satrio software and so i had this
0.56 list going
0.56 and when i was telling what after the
0.56 first trips looking at zipline uh
0.56 looking at what became zipline i was
0.56 like i&;m really thinking about this and
0.56 she started making fun of me she&;s like
0.56 you were staring at that list too much
0.56 because everything on the list
0.56 of don&;t put in startup is in this
0.56 potential thing
0.56 uh and i think it&;s uh anyway so
0.56 it always makes me laugh because a lot
0.56 of the really challenging things about
0.56 zipline have been
0.56 in in that world right selling the
0.56 government&;s just really hard to do
0.56 the regulation stuff just really because
0.56 like
0.56 we work in a regulated environment
0.56 without regulations which is like the
0.56 worst right
0.56 there aren&;t drone regulations that work
0.56 for ziplines so we&;re always every
0.56 country we go to we&;re working
0.56 to basically create the regulations that
0.56 enable zipline and that&;s
0.56 um that&;s been an adventure and then of
0.56 course all the safety critical aspects
0.56 of this
0.56 has been really wild all of this sounds
0.56 uh
0.56 like it has many many challenge many
0.56 layers of challenges
0.56 maybe being an engineer in in training
0.56 um i&;m really curious about the
0.56 technical side
0.56 um and some some challenges that you
0.56 might have
0.56 run into there because i mean
0.56 you don&;t just build a drone like that i
0.56 mean you you gotta somehow
0.56 decide what drone you even build how
0.56 you&;re going to build it
0.56 um say a bit more about that how does
0.56 this
0.56 become what it is now this is fun
0.56 because when we started zip line i
0.56 actually thought i was going to buy
0.56 drones
0.56 and then maybe modify them for package
0.56 delivery and that&;s how we get going
0.56 and so the first kind of like uh let&;s
0.56 call it rough lesson was
0.56 talking to all the drone manufacturers
0.56 who could who could potentially
0.56 build drones to do what we wanted to do
0.56 and basically learning that
0.56 none of them worked for what we wanted
0.56 to do i still remember the most
0.56 the closest we got to that was a drone
0.56 manufacturer who wanted to charge us
0.56 200 000 a drone with a 200 flight
0.56 warranty
0.56 if we didn&;t fly in the rain
0.56 and it was like it was it was just like
0.56 so impossibly
0.56 not you know useless for what we wanted
0.56 to try to do uh that was like oh wow we
0.56 actually have to start in order to our
0.56 first customer we have to build a drone
0.56 ourself
0.56 um i remember that was one of those like
0.56 wake up call moments of like okay
0.56 there&;s more work here than we thought
0.56 can you contrast the 200 000
0.56 with the rough cost of building a drone
0.56 yourself
0.56 sure sure like the the i mean our drones
0.56 are now it&;s a little hard to like
0.56 describe the drone itself but think of
0.56 our drones as like low
0.56 very low tens of thousands right so and
0.56 they also last a very long time and they
0.56 fly through
0.56 rain and all the time and they&;re very
0.56 easy to maintain and they have a lot of
0.56 other things that we could not get off
0.56 the market if you will
0.56 um yeah um one really interesting
0.56 technical challenge like
0.56 i think the oftentimes
0.56 like we like if
0.56 if you&;ve been trained by anybody right
0.56 in engineering they teach you to think
0.56 about risk right so you&;re kind of
0.56 thinking you&;re like
0.56 making your list so like what are all
0.56 the things they&;re going to get me
0.56 and one of those but what really burns
0.56 you is the things you don&;t
0.56 really think are going to be big risks
0.56 right and so
0.56 one of the advantages we have is we have
0.56 a plane and a plane flies really
0.56 efficiently and that&;s you know
0.56 if you stand 100 feet back from our
0.56 plane it looks a lot like all the other
0.56 planes right in terms of
0.56 got a wing and a body and a tail and you
0.56 know you&;re good to go
0.56 um and so you know one of the things we
0.56 wanted to do was fly through weather and
0.56 so we went to the experts in the
0.56 industry
0.56 uh places like nasa and said hey like
0.56 how do we make sure this thing can fly
0.56 through weather
0.56 and they gave us their models they&;ve
0.56 been developed over many years of how
0.56 to simulate and how to basically set
0.56 your requirements for
0.56 flying through weather um and we did
0.56 that okay
0.56 followed all that follow their their
0.56 recommended design practices their
0.56 requirements and so on uh
0.56 we went to rwanda we started flying and
0.56 it didn&;t work right
0.56 we have a parachute landing system on
0.56 our drones uh we were parachute landing
0.56 way too often way more often than that
0.56 that analysis and modeling suggested we
0.56 could
0.56 and then we turned around and really
0.56 started thinking about the problem went
0.56 back to those experts
0.56 and we&;re like okay how did you build
0.56 these models and so on
0.56 and it turns out we were actually doing
0.56 something that basically nobody does
0.56 nobody flies near the ground in storms
0.56 um it&;s just not done and now in our
0.56 head even i was like yeah but like you
0.56 know
