WEBVTT - Problems Solved: Drones, Bananas and Real Estate* 0:00:15.316 --> 0:00:22.316 Pushkin. It's been a year since we started this show, 0:00:22.956 --> 0:00:25.116 and we started it because I wanted to talk to 0:00:25.196 --> 0:00:29.476 people who are trying to make technological progress, people who 0:00:29.476 --> 0:00:32.996 are struggling with big, interesting problems, who go to work 0:00:33.036 --> 0:00:34.916 every day and try to figure out how to do 0:00:35.036 --> 0:00:38.156 things that no one in the world knows how to do. 0:00:39.156 --> 0:00:41.636 And now enough time has passed since we started the 0:00:41.676 --> 0:00:43.956 show that some of the people we've talked to have 0:00:44.196 --> 0:00:48.036 in fact figured some things out, they have solved some problems. 0:00:52.036 --> 0:00:55.036 I'm Jacob Goldstein, and this is What's Your Problem? On 0:00:55.156 --> 0:00:58.396 today's show, we're following up with three people who we've 0:00:58.436 --> 0:01:02.716 spoken to over the past year. First Keenan Wirobek. He 0:01:02.796 --> 0:01:05.716 may have solved the key problem that has been preventing 0:01:05.996 --> 0:01:09.796 drone delivery from taking off in the United States. Then 0:01:10.116 --> 0:01:13.316 Katherine Sizov, who seems to have figured out how to 0:01:13.356 --> 0:01:16.876 get ripe bananas onto grocery store shelves at the right time. 0:01:17.396 --> 0:01:20.676 And finally Glenn Kelman. He is the CEO of the 0:01:20.756 --> 0:01:24.036 real estate company Redfinn and since we talked to him 0:01:24.116 --> 0:01:26.076 last he, like a lot of people in the real 0:01:26.196 --> 0:01:29.316 estate industry, has had something of a tough time. What 0:01:29.436 --> 0:01:32.636 I've been through just personally has sometimes felt like a 0:01:32.756 --> 0:01:36.596 spiritual death of the past six months, but he has 0:01:36.596 --> 0:01:40.276 come through it with some really illuminating insights. We'll hear 0:01:40.316 --> 0:01:46.596 those later in the show. First up is Keenan Wirobek, 0:01:46.876 --> 0:01:50.316 co founder and CTO of Zipline. Zip Line is a 0:01:50.396 --> 0:01:53.036 drone delivery company. They uperated in a bunch of countries 0:01:53.076 --> 0:01:55.836 around the world. They've done hundreds of thousands of flights 0:01:55.916 --> 0:02:00.556 in Rwanda, where their drones deliver medicines across the entire country. 0:02:00.876 --> 0:02:02.956 But when I talked with Keenan last year, he was 0:02:03.036 --> 0:02:07.236 up against this kind of combination technical slash regulatory problem 0:02:07.396 --> 0:02:09.596 that was making it hard for him and other drone 0:02:09.636 --> 0:02:13.156 delivery companies to expand in the United States. Here's what 0:02:13.196 --> 0:02:17.196 the problem was. In other countries, there are strict rules 0:02:17.236 --> 0:02:21.276 that require planes to have transponders, these devices that allow 0:02:21.396 --> 0:02:24.556 drones and planes to detect and avoid each other. But 0:02:24.756 --> 0:02:28.436 rules about transponders are more lax in the US, and 0:02:28.596 --> 0:02:32.036 as a result, commercial drone companies have to figure out 0:02:32.196 --> 0:02:38.156 how to detect and avoid planes that are flying without transponders. Today, 0:02:38.636 --> 0:02:42.156 drone delivery companies like zip Line in the US solve 0:02:42.236 --> 0:02:46.156 for this in a kind of ridiculously low tech and 0:02:46.356 --> 0:02:50.156 labor intensive way. They actually pay people to sit on 0:02:50.156 --> 0:02:53.236 the ground and stare at the sky watching for planes. 0:02:54.476 --> 0:02:56.996 This obviously is not going to scale. We're not going 0:02:57.036 --> 0:02:59.356 to have a drone delivery business when you have to 0:02:59.356 --> 0:03:01.556 pay people to sit on the ground and stare at 0:03:01.596 --> 0:03:05.596 the sky for drone delivery to be a real economically 0:03:05.676 --> 0:03:09.116 viable service in this country, someone needs to figure out 0:03:09.196 --> 0:03:13.796 how drones can automatically detect and avoid planes. When I 0:03:13.836 --> 0:03:18.476 talked to Keenan last year, that was ziplines big problem. Now, 0:03:18.556 --> 0:03:21.516 after years of trial and error, he thinks they've solved it. 0:03:21.916 --> 0:03:24.396 Keenan told me the story of how they figured it out. 0:03:24.916 --> 0:03:27.996 So we went on a journey here. Um we started. 