0.56 uh search and rescue helicopters they
0.56 must do it so get on the phone literally
0.56 talk to the team
0.56 in the california coast guard talk to
0.56 the team on zermatt right the biggest
0.56 mountain is the swiss alps
0.56 like their rescue team and both of them
0.56 said the same thing oh
0.56 we don&;t fly in weather anymore that was
0.56 20 years ago too many people died we
0.56 stopped we stopped we like
0.56 we wait till the weather clears right
0.56 and so we&;re sitting here going like
0.56 okay we&;re flying in this
0.56 really interesting uh domain that nobody
0.56 flies in
0.56 and anyway so we had to go learn about
0.56 this domain develop our own models
0.56 and actually pretty much completely
0.56 redesign our drone
0.56 to actually be able to survive in the
0.56 crazy sort of
0.56 updrafts and downdrafts and sort of
0.56 stormy conditions that happen and this
0.56 we call them energy concentrators as you
0.56 have winds flying over hills and
0.56 mountains
0.56 these crazy basically energy sources uh
0.56 we had to redesign the plane the control
0.56 system pretty much everything to
0.56 actually do this reliably like we did
0.56 today
0.56 and we anyway so those are two steps
0.56 first i didn&;t even think i&;d make the
0.56 drone
0.56 and then i thought well at least nasa
0.56 knows how to fly in this in this kind of
0.56 environment so i
0.56 listened to them and both of those
0.56 things turned out to be
0.56 learning experiences we are dropping new
0.56 interviews
0.56 every week so subscribe to the robot
0.56 brains
0.56 on whichever platform you listen to your
0.56 podcasts
0.56 ai of course is powering many many
0.56 systems these days and i&;m curious
0.56 uh when today as it you know flies off
0.56 and comes back
0.56 what kind of ais involved in that
0.56 process today
0.56 yeah so today in operations the
0.56 the ai in operations is mostly around
0.56 weather um
0.56 now there&;s no coincidence that our
0.56 biggest reliability challenge today in
0.56 operations is also weather
0.56 and so that&;s what we put that&;s where
0.56 we put our big guns in terms of what we
0.56 really
0.56 solve problems around in our development
0.56 process at our test sites and things
0.56 it&;s around aircraft detection and then
0.56 soon it&;s going to be
0.56 around things like ground obstacle
0.56 detection and delivery site selection
0.56 and getting that package right where you
0.56 want it on your doorstep
0.56 one of the things that
0.56 happened kind of in during i mean
0.56 zipline already existed but kind of
0.56 started happening last five years really
0.56 is
0.56 ai has really broken through and got a
0.56 lot of new capabilities that weren&;t
0.56 possible before and image recognition
0.56 speech recognition and so forth right
0.56 and all powered under the hood with with
0.56 deep learning
0.56 and i&;m very curious
0.56 if you know deep learning plays a role
0.56 at zipline now or
0.56 what what what do you see kind of the
0.56 future of deep learning in the context
0.56 of zipline
0.56 absolutely absolutely and this this
0.56 comes back so so
0.56 ai in general&;s role at zipline is
0.56 growing so in the past it&;s only been
0.56 offline and that&;s been very intentional
0.56 because
0.56 i was dreading having to explain you
0.56 know ai basically to regulators
0.56 uh in the early days because it was
0.56 already hard enough to explain the more
0.56 way more
0.56 sort of deterministic stuff we were
0.56 doing um
0.56 but this the ai has a big role in the
0.56 future of flying for us
0.56 uh and it&;s really connected to that
0.56 that march we&;re going down
0.56 starting with hospitals health clinics
0.56 and outpatient home delivery
0.56 right because with hospitals from one
0.56 distribution center you know
0.56 maybe 10 to 20 hospitals we can reach
0.56 depends where we are in the world
0.56 health clinics you know now we&;re
0.56 talking about thousand health clinics
0.56 but now they&;re in patient home delivery
0.56 we&;re talking about hundreds of
0.56 thousands of delivery sites
0.56 from one distribution center and so a
0.56 lot of the things we did
0.56 uh that make economic sense and scale
0.56 for those those
0.56 those uh for health clinics and
0.56 hospitals don&;t make sense anymore and
0.56 really
0.56 aren&;t going to create the customer
0.56 experience we want for this next level
0.56 of operation
0.56 so uh and it really it comes down to
0.56 things like
0.56 well detecting and avoiding other
0.56 aircraft uh detecting and avoiding uh
0.56 ground obstacles you know cell towers
0.56 and power lines and trees and things
0.56 like that
0.56 uh and then uh the last place also
0.56 related to perception that we&;re working
0.56 hard on is really around the delivery
0.56 experience
0.56 and being able to get dealing dealing
0.56 with that you know
0.56 the people want their deliveries in
0.56 their backyards uh from drones
0.56 and so being able to actually sort of
0.56 navigate if you will and find a good
0.56 delivery place in their backyard that&;s
0.56 going to work and be safe
0.56 this is a big part of where we&;re going
0.56 with it um
0.56 and so uh i can i&;ll talk a little more
0.56 about that in a second the other place
0.56 that we&;ve