0:03:28.036 --> 0:03:30.836 We actually started with radar because radar was the thing 0:03:30.876 --> 0:03:34.676 that sort of the holy grail of FA regulators understand radar. 0:03:35.516 --> 0:03:37.276 It has the benefit of being able to see through 0:03:37.276 --> 0:03:40.476 clouds and things like this. Okay, so why didn't radar work. 0:03:40.836 --> 0:03:44.476 The regulators want drones to be able to sense aircraft 0:03:44.516 --> 0:03:47.076 coming from any direction and avoid them, and this is 0:03:47.116 --> 0:03:49.436 a big challenge. And so if you put enough radar 0:03:49.516 --> 0:03:52.676 on an aircraft, to see in all directions up, down, 0:03:52.796 --> 0:03:56.636 you know, left, right, front, back. It weighs something like 0:03:56.716 --> 0:03:59.436 five times more than our drone does. Okay, it's a 0:03:59.476 --> 0:04:03.116 massive radar system, and it's it's basically there's just no 0:04:03.156 --> 0:04:05.956 way to fly it. Okay, So it's too heavy. The 0:04:06.076 --> 0:04:08.756 simple answer is too heavy. Won't work. It's too heavy, 0:04:09.236 --> 0:04:11.196 too heavy, And so we went from there to cameras. 0:04:11.396 --> 0:04:14.196 It's like, okay, can you just look now? Cameras have 0:04:14.236 --> 0:04:16.956 some really interesting challenges, and I think the first challenge 0:04:16.996 --> 0:04:18.876 is easy to If you ever tried to take out 0:04:18.916 --> 0:04:21.036 your cell phone and like look for a plane in 0:04:21.076 --> 0:04:23.476 the sky, you've seen the first challenge. You almost it's 0:04:23.476 --> 0:04:26.156 almost impossible to see it on your phone. Right to 0:04:26.236 --> 0:04:27.836 our eyes, we can kind of figure out, oh, there's 0:04:27.836 --> 0:04:30.076 a plane there, but it's literally a couple of pixels 0:04:30.076 --> 0:04:32.596 to a camera, even a very high resolution camera, and 0:04:32.636 --> 0:04:34.916 so being able to tell the difference between you know, 0:04:35.036 --> 0:04:38.316 smudge on the lens a bird and a plane very 0:04:38.316 --> 0:04:42.516 difficult to do that reliably. Okay, So radar won't work, 0:04:42.836 --> 0:04:47.996 cameras won't work. What do you try? So this is 0:04:48.036 --> 0:04:49.676 sort of a back to the drawing board moment of 0:04:49.716 --> 0:04:52.396 like what on Earth could work. So if you remember 0:04:52.396 --> 0:04:54.436 the early hearing aids, that were these cones that you 0:04:54.476 --> 0:04:56.916 would stick in your ear and use still hear something, 0:04:57.436 --> 0:05:00.756 So personally remember them. But I have seen photos. Yes, 0:05:00.836 --> 0:05:05.836 you see photos. Okay, yes, the So before radar was 0:05:05.876 --> 0:05:08.956 invented the Brits in the United States, we built each 0:05:09.276 --> 0:05:13.476 huge cones, like twenty foot size cones that humans would 0:05:13.476 --> 0:05:15.276 stick their ear onto the end of the twenty foot 0:05:15.276 --> 0:05:17.996 size cone and this they point the cone at the 0:05:18.036 --> 0:05:21.236 horizon to try to hear aircraft coming before you could 0:05:21.236 --> 0:05:24.636 see them. Wow. So when was that in use? World 0:05:24.636 --> 0:05:27.636 War One and then early World War Two until radar 0:05:27.996 --> 0:05:30.276 was invented, which was mid World War two. So just 0:05:30.356 --> 0:05:32.876 be a dude with his ear on a giant cone 0:05:32.956 --> 0:05:35.876 and if he heard a plane, he'd say whatever, everybody 0:05:35.876 --> 0:05:40.596 into the bunker, planes coming, scramble the planes intercept this 0:05:40.676 --> 0:05:43.876 kind of thing. Yeah, and you know it kind of 0:05:43.876 --> 0:05:45.876 makes sense, right, Like as humans were pretty good, we're 0:05:45.876 --> 0:05:47.836 walking around, you hear a plane, you look up, you're like, oh, 0:05:47.836 --> 0:05:50.156 there's the plane. Right, Um, we don't see a plane first. 0:05:50.196 --> 0:05:53.756 We hear a plane first. And that was the inspiration 0:05:54.276 --> 0:05:56.356 and it was like, okay, well, we've tried the other 0:05:56.516 --> 0:05:59.716 the things that seemed more obvious, let's try this. And 0:05:59.716 --> 0:06:01.716 that this was one in turn was working with me, 0:06:02.036 --> 0:06:04.156 and it was basically saying, hey, let's you know, the 0:06:04.396 --> 0:06:07.596 philosophy is fail fast, right, It's basically, let's try to 0:06:07.636 --> 0:06:10.996 understand the technical reasons why this is impossible as quickly 0:06:11.036 --> 0:06:14.276 as possible. Let's rule it out. Yeah. So the number 0:06:14.276 --> 0:06:16.596 one reason to be skeptical is it you've ever been, like, 0:06:16.676 --> 0:06:18.836 you know, on a bike or you know, in a 0:06:18.916 --> 0:06:22.036 motorcycle or a car and it's quiet in the car, 0:06:22.236 --> 0:06:23.956 But on a motorcycle you put your head, you're out 0:06:23.996 --> 0:06:25.556 the window, and you hear all this noise. You can 0:06:25.556 --> 0:06:28.076 barely hear the person, you know, biking next to you. 0:06:28.516 --> 0:06:31.196 That's called aero acoustic noise. It's basically the noise made 0:06:31.196 --> 0:06:34.156 by the rush of air over your well. Yeah, when 0:06:34.156 --> 0:06:35.916 you roll down the window in the car when you're 0:06:35.916 --> 0:06:38.996 on the freeway, it's super loud. It's super loud, right, 0:06:39.036 --> 0:06:41.556 And so these planes, the microphones are in the rushing 0:06:41.556 --> 0:06:43.676 wind created by the plane in the sky, by the 0:06:43.716 --> 0:06:46.396 drone in the sky exactly, So is there Basically the 0:06:46.476 --> 0:06:48.636 question was, with all that noise from that rushing wind, 0:06:48.676 --> 0:06:51.996 would we still hear those aircraft far enough away to 0:06:52.036 --> 0:06:54.396 be able to avoid them? That was the biggest thing 0:06:54.436 --> 0:06:58.196 that we were worried about. Um, but we we we 0:06:58.316 --> 0:07:02.116 found a way using a microphone array. So basically, you know, 0:07:02.116 --> 0:07:04.876 a set of microphones on the aircraft to be able 0:07:04.916 --> 0:07:07.876 to literally listen for where aircraft are and use that 0:07:07.916 --> 0:07:10.156 to avoid them. So so you're saying it works, You 0:07:10.236 --> 0:07:13.996 figured it out, Tell me how it works. Yeah, So 0:07:13.996 --> 0:07:18.236 so the final system across the wing, um, there's eight 0:07:18.276 --> 0:07:20.956 microphones and they're on. They're on, and they're on little 0:07:20.996 --> 0:07:22.836 they're on little polls. They look like basically like a 0:07:22.956 --> 0:07:24.916 you know, I don't know the length of a straw, 0:07:24.996 --> 0:07:27.316 like a drinking straw coming out the front of the 0:07:27.396 --> 0:07:30.876 of the micro of the wing. There's eight microphones and 0:07:31.196 --> 0:07:33.836 now you need a bunch of microphones to do basically 0:07:33.836 --> 0:07:36.196 what your ears do, which is beam form, which is 0:07:36.236 --> 0:07:39.276 basically you figure out how the sound hits one microphone 0:07:39.316 --> 0:07:41.236 a little sooner than the other one, and it helps 0:07:41.236 --> 0:07:43.436 you understand, oh, the noise is coming from over there 0:07:43.556 --> 0:07:45.476 versus you know, coming from your right side versus your 0:07:45.516 --> 0:07:47.716 left side, so that that's what it looks like on 0:07:47.756 --> 0:07:51.156 the aircraft. And then okay, so this is the hardware piece. 0:07:52.036 --> 0:07:54.636 Is there a sort of software layer. I've got the 0:07:54.636 --> 0:07:58.916 hardware piece in my mind. You're gathering this sound. I mean, 0:07:59.676 --> 0:08:01.836 I mean, I know everything sounds like an AI problem now, 0:08:01.876 --> 0:08:04.076 but this one really does sound like an A problem. Right. 