0.56 we&;re using deep learning machine
0.56 learning is actually weather forecasting
0.56 we have our own weather forecasting we
0.56 call it now casting because we care
0.56 about like
0.56 basically what&;s right in front of the
0.56 drones and we feed data in from
0.56 satellites and
0.56 from our drones to guests uh there&;s
0.56 basically a very particular type of
0.56 weather event
0.56 we can&;t fly through uh it&;s called a
0.56 micro burst
0.56 and so we really wanna oh what&;s
0.56 microburst so micro versus what happens
0.56 right at the beginning of a thunderstorm
0.56 when that thunderstorm head forms
0.56 um you end up with this huge updraft and
0.56 and then outside of a downdraft and i
0.56 say huge updraft i&;m talking like
0.56 we we have recorded data at like 40 50
0.56 meters per second
0.56 of the the vertical component of the
0.56 wind on the middle and on the outside
0.56 it&;s
0.56 you know coming down um and uh well
0.56 let&;s just say that&;s more than our
0.56 drones can handle then that
0.56 if we often not every time often if we
0.56 end up in those
0.56 we end up parachuting the drone um
0.56 and so being able to forecast those has
0.56 been something that that we actually end
0.56 up rolling ourselves because this is
0.56 another one of those things that
0.56 we care a lot about uh it&;s a very one
0.56 of these near ground effect
0.56 uh things uh and it turns out the the
0.56 rest of the weather industry isn&;t
0.56 working on it so
0.56 uh uh the the brilliance of ai today one
0.56 really good engineer
0.56 uh has has made a huge impact on this
0.56 for zipline already
0.56 i&;m kind of curious because when we
0.56 think about ai if we talk about this
0.56 specific one for a moment
0.56 sure it&;s like the typical model is that
0.56 you you collect data
0.56 you have observations and then the ai
0.56 often a neural network has to turn those
0.56 observations
0.56 coming from satellites radar and so
0.56 forth
0.56 turn that into a prediction yep but the
0.56 way it&;s
0.56 it&;s trained is by also having access to
0.56 ground
0.56 truth predictions that you have as
0.56 examples
0.56 and so it seems almost like a chicken
0.56 and egg problem because you gotta
0.56 predict those you gotta find those
0.56 microbursts
0.56 so you can collect your data so you can
0.56 say a little bit about that sure
0.56 so there&;s basically three ways that we
0.56 there&;s three sources of ground truth
0.56 for this problem that we use
0.56 one source of ground truth truths are
0.56 our historical logs from every flight
0.56 we&;ve ever done
0.56 uh places where we&;ve flown into
0.56 microburst or flow near them
0.56 uh because if you&;re right in the center
0.56 it&;s it&;s one thing but if you&;re near
0.56 them you can still
0.56 see the sense them that&;s one source uh
0.56 the other source
0.56 is is historical satellite imagery right
0.56 so you know
0.56 you have the satellite you have the
0.56 satellite imagery and you can actually
0.56 these microbursts have very unique uh
0.56 they look very unique visually in these
0.56 in this imagery and so you can see when
0.56 it happens and then you try to
0.56 train a model on basically what came
0.56 before uh you try to
0.56 predict based on what came before when
0.56 it&;s about to happen
0.56 uh and then there&;s another thing we&;ve
0.56 been using which is basically
0.56 um so computational fluid dynamics right
0.56 lets you sort of build like a sort of a
0.56 3d model
0.56 uh of how the wind moves over surfaces
0.56 and so this is
0.56 historically done over like your arrow
0.56 your airplane wing
0.56 uh but we&;re also doing this over
0.56 terrain so we actually have 3d models of
0.56 countries
0.56 and that lets us build very detailed
0.56 models that helps us
0.56 basically fill in the gaps between those
0.56 other two sources i mentioned right
0.56 so if we&;re flying if we have data from
0.56 a parachute landing where we flew into
0.56 one of these
0.56 uh we will go back we take the data from
0.56 our plane flying in
0.56 we take the weather data from that you
0.56 know from that moment of time that
0.56 whatever we can get our hands on
0.56 we run that through these sort of macro
0.56 uh models of
0.56 computational fluid dynamics models to
0.56 predict what was actually happening
0.56 basically ahead of where we parachute
0.56 landed right the thing we didn&;t quite
0.56 get to see right
0.56 uh and then that helps us fill in some
0.56 of the gaps so this is fascinating
0.56 because
0.56 i mean this is this is definitely not a
0.56 mainstream ai problem that
0.56 you would find in the research
0.56 literature this but it is
0.56 exactly the same kind of methodology as
0.56 you would use for the kind of more
0.56 commonplace
0.56 vision and speech problems it seems
0.56 exactly exactly
0.56 and it&;s also one of those fun sort of
0.56 necessity problems right
0.56 you know when we launched in rwanda and
0.56 ghana this was not our biggest challenge
0.56 for reliability
0.56 right but over time we&;ve gotten to the
0.56 point where it is uh and now we&;ve been
0.56 tackling it
0.56 um and uh and we like you said we have
0.56 because we have this
0.56 historical data it helps sort of jump
0.56 start that chicken and egg cycle