0:08:04.076 --> 0:08:06.996 You're pattern matching because you want to say it, does 0:08:07.076 --> 0:08:09.316 this sound like a plane? Let's test it against a 0:08:09.356 --> 0:08:12.116 billion other sounds? And I don't know how what's the 0:08:12.156 --> 0:08:15.796 software piece work? Like, Yeah, there's a big software piece. 0:08:15.796 --> 0:08:18.756 The software has been far and away the biggest effort here. 0:08:18.796 --> 0:08:21.036 There's a there's a tiny team of hardware engineers and 0:08:21.076 --> 0:08:23.956 a big team of software engineers working on this. So 0:08:23.996 --> 0:08:26.596 how does this software work? So the software has layers, 0:08:27.236 --> 0:08:29.676 and the first layer is a layer called beam forming, 0:08:29.716 --> 0:08:32.956 and beam forming is this technique very similar to what 0:08:32.956 --> 0:08:35.116 our ears used to figure out? Okay, where are the 0:08:35.156 --> 0:08:38.476 sound sources coming from? Right now? That's the first layer. Okay, 0:08:38.556 --> 0:08:40.236 that's cool. So now you know where the sound is 0:08:40.236 --> 0:08:42.276 coming from, but you don't know what it is yet 0:08:43.116 --> 0:08:45.356 exactly so, and then the next layer is an AI 0:08:45.436 --> 0:08:50.756 layer where we've we've done thousands of flights with different aircraft, 0:08:50.796 --> 0:08:52.996 different sound sources, and we've used that data to train 0:08:53.036 --> 0:08:58.836 AI model to actually figure out basically it reinforces where 0:08:58.836 --> 0:09:00.556 it's coming from and to figure out, hey, this is 0:09:00.556 --> 0:09:02.876 probably an aircraft, this is probably not an aircraft. And 0:09:03.356 --> 0:09:06.476 then the AI just says, yes, it's an aircraft. I've 0:09:06.556 --> 0:09:09.876 detected and now avoid avoid avoid or what what's the 0:09:09.956 --> 0:09:12.156 last So there's two there's two more layers here. So 0:09:12.356 --> 0:09:15.316 the next layer is it's called the tracking layer, and 0:09:15.316 --> 0:09:17.716 the tracking layer is you kind of think of that 0:09:17.716 --> 0:09:20.116 that layer is the layer that remembers over time, right, 0:09:20.156 --> 0:09:23.076 So this takes the information from this AI layer and 0:09:23.236 --> 0:09:25.636 over time says, oh, you think something's kind of over 0:09:25.676 --> 0:09:27.516 in that direction, I'm gonna I'm gonna keep track of that, 0:09:27.836 --> 0:09:30.556 and every every you know, every fraction of a second, 0:09:30.556 --> 0:09:32.196 it gets an update and it kind of updates that 0:09:32.236 --> 0:09:34.356 track of sort of where it thinks that sound is 0:09:34.396 --> 0:09:37.276 coming from. And then the final layer takes those tracks 0:09:37.316 --> 0:09:40.636 of where uh these these and these tracks are estimates, 0:09:40.636 --> 0:09:42.316 all right, you kind of think of them as like, hey, 0:09:42.316 --> 0:09:44.676 I'm there's you know, there's an there's an aircraft in 0:09:44.716 --> 0:09:47.396 that direction kind of over there. And the next layer 0:09:47.516 --> 0:09:49.276 is the layer the planning layer, the layer that takes 0:09:49.276 --> 0:09:51.716 that information and says, okay, there's an aircraft over there, 0:09:51.996 --> 0:09:55.396 I better go over here to say away, I better 0:09:55.436 --> 0:10:01.636 go in the other direction, exactly exactly. So okay. So 0:10:01.636 --> 0:10:05.836 so now you have solved this non trivial technical problem. 0:10:06.036 --> 0:10:08.996 There is also a regulatory problem, right, you have to 0:10:09.236 --> 0:10:13.836 get approval to fly your commercial drones without people on 0:10:13.876 --> 0:10:15.956 the ground looking at the drone the whole way. The 0:10:15.996 --> 0:10:18.636 FAA has to say, yes, we believe that this works 0:10:18.636 --> 0:10:23.036 and you can fly your drones. Do you have that approval? Yes, 0:10:23.076 --> 0:10:25.316 But let me explain it because, as in all things 0:10:25.356 --> 0:10:28.676 with from a regulatory perspective, it's subtle. So what's really 0:10:28.676 --> 0:10:32.676 exciting is we just got yeah, yes, Asterius. So there, 0:10:32.716 --> 0:10:35.996 basically we're about to start this walk crawl run sort 0:10:36.036 --> 0:10:38.996 of progression, which is how you should start a new 0:10:39.036 --> 0:10:41.716 technology like this. So we just got approved to fly 0:10:42.236 --> 0:10:45.556 this in operations in the United States. So in the 0:10:45.636 --> 0:10:47.996 crawl walk run, the crawl phase which we're about to start, 0:10:47.996 --> 0:10:51.156 which is basically this technology turned on, so doing the 0:10:51.156 --> 0:10:54.156 avoid les, sensing and avoidance, but with still these visual 0:10:54.156 --> 0:10:56.956 observers standing on the ground staring at the sky and 0:10:56.996 --> 0:10:59.036 then later this year we go through a process with 0:10:59.036 --> 0:11:01.476 the FA to remove the visual observers and keep operating. 0:11:01.676 --> 0:11:04.796 And that's the holy grail moment that's coming soon. So 0:11:06.316 --> 0:11:09.796 the holy grail moment is flying a commercial drone flight 0:11:09.996 --> 0:11:14.676 in the United States without someone watching from the ground exactly. 0:11:15.636 --> 0:11:19.116 And when do you think that will happen? If I'm 0:11:19.156 --> 0:11:21.476 being optimistic, it's about three months from now. If I'm 0:11:21.476 --> 0:11:24.516 being more realistic, three to six months from now. It 0:11:24.596 --> 0:11:28.196 seems like this has been the big underappreciated by the 0:11:28.236 --> 0:11:33.836 general public problem that has prevented commercial drone delivery from 0:11:33.876 --> 0:11:38.716 taking off. Excuse my fun in this country? Is that right? 0:11:38.876 --> 0:11:40.676 And like you think you've solved it, and the FA 0:11:40.676 --> 0:11:45.956 thinks you've probably solved it. Yes, this is Yes, it 0:11:45.996 --> 0:11:48.516 has been a long few years to get through this journey, 0:11:48.676 --> 0:11:51.156 but I am over the moon excited about this because 0:11:51.156 --> 0:11:55.196 this is this is the big unlock. Great, good to 0:11:55.236 --> 0:11:58.076 talk to you. Can we talk in a year, Let's 0:11:58.116 --> 0:12:02.476 do it, okay, twenty twenty four, back to you then, excellent. 0:12:05.676 --> 0:12:07.916 One other piece of zip Line news that Tenan told 0:12:07.956 --> 0:12:11.036 me about. The company is working on this new kind 0:12:11.076 --> 0:12:14.116 of drone. It's this wild two part system where the 0:12:14.196 --> 0:12:17.876 drone itself hovers like three hundred feet above the delivery site, 0:12:18.116 --> 0:12:21.156 and then it lowers down this little box on some 0:12:21.276 --> 0:12:23.356 kind of wire. The box goes all the way down 0:12:23.356 --> 0:12:26.276 to the ground, opens up, delivers the package, and goes 0:12:26.316 --> 0:12:28.276 back up into the drone. They hope to have that 0:12:28.476 --> 0:12:31.996 in operation by early twenty twenty four, so that'll be 0:12:32.036 --> 0:12:34.036 one more thing for me and Keenan to talk about 0:12:34.076 --> 0:12:39.076 next year. Still to come on today's very special episode 0:12:39.116 --> 0:12:44.436 of What's Your Problem, The Banana Ripening Frontier also how 0:12:44.556 --> 0:12:47.996 higher interest rates are profoundly changing not only the real 0:12:48.116 --> 0:13:00.316 estate business, but lots and lots of businesses. Next up, 0:13:00.996 --> 0:13:04.356 Katherine says Off, founder and CEO of a company called 0:13:04.636 --> 0:13:09.956 Strella Biotech. Catherine and Strella, they developed these sensors that 0:13:10.076 --> 0:13:14.196 detect when apples and pears are getting ripe. They work 0:13:14.316 --> 0:13:16.156 not like in the grocery store in your house, but 0:13:16.236 --> 0:13:20.196 in these giant storage rooms where apples and pears sit 0:13:20.276 --> 0:13:24.476 and ripen. The idea behind the product was to reduce waste, 0:13:24.636 --> 0:13:28.276 increase efficiency, that kind of thing, and the sensors are 0:13:28.276 --> 0:13:31.276 based on this thing that happens in nature. There are 0:13:31.396 --> 0:13:33.876 lots of kinds of fruit where when they get ripe, 0:13:34.156 --> 0:13:37.756 they emit ethylene gas, and that gas is like a 0:13:37.796 --> 0:13:41.916 signal that causes nearby pieces of fruit to ripen more quickly. 