0.56 that we can use that to really uh well
0.56 sort of mind that for any information
0.56 that we can use to train these models
0.56 on that note of data is are there other
0.56 things that
0.56 are on your mind for the future where
0.56 it&;s okay thanks to flying these drones
0.56 we have
0.56 very interesting data that can allow us
0.56 to build
0.56 more advanced systems that can do even a
0.56 better job or something new
0.56 yeah yeah i know absolutely and i the so
0.56 you
0.56 well already with weather generally it&;s
0.56 something we&;ve been thinking a lot
0.56 about because
0.56 our drones sort of represent these
0.56 flying uh weather stations
0.56 that that are very good uh
0.56 we care a lot about them being precise
0.56 so our planes fly well uh
0.56 and and and they and they also capture
0.56 data
0.56 that&;s very clean so if you think about
0.56 the weather challenge in general weather
0.56 forecasting in general
0.56 um you know it&;s it&;s all about where
0.56 your data comes from and if you&;re
0.56 collecting data on the ground
0.56 that&;s like the that&;s like the worst
0.56 data ever because it&;s a wind especially
0.56 if just it&;s all affected by
0.56 what&;s around you right so a weather
0.56 station next to a tree
0.56 is not as good as a plane nice in the
0.56 nice sort of free stream of the air
0.56 um and we&;d love to sort of use that to
0.56 help with weather forecasting in general
0.56 is that what you&;re asking are you
0.56 you&;re going more for like internal
0.56 yeah i was thinking just generally
0.56 kind of things that because you have the
0.56 data new ideas that have come to your
0.56 mind as you know
0.56 opportunities you you could leverage the
0.56 data for yeah
0.56 yeah so i&;ll be honest most of the most
0.56 my efforts of how to use our existing
0.56 data are really about how to
0.56 get past the next hurdle for the company
0.56 exactly exactly uh so some of the things
0.56 we&;ve looked at which are
0.56 starting to bear fruit are essentially
0.56 using
0.56 so we have logs from every flight we&;ve
0.56 ever done ever and we and we&;ve
0.56 invested quite a bit in the data systems
0.56 that make uh using those logs really
0.56 um easy uh and of course like
0.56 a lot of what those logs get used for is
0.56 like intentional right so let&;s say we
0.56 have a problem and we use those logs to
0.56 study
0.56 data to help us root cause a problem
0.56 another example of intentional of course
0.56 is
0.56 a design problem we want to understand
0.56 what are the loads that
0.56 a wing sees right and so we can look
0.56 through all of our flight data to get a
0.56 nice really good statistical picture of
0.56 what loads of wings sees but of course
0.56 there&;s there&;s definitely information
0.56 in those logs that we don&;t
0.56 know to go look for but there&;s a
0.56 fantastic amount of data and so we&;ve
0.56 actually
0.56 we&;ve been starting to look at basically
0.56 sort of anomaly detectors to help us
0.56 flag
0.56 like what is weird right um
0.56 which can now be a predictor sort of a
0.56 near-miss if you will signal
0.56 uh that hey we&;re going to have a
0.56 problem coming and we treat this as a
0.56 near-miss now we dig into it
0.56 we may be able to basically head off a
0.56 reliability challenge that we don&;t
0.56 understand
0.56 yet that&;s sort of order magnitude
0.56 beyond the reliability we are
0.56 today if that makes sense right it seems
0.56 like you could do preventative
0.56 maintenance
0.56 in really nice ways at the scale that
0.56 you&;re starting to operate exactly
0.56 exactly we&;re really excited about that
0.56 too yeah going back to
0.56 to what you mentioned earlier the other
0.56 main application
0.56 that you&;re already seeing now for for
0.56 deep learning and ai in general uh on
0.56 the image recognition side and you said
0.56 something about
0.56 landing or destinations can you say a
0.56 bit about that
0.56 sure absolutely so there&;s a number of
0.56 things we&;re working on so
0.56 um let me sort of paint a little bit of
0.56 a picture so the first thing we&;ve been
0.56 working on
0.56 using ai and perception is around
0.56 aircraft detection
0.56 uh and we actually we have our first
0.56 submission into the fa uh in november
0.56 for that
0.56 which by the way fun fact as far as we
0.56 know as far as the fa people
0.56 we&;re working with know this is the
0.56 first time the fa has ever reviewed
0.56 uh basically a trained model or an ai
0.56 based ai
0.56 functionality for use in the national
0.56 airspace
0.56 uh so this this will be a this is a this
0.56 is a this is i&;m saying that for two
0.56 reasons because it&;s not approved yet to
0.56 be clear
0.56 but it is approved uh it&;s a big
0.56 milestone for the fa as well because
0.56 they&;re they&;re basically
0.56 uh sort of learning how to think about
0.56 this kind of performance-based uh
0.56 way of looking at these problems um
0.56 which is uh which it&;s a big lift
0.56 uh it&;s it&;s i mean to be clear it&;s a
0.56 big lift for them and i know that
0.56 because it&;s a big lift for us
0.56 even for us to get the confidence of
0.56 like hey
0.56 we are confident in the safety
0.56 performance of the system in order to
0.56 then submit it to them