0:13:42.796 --> 0:13:45.196 When Catherine and I talked last year, she was trying 0:13:45.196 --> 0:13:48.356 to solve ripening problems for other kinds of fruit besides 0:13:48.396 --> 0:13:51.356 apples and pears. When I followed up with her earlier 0:13:51.356 --> 0:13:53.596 this month, I asked her, is there some new kind 0:13:53.596 --> 0:13:56.636 of fruit that you've cracked since we last talked? Bananas? 0:13:57.076 --> 0:14:01.076 Actually bananas. Yeah, it's a big one. Yeah, it is 0:14:01.076 --> 0:14:04.956 a one. Tell me about bananas. What happened and when 0:14:04.956 --> 0:14:07.276 did it happen? Yeah? I guess my first question is 0:14:07.316 --> 0:14:10.996 how do you like your bananas? Because, says a heated topic, 0:14:11.836 --> 0:14:14.236 I prefer them a little too ripe as opposed to 0:14:14.236 --> 0:14:17.516 a little bit unripe, Like if it's still kind of green, 0:14:17.556 --> 0:14:19.596 if it's still hard to peel, if when you peel 0:14:19.636 --> 0:14:22.276 it it's hard to peel, and then you eat it 0:14:22.276 --> 0:14:25.796 it's a little chalky. That's that's not my favorite banana. 0:14:25.876 --> 0:14:27.916 I'm one of those freaks that like, say, a little 0:14:27.956 --> 0:14:30.596 bit underripe, actually, But I think you're in the majority. 0:14:30.916 --> 0:14:34.356 So what have you figured out about bananas, Like what 0:14:34.436 --> 0:14:37.436 was the problem and then what did you figure out, 0:14:37.556 --> 0:14:40.116 Like what was the status quo whatever a year ago? 0:14:40.156 --> 0:14:42.996 What was the banana problem? So the banana problem is 0:14:43.036 --> 0:14:44.916 that every time we go to the grocery store, the 0:14:44.956 --> 0:14:47.156 bananas tend to be a little bit on the green side. 0:14:47.236 --> 0:14:49.636 I don't know if you have this from personal experience, 0:14:49.716 --> 0:14:52.396 but they tend to actually edge towards green and then 0:14:52.436 --> 0:14:54.676 sometimes it goes to the other direction where they're way 0:14:54.716 --> 0:14:56.716 too ripe and no one wants to buy them. Right. 0:14:56.996 --> 0:15:00.076 But basically what happens is that they're brought into North 0:15:00.076 --> 0:15:04.276 America totally under ripe, so they haven't they're not mature whatsoever, 0:15:04.316 --> 0:15:07.196 and they're green. When you peel them, it's very hard 0:15:07.196 --> 0:15:11.236 to peel them. They ooze latex, so they're like if 0:15:11.276 --> 0:15:13.236 you try to peel it, it's it's kind of almost 0:15:13.276 --> 0:15:17.036 like bleeding latex. And then what happens is these bananas 0:15:17.036 --> 0:15:20.876 are put into what are called ripening rooms. So basically 0:15:21.476 --> 0:15:24.036 bananas go into these rooms and then they're dosed with 0:15:24.076 --> 0:15:28.476 ethylene gas, which is what makes them ripen. And this 0:15:28.676 --> 0:15:30.996 is just to restate when we talked about last time, 0:15:31.036 --> 0:15:34.476 like in the wild a ripening piece of fruit or 0:15:34.556 --> 0:15:37.716 ripening banana emits ethylene gas and that is a signal 0:15:37.756 --> 0:15:40.236 to other surrounding bananas that they should also ripen. Right, 0:15:40.236 --> 0:15:43.036 So the dosing with ethylene gas is just doing what 0:15:43.116 --> 0:15:46.396 happens in nature. That's exactly right. So they're put into 0:15:46.436 --> 0:15:49.396 what's called a ripening room and they're dosed with ethylene gas. 0:15:49.756 --> 0:15:52.396 And then comes the art of it, which is that 0:15:52.556 --> 0:15:56.556 every day a ripener that's a job will go into 0:15:56.596 --> 0:15:58.916 that room and say, hey, or these bananas more green 0:15:58.996 --> 0:16:01.516 or less green than what I was expecting, And they 0:16:01.556 --> 0:16:05.076 just look at them visually and they say, mm, doesn't 0:16:05.076 --> 0:16:08.076 look quite yellow enough, and then they adjust the conditions 0:16:08.076 --> 0:16:12.276 of the ripening room base on that. So it's a superhuman, 0:16:12.556 --> 0:16:16.196 super subjective way to do it. And we thought, you 0:16:16.196 --> 0:16:18.516 know what, this is probably the cause of a lot 0:16:18.516 --> 0:16:22.596 of inconsistencies. Every single ripener has a different idea of 0:16:22.596 --> 0:16:25.196 what a stage seven banana looks like or stage four 0:16:25.236 --> 0:16:27.916 banana looks like. And so let's take the human out 0:16:27.916 --> 0:16:30.316 of this, let's take the art out of it. We 0:16:30.436 --> 0:16:33.476 know about fruit physiology, let's turn this into a science. 