0.56 just getting that done internally was
0.56 was uh uh it required people better at
0.56 math than i am
0.56 well you got some really great people
0.56 apparently um now
0.56 when i think about this i mean you build
0.56 a vision system to recognize other
0.56 aircraft
0.56 um what does that look like i mean how
0.56 do you build a system to detect other
0.56 aircraft
0.56 yeah how do you build a system detect
0.56 other aircraft so uh the first thing
0.56 i&;ll mention is the thing we submitted
0.56 to the faa
0.56 is not what you think it is it&;s not
0.56 vision and i can&;t tell you what it is
0.56 quite yet but it&;s
0.56 very cool and it works incredibly well
0.56 so i&;m very excited about it but that&;s
0.56 that&;s enough teasing that idea
0.56 we actually do have we actually i do
0.56 have vision capabilities in development
0.56 right now
0.56 and those are focused around what we
0.56 call the delivery experience so this is
0.56 finding the spot you know at your house
0.56 or your backyard where we can deliver
0.56 right
0.56 so we we want to see what we want to
0.56 find the trees so we don&;t put it
0.56 delivering to the trees is a bad
0.56 customer experience uh we want to figure
0.56 out like
0.56 where&;s a good you know we don&;t want to
0.56 deliver in your pool uh there&;s a lot of
0.56 things like that that
0.56 they want to really understand well um
0.56 and so a lot of that is uh
0.56 you know it&;s it&;s gathering data from
0.56 just operations using our sort of
0.56 existing methods of of getting operating
0.56 uh it&;s it&;s it&;s also gathering data
0.56 sort of by hook and by cook right like
0.56 we put we&;ll put a camera payload on a
0.56 fishing pole or basically a sort of
0.56 extendo pole and send it home with
0.56 people
0.56 to kind of build out data sets and and
0.56 things like that to help make sure we&;re
0.56 getting
0.56 uh um diversity in our data sets uh
0.56 and yeah and then it&;s just a lot of
0.56 really understanding the problem
0.56 uh and then being really clever about
0.56 how do we sort of
0.56 reduce the dimensionality of the problem
0.56 how do we make this problem tractable
0.56 right how do we how do we stop thinking
0.56 about it is like a brute force like
0.56 we&;re just going to solve the whole
0.56 thing to like how do we
0.56 how do we make this something we can
0.56 really get our head around uh train a
0.56 network to solve in a way that
0.56 that actually works for the customer and
0.56 and uh and makes sense too that&;s one of
0.56 the last pieces i think really important
0.56 when you&;re dealing with people you know
0.56 if it doesn&;t make sense uh to anybody
0.56 but the
0.56 ai team like it&;s not it&;s hard people
0.56 have to look at they have to be like i
0.56 want that or i
0.56 understand that i gotta get what&;s
0.56 happening and that has to come together
0.56 in a reasonable way
0.56 now one thing i&;m very curious about a
0.56 lot of the challenges of putting ai to
0.56 work
0.56 in in the real world relate to kind of
0.56 the long tail always new things happen
0.56 is that is that the case are you trying
0.56 to address that or
0.56 you know does a customer have to satisfy
0.56 specific requirements ahead of time that
0.56 you specify before
0.56 you know it it&;s it&;s a place you can
0.56 deliver it&;s very much about the long
0.56 tail and this is part of the reason we
0.56 pushed really hard to find a use case
0.56 and a customer
0.56 that we could get operating without
0.56 requiring ai first
0.56 because what that means is you can kind
0.56 of think of our ai-based capabilities
0.56 as essentially adding one of two things
0.56 improving our safety
0.56 case right or basically improving our
0.56 efficiency uh or in some cases enabling
0.56 experiences we couldn&;t have done
0.56 otherwise
0.56 um but when you think about a safety
0.56 case for example like
0.56 so safety cases sort of this in
0.56 aerospace is this end to end you know
0.56 safe sort of statistical argument about
0.56 how safe are you
0.56 um and uh and and there&;s a lot of ways
0.56 to solve the problems in that chain
0.56 right and some you know some of them
0.56 aren&;t very efficient or
0.56 aren&;t very elegant but you do them
0.56 anyway because you want you got to be
0.56 safe at all times
0.56 but because we have that baseline today
0.56 we&;re able to layer in these
0.56 capabilities
0.56 in ways that enable us to scale while
0.56 maintaining safety for example which is
0.56 a lot of what we&;re doing
0.56 um but without requiring perfection at
0.56 the beginning right as long as now it&;s
0.56 important that we know how imperfect it
0.56 is
0.56 because otherwise you can&;t stand behind
0.56 the safety case but if you know how
0.56 imperfect it is
0.56 you can you can you can launch it you
0.56 can operate on it you can
0.56 learn about it um and then you can
0.56 improve it over time
0.56 which improves the safety case improves
0.56 your ability to scale uh and that&;s sort
0.56 of you know we&;re kind of in that dream
0.56 situation
0.56 where you know a lot of people describe
0.56 it similar like tesla like tesla makes
0.56 plenty of money selling cars
0.56 and they can use that to then sort of
0.56 basically improve their cars if you will
0.56 and make them more and more valuable
0.56 by adding a base capabilities which is a
0.56 great way to enter the market versus
0.56 having to have
0.56 an ai capability that just like nails
0.56 your product on the first go
0.56 which is hard to do you talk about it so
0.56 concrete about
0.56 you know deliveries and so forth and i&;m
0.56 kind of curious
0.56 um could i personally expect
0.56 a drone delivery from zipline as just as
0.56 a consumer
0.56 just buying something online is that
0.56 something that&;s on the near horizon
0.56 here in let&;s say california i mean it
0.56 all depends on your definition of your
0.56 horizon but we&;ll
0.56 get we&;ll get it done we will get that
0.56 done for sure
0.56 it will definitely initially be health
0.56 related um and it might be as simple as
0.56 like you have a telehealth visit
0.56 and like while the telehealth is
0.56 happening on your phone like the doctor
0.56 sends you something directly to your
0.56 home
0.56 like that&;s what it&;ll be initially uh
0.56 to homes here in the u.s
0.56 um and you know some things like walmart
0.56 if you&;re in more remote areas
0.56 uh and then we&;re gonna go from there
0.56 and keep like th there&;s this awesome
0.56 thing happening in the u.s around
0.56 healthcare at home
0.56 accelerated by covid and the reason i
0.56 can say it&;s awesome is actually africa
0.56 is way ahead of us europe&;s way ahead of
0.56 us
0.56 and i&;ve got to experience myself a few
0.56 times in my life
0.56 and just how cool it is to have really
0.56 good health care at home whenever you
0.56 can
0.56 uh and here in the states it&;s going to
0.56 happen too and we want to be a
0.56 really big part of accelerating it
0.56 because it&;s just it&;s just so much
0.56 better
0.56 it&;s so much better yeah well i look
0.56 forward to that um
0.56 i i definitely know about the
0.56 discrepancies between
0.56 healthcare and us and europe so a lot of
0.56 gaps there
0.56 now when a lot of people think about
0.56 drones
0.56 i think the most natural thing they
0.56 still think about today
0.56 is dji drones because those are the the
0.56 toys that
0.56 you buy and you have fun with and it&;s a
0.56 consumer product right
0.56 yeah and so i&;m kind of curious how do
0.56 you see the the broader
0.56 kind of landscape um you know
0.56 what&;s happening with with drones beyond
0.56 what&;s happening
0.56 at zipline and what are some exciting
0.56 things that are happening i mean the
0.56 most exciting thing that&;s happening
0.56 with drones at zipline and elsewhere is
0.56 really around autonomy
0.56 right it&;s it&;s really around these
0.56 systems being able to take care of a lot
0.56 of the basics of like don&;t hit things
0.56 uh whether it&;s whether it&;s a drone
0.56 like what uh you know skydio makes which
0.56 is sort of a
0.56 dji that basically has a lot of autonomy
0.56 in it so you could basically it&;s hard
0.56 hard to crash it
0.56 almost impossible it just flies itself
0.56 um and those kind of capabilities for
0.56 uh for all manner of things the the the
0.56 skydio has this amazing product out for
0.56 insurance adjusters where basically
0.56 you walk up to a house you push a button
0.56 and it gives you like a you know
0.56 this like you know super high resolution
0.56 and imagery of the roof for that you
0.56 know in a matter of seconds you don&;t
0.56 fly it it just flies itself
0.56 and so there&;s really cool stuff
0.56 happening from companies like that
0.56 around for industries like that so take
0.56 here in california right
0.56 pg e is like you know has this huge
0.56 problem with uh
0.56 starting fires right and
0.56 you know last year they&;re like look we
0.56 have i forgot the number was you know
0.56 i think it&;s millions of of kilometers
0.56 of power lines or
0.56 maybe hundreds of thousands just a
0.56 fantastically large number way more than
0.56 you can expect with helicopters which is
0.56 what they used to do
0.56 and they came to the drone industry and
0.56 said hey can we do this at scale
0.56 right and we&;re at the cusp of saying
0.56 yes right
0.56 and that really comes back to things
0.56 like autonomy you know being
0.56 really good at sort of the range and
0.56 those kind of capabilities and
0.56 and of course once we unlock these you
0.56 know then the awesome next step that&;s
0.56 going to happen i&;m very confident in
0.56 our lifetime is the flying car step
0.56 right we&;re going to start we&;re going
0.56 to start commuting and flying drones and
0.56 that&;s going to be awesome too yeah i
0.56 mean i think that
0.56 that for a lot of people will probably
0.56 the most impact
0.56 biggest change in in their day-to-day
0.56 life if they if they fly to work instead
0.56 of drive to work though a lot of people
0.56 working from home of course now but
0.56 um and just think about commuting in 3d
0.56 instead of on
0.56 on the roads it&;s going to be so fun
0.56 it&;s going to be so great
0.56 and so quick it&;s going to change
0.56 everything the flying cars are going to
0.56 change the
0.56 economies in really big ways can you say
0.56 a bit more
0.56 well just think of the basics of just
0.56 home prices right like
0.56 you know but you know if you&;re in the
0.56 san francisco bay area for example right