0:16:34.076 --> 0:16:36.916 So what we did was we put our sensors inside 0:16:36.916 --> 0:16:39.396 these ripening rooms, and we can tell how the bananas 0:16:39.436 --> 0:16:42.636 are reacting to their conditions and how they're ripening. And 0:16:42.676 --> 0:16:44.796 then we throw some machine learning on top of that 0:16:45.116 --> 0:16:47.556 and we say, hey, this banana is going to be 0:16:47.836 --> 0:16:50.156 under ripe if you keep going the way that you're going. 0:16:50.436 --> 0:16:53.116 So we're basically going to the retailer and helping them 0:16:53.196 --> 0:16:56.796 ripen their bananas for the store shelf. Oh huh, so 0:16:57.356 --> 0:16:59.516 how does that work? So a grocery store can you 0:16:59.596 --> 0:17:01.636 name it? Are you working with some grocery store chain? 0:17:01.676 --> 0:17:05.156 I would have heard of a big one in Canada. Okay, 0:17:05.756 --> 0:17:10.556 can you say the name? I can't, sorry, like hundreds 0:17:10.556 --> 0:17:13.796 of stores across Canada or something. Yes, So, so how 0:17:13.836 --> 0:17:15.916 do you go into the grocery store? So you go 0:17:15.996 --> 0:17:18.236 talk to a ripener and they say, you know what, 0:17:18.436 --> 0:17:20.956 You're totally right. This is a huge problem for me 0:17:20.996 --> 0:17:22.756 and I have a million other things to do, and 0:17:22.796 --> 0:17:25.036 I will Oh. I wouldn't have expect that. I would 0:17:25.036 --> 0:17:28.676 have expected them to say I'm a genius at ripening. 0:17:28.836 --> 0:17:33.676 Go away. That happens too sometimes. But I think the 0:17:33.716 --> 0:17:36.396 ones that you know in general, when you're when you're 0:17:36.556 --> 0:17:38.716 good at your job and a new tool comes along, 0:17:38.756 --> 0:17:42.316 You're like, hell, yeah, that's awesome. So I think most 0:17:42.316 --> 0:17:47.716 people are really interested. So okay, so you go, you 0:17:47.756 --> 0:17:50.356 talk to the ripeners. Keep going. So we talk to 0:17:50.356 --> 0:17:52.556 the ripeners. They're like, I have a million things on 0:17:52.596 --> 0:17:54.436 my plate. Ripening in bananas is one of them. I 0:17:54.476 --> 0:17:56.236 don't know if I'm doing a good job. Would love 0:17:56.276 --> 0:17:58.796 to see what's up. So then we put up We 0:17:58.916 --> 0:18:01.636 set up shop in a couple of ripening rooms. We 0:18:01.716 --> 0:18:04.556 work with the ripener to understand what they're doing, why 0:18:04.596 --> 0:18:06.716 they're doing it, and train our model on it. And 0:18:06.756 --> 0:18:10.356 then we expand and how long did it take you 0:18:10.796 --> 0:18:14.196 for your sensors to be better than a ripener? It's 0:18:14.196 --> 0:18:17.036 we're still going. So we're kind of well, let me 0:18:17.076 --> 0:18:20.516 ask you, are you better than a human ripener? Now? Well, 0:18:20.556 --> 0:18:23.516 it depends, right, So if what we can do is 0:18:23.556 --> 0:18:25.636 we can basically give a tablet to a person off 0:18:25.636 --> 0:18:27.236 the street and you can go ripen a bunch of 0:18:27.276 --> 0:18:30.076 bananas and not do a terrible job at it. So 0:18:30.396 --> 0:18:32.596 that's pretty important because it usually it takes three to 0:18:32.636 --> 0:18:36.156 five years to train a ripener, yeah, in the art 0:18:36.156 --> 0:18:38.996 of banana ripening, But now we can Now we can 0:18:39.036 --> 0:18:41.196 kind of nagate all that in terms of like the 0:18:41.236 --> 0:18:46.036 ogs of banana ripening, I don't necessarily think so, but 0:18:46.396 --> 0:18:48.916 with a little bit more time than I think, we'll 0:18:48.916 --> 0:18:53.076 get there. So basically you can make anybody as good 0:18:53.076 --> 0:18:56.916 as the sort of average five years of training banana ripener. 