0.56 if you&;re sort of near
0.56 the big companies your home prices is
0.56 twice as much as you are if you&;re
0.56 20 30 miles away but when i say 20 30
0.56 miles away i mean like
0.56 as the drone flies right and like uh you
0.56 can buy a house for literally half the
0.56 cost
0.56 uh and that that drone flight will take
0.56 you all of like 10 minutes
0.56 uh but driving through the highway
0.56 system can take you an hour and a half
0.56 so people will pay
0.56 for that really expensive house instead
0.56 of instead of living over the
0.56 side of the hills or whatever it might
0.56 be what i&;m personally very curious
0.56 about
0.56 but related to the applications you
0.56 mentioned and curious if you&;ve seen
0.56 anything is
0.56 um fire prevention and fire fighting
0.56 any thoughts on that yeah definitely i
0.56 mean the fire fighting is really
0.56 exciting because
0.56 just the tanker style operations being
0.56 able to do that without people in those
0.56 in those planes is going to be
0.56 uh it&;s i mean it&;s just it&;s an
0.56 incredibly dangerous thing to have
0.56 people flying these tankers into fires
0.56 um going in and doing inspection for a
0.56 spot like looking for hot spots
0.56 also really powerful to do with drones
0.56 uh this is one thing that i
0.56 this is one thing that happens i think
0.56 there&;s a company boeing has a company
0.56 that does this for
0.56 firefighters at scale here in california
0.56 it&;s not very well advertised
0.56 uh but they will fly these drones that
0.56 are designed for military surveillance
0.56 into the fires to look for hot spots
0.56 um and it&;s uh it really helps the
0.56 firefighters know what&;s happening in
0.56 the fire and fire
0.56 fight the fire much faster and more
0.56 safely
0.56 now seems like those are things we&;ve
0.56 all imagined are going to be part of our
0.56 future
0.56 and maybe the big question at least on
0.56 my mind is you know
0.56 when you talk about these applications
0.56 what kind of timelines do you think
0.56 because i think we&;ve all seen the
0.56 jetsons with flying cars
0.56 and we know 100 years from now sure but
0.56 what do you think that&;s a great
0.56 question i i
0.56 drone delivery to homes in the u.s you
0.56 know it&;s it&;s
0.56 it&;s going to start this year i i think
0.56 i&;m almost 100 sure of it
0.56 wow this would be a more remote
0.56 locations um
0.56 i think the more the more dense
0.56 locations you know three four years
0.56 feels very practical to me um that&;s
0.56 where the technology is and i think
0.56 i&;m reasonably confident that&;s where
0.56 the regulators will be here in the us
0.56 um i think flying cars like flying cars
0.56 have a
0.56 longer road to hoe but i if that took 20
0.56 years i&;d be surprised i&;d be surprised
0.56 if it took that long
0.56 i think it&;s more like the 10-year time
0.56 frame
0.56 for a number of reasons um i think
0.56 there&;s there&;s a chance 20 years of
0.56 time frame for like where the with
0.56 economics because make it mainstream
0.56 and 10 years is sort of a more of like a
0.56 premium service
0.56 um but uh yeah there&;s there&;s not that
0.56 much technologically in the way anymore
0.56 of flying cars
0.56 batteries today are good enough and
0.56 they&;re just getting better uh
0.56 learning how to fly in this kind of
0.56 weather is something that people like us
0.56 are
0.56 like really learning how to master um
0.56 the safety and the safety for the safety
0.56 systems around it like it&;s
0.56 it&;s in a lot of ways it&;s it&;s you know
0.56 more tractable
0.56 from an autonomy perspective than you
0.56 know well an autonomous car on the road
0.56 right for a lot of reasons well just to
0.56 be clear when you said
0.56 flying is something people like us are
0.56 mastering
0.56 you mean is it going to be automated or
0.56 are
0.56 you saying that people are going to be
0.56 actively flying
0.56 no i mean well the very early flying car
0.56 stuff that people will be flying those
0.56 but that&;s i i think that&;s more from a
0.56 regulatory
0.56 sort of kind of walk crawl run than
0.56 anything else
0.56 um and uh when i say people are
0.56 mastering i mean companies like zipline
0.56 are getting really good at flying
0.56 in this kind of weather and this kind of
0.56 in this sort of low altitude like how to
0.56 fly reliably
0.56 um and i think that what
0.56 in the long run you know autonomy is
0.56 going to win there there&;s there&;s
0.56 there&;s as my pilot friend is always
0.56 saying
0.56 you know when you look at it when you
0.56 look at the cockpit in his little in his
0.56 little two-speeder
0.56 most of the buttons there he has to
0.56 remind himself not to push because if he
0.56 does like
0.56 you know bad things will happen right uh
0.56 and
0.56 it turns out that like uh you know this
0.56 this is something that
0.56 zipline has gotten very good at is the
0.56 the time like what
0.56 we think we call it the low-level
0.56 autonomy this is like we call it sort of
0.56 even at that health and status of like
0.56 hey you have a problem with this part of
0.56 your hardware what are you gonna do
0.56 about it
0.56 like we have software systems that are
0.56 very good at doing the right thing about
0.56 that