0:18:57.036 --> 0:18:59.076 That's right now, yep. How long did it take you 0:18:59.156 --> 0:19:01.716 to figure to do that? Oh? My god, well, I 0:19:01.716 --> 0:19:03.876 mean I have. I didn't talk to you too long ago, 0:19:03.956 --> 0:19:06.476 but to me it feels like basically an eternity. It 0:19:06.596 --> 0:19:08.156 was a year. It was a lot of year for you. 0:19:08.516 --> 0:19:12.036 You did a lot of very long year. When you 0:19:12.076 --> 0:19:13.676 think you'll be doing it in the US, when do 0:19:13.676 --> 0:19:16.516 you think some US grocery store will be buying your service? 0:19:17.236 --> 0:19:22.156 We're in trials with several grocery stores and distributors, now, okay, 0:19:22.396 --> 0:19:30.196 in the US, yes, anybody you can name. No, okay. 0:19:30.676 --> 0:19:35.196 Automating the task of figuring out how to ripen bananas 0:19:35.436 --> 0:19:39.476 is not some world changing technology, But you can zoom 0:19:39.476 --> 0:19:42.956 out a little bit here. Creating machines to automate one 0:19:43.036 --> 0:19:46.356 task or another has been at the very core of 0:19:46.476 --> 0:19:50.596 technological progress for hundreds of years. In the long run, 0:19:51.196 --> 0:19:55.316 little automations like this are like the fundamental efficiency game, 0:19:55.676 --> 0:19:57.916 getting more and more efficient at lots and lots of 0:19:57.956 --> 0:20:00.716 little things. That is how over the long run we 0:20:00.756 --> 0:20:04.156 have raised living standards for most of the people on Earth. 0:20:06.116 --> 0:20:09.916 In a minute, we'll talk tech and real estate giant 0:20:10.156 --> 0:20:15.436 macro economic fluctuations with Glenn Kelman, the CEO of Redfinn. 0:20:23.636 --> 0:20:27.036 About a decade ago, this new kind of thing popped 0:20:27.116 --> 0:20:30.316 up in the real estate world. It was called ie buying. 0:20:30.796 --> 0:20:33.276 If you wanted to sell your house, you could go 0:20:33.316 --> 0:20:35.636 on to a few sites on the internet, enter some 0:20:35.676 --> 0:20:39.436 information about your house, and a big company like say Zillo, 0:20:39.836 --> 0:20:43.676 would actually buy your house for cash. Then they would 0:20:43.756 --> 0:20:45.676 fix it up a little turn around, and try and 0:20:45.716 --> 0:20:49.876 sell it for a profit. Zillo famously lost a ton 0:20:50.076 --> 0:20:52.636 of money on this business and they got out in 0:20:52.796 --> 0:20:58.116 late twenty twenty one. Another big eyebuyer was Redfinn. Redfinn 0:20:58.156 --> 0:21:00.956 is a big brokerage operates all around the country. And 0:21:01.036 --> 0:21:04.036 when I talked to Glenn Kelman, the company CEO, last year, 0:21:04.316 --> 0:21:07.956 Redfinn was still in the eye buying business. But Glenn 0:21:08.036 --> 0:21:10.876 knew that if house prices started falling, it would be 0:21:10.916 --> 0:21:12.716 tough to stay in that business, and he told me, 0:21:12.796 --> 0:21:15.996 so in very dramatic terms. I brought this up with 0:21:16.076 --> 0:21:19.356 him when I talked to him again earlier this year. So, look, 0:21:19.556 --> 0:21:23.076 the last time we talked was in the first half 0:21:23.076 --> 0:21:26.996 of last year, and we talked about iBuying, and you 0:21:27.116 --> 0:21:30.356 said that I buying would be good on the way 0:21:30.476 --> 0:21:33.836 up and on the way down you said, in your words, 0:21:34.076 --> 0:21:39.196 it would be a roller coaster ride to hell. Did 0:21:39.436 --> 0:21:47.516 say that? Remarkable prescience, remarkable prescience. So tell me what 0:21:47.716 --> 0:21:51.156 happened with Redfin's I buying business since the last time 0:21:51.156 --> 0:21:53.996 we talked in the first half of last year. Actually, 0:21:54.196 --> 0:21:57.316 we closed the business, and we did that for two reasons. 0:21:57.476 --> 0:22:01.556