0.56 and they&;re not they&;re good at it they
0.56 do it in a fraction of a second
0.56 they make the right decision a fraction
0.56 of a second and and they&;re they&;re you
0.56 know
0.56 they&;re auditable they have all the
0.56 features you want uh that the regulators
0.56 want but don&;t know they need yet
0.56 um and to do this at scale and so um i
0.56 think that&;s gonna be a big part of it
0.56 is is uh well it&;s basically you know
0.56 you have to learn this
0.56 it&;s all a learning process uh and and i
0.56 don&;t see anything technologically sort
0.56 of fundamentally blocking flying cars
0.56 it&;s just it&;s just it&;s basically
0.56 already here um
0.56 and once the batteries get a little bit
0.56 better it&;s going to be here in a cheap
0.56 way too
0.56 that&;s very exciting yeah much looking
0.56 forward to that
0.56 you do now another thing i&;m really
0.56 curious about you
0.56 i mean we talked a lot about zipline but
0.56 you also did another company before
0.56 willow garage and i believe between
0.56 willow garage and zipline you were part
0.56 of a few other startups
0.56 and some kind of curious zooming out
0.56 um i mean there&;s a lot of people out
0.56 there that
0.56 you know did love to start a company and
0.56 and it generally seems a great way to
0.56 have a lot of impact in the world by
0.56 starting a company versus joining an
0.56 existing company
0.56 you have some advice that uh you can
0.56 share from your experiences
0.56 my high level advice on startups is
0.56 pretty simple like
0.56 do it because you have to right like if
0.56 someone has to talk you into doing it
0.56 don&;t do it you won&;t enjoy it because
0.56 it&;s hard but you and you will enjoy the
0.56 hard if it&;s something that you really
0.56 have to do
0.56 i mean you have to do it like you just
0.56 like you can&;t control yourself right
0.56 you&;re just really excited to go do that
0.56 that&;s the first thing um and the second
0.56 thing is is you know
0.56 customers first right and literally
0.56 especially for your technology company
0.56 so many technology companies make the
0.56 excuse of like ah
0.56 there&;s too much technology to do here i
0.56 need to make some demos or whatever
0.56 before finding the customer
0.56 i have come to believe that like that is
0.56 never true like it doesn&;t matter how
0.56 complex your technology is if you find
0.56 customers first
0.56 and then develop the product based on
0.56 the technology with those customers
0.56 like the everything gets easier when i
0.56 say everything i literally mean
0.56 you know the investors i mean most
0.56 investors want to invest in customers
0.56 right if you have customers investment
0.56 comes
0.56 hiring great people like most you want
0.56 to inspire someone to join your crazy
0.56 adventure
0.56 and you can talk about an actual
0.56 customer with an actual need that you
0.56 really understand that will inspire
0.56 people to join you
0.56 um product decisions technical decisions
0.56 what you can cut
0.56 and what you can&;t right if that&;s if
0.56 those debates in your company are
0.56 anchored in
0.56 actual customers you can literally go
0.56 visit and get to know
0.56 those decisions become easy like
0.56 literally all the hard parts of a
0.56 startup become much much much easier
0.56 if it&;s all guided by those customers um
0.56 and yeah i find if you do those two
0.56 things it&;s almost hard to screw up
0.56 if you&;re really listening to your
0.56 customers don&;t get me wrong as a geek
0.56 it&;s easy to act think you know talk
0.56 convince yourself if you&;re listening to
0.56 your customer or not but if you really
0.56 listen to them and are willing to like
0.56 you know let them drag you a bit
0.56 uh i don&;t mean like go ask them like
0.56 what do you want but go really
0.56 understand their problem
0.56 and like let that process of
0.56 understanding their problem
0.56 you know guide you off of your initial
0.56 course
0.56 like you you&;ll be successful it&;s it&;s
0.56 like success is almost guaranteed if you
0.56 really
0.56 listen to the customer keenan it&;s been
0.56 absolutely wonderful to have you on
0.56 it&;s been far too long since we got to
0.56 catch up
0.56 and i&;m absolutely stunned by the
0.56 tremendous impact zipline has had
0.56 saving lives hard to imagine a better
0.56 story of robotics being taken out of
0.56 research labs into
0.56 the real world thank you peter this has
0.56 been absolutely fantastic
0.56 it&;s so cool to catch up and well anyway
0.56 it&;s also just been so fun to follow
0.56 both your lab work and your
0.56 startup work and uh it&;s very inspiring
0.56 and uh well
0.56 literally educational we read your
0.56 papers
0.56 well i&;m sure you managed to educate
0.56 a lot of people listening in today but
0.56 now if anybody listening in
0.56 just like me would like to keep learning
0.56 even more from kenan
0.56 i highly recommend checking out the
0.56 company&;s website
0.56 flyzipline.com and also highly recommend
0.56 following keenan on twitter where he
0.56 tends to share
0.56 ziplines latest innovations and impact
0.56 that&;s being achieved
0.56 on twitter that&;s add kitten wyroback
0.56 and the same goes for linkedin
0.56 thanks again keenan cool thank you
0.56 you
.