Autonomous drone delivery in a Zip (Interview)

24:35It's too easy. Again, depot.dev is where to go. It all begins at depot.dev.

24:44How much does regulation play into this rollout? I assume every area probably has different rules about how you can go about flying your drones at 300 feet. Yeah, I wish everybody had rules. I think that everyone we've been. Usually people say they want less rules when they're trying to scale up. Yeah. I mean, when you're a startup and you go to a regulator and you're like, cool, I would like to do this thing. And they're like, well, that's not allowed. And then basically from there, it's like all by exception. And so you have to kind of work it out, the rules by which you operate at the detail.

25:19And as a startup, that's a slow process. Right. And startups need to go fast to survive. And, yeah, that's why I say I wish there was some precedent. In some ways, obviously, there's some luxury in getting to help create the precedent. But, yeah, it's a lot of working with the regulators. You know, some things about airspaces between countries are very similar, like harmonized. But other things are very different. And so, yeah, like things we have to do here in the U.S. have been very different than Rwanda. Yeah, it's a big part of what we do. But it's also, we've gone from that being a big challenge, you know, maybe 10 years ago to now it's something that we've gotten quite good at and quite good at working with these partners and these regulators as partners and working with them proactively enough that we can generally work out what needs to be worked out before we need it to serve a customer.

26:08But it takes a lot of effort. I will say that. Yeah, I was thinking of the luxury of not having rules that are already defined because those can slow you down, but actually not having rules and not having a clear path towards engagement. That means you have to trailblaze and to trailblaze, you got to get out a machete and hack away a bunch of stuff, you know, and that can be even slower and more laborious than if somebody else had already cleared the path for you. So that makes sense. There's a great story. Here in the United States, starting about six years ago, we started working on the sensor system on the aircraft that can sense and avoid other aircraft.

26:43This is a system that uses machine learning. And machine learning is a class of software that the regulators have no experience with, literally. And it took us about five years of working with them to go from the beginning of this to them approving us to operate. And that was the first time they've ever approved AI, you know, ML-based software for safety in the airspace. And so it was near daily conversations with their teams figuring that out over that five-year journey. And there's some people at the FAA and at Zipline who I respect a ton for working through the minutiae they had to work through to figure out how do you, what would the requirements be?

27:23And how would you validate the system against those requirements? And then actually doing all that work and all that analysis to bring it back to the FAA and be like, OK, we did what we agreed to. Like, you know, let's go through this together. And, yeah, that was a five-year journey there. And it was a hard-won battle. And when I say battle, I don't mean like Zipline versus the FAA. It was very, very much in it together because they definitely wanted to figure it out, too. Yeah, it was something else. This is an example where the word regulation, which you haven't quite said yet, gets really weird.

27:56weird and I don't want to go into politics, but it seems to be the word that gets thrown around regulation, less regulation, more regulation. But as a developer, you know, I really lean back on what you said. There was not a presence of which was specifications, which is really what regulation is. It's an adherence to a specification that everyone agrees on for obvious reasons. But you're in this kind of unique high stakes scenario where, you know, maybe regulation could be less, you know, less pressured, so to speak. and in your case, you got to be, you know, a high stakes scenario where if there's a crash or if there's a, you know, break a leg situation, which I'm not sure you like the word crash.

28:32I wouldn't like the word crash if I was in your business personally, but I just think about the need for that. Can you speak to how that's played as an ally versus a foe? Yeah. I think there's a couple layers for how we think about this. The first one is because everywhere we've been everywhere we've gone, there hasn't been some, like, like you said, specification to follow. We've taken the mantra that like our first job is to convince ourself and not in like a hand wavy way. It's like, no, no. Like, you know, this is over our house over our kids. Like, you know, are we convinced that we've taken a rigorous approach before we suggest that approach to a regulator?

29:07And, um, that's gone a long way, right? It's a, that mindset has attracted some of the best talent is figuring some of this stuff out is really complicated. It's trying to attract the talent we need to actually figure this out. And then it's also forced us to do the work, right? To convince ourself, the analysis, the, you know, all the, all the testing, all that data review to convince ourself is also what the regulators want to see, right? So if we've done that to ourselves and we've really been skeptical of our, uh, internally first, um, and it helps a ton when, when you're then having these in-depth conversations with these folks on the regulatory side who are, They have the completely thankless job of like, okay, we got to figure out the specification for this new thing that no one's done before.

29:51And so if they underdo it and someone gets hurt, they look really bad. And of course, if they overdo the regulation, then everybody in the industry complains of like, oh, they wrote down these regulations that are completely impractical. And so they're stuck between a rock and a hard place. And so by doing that legwork up front and internally proving to ourselves something that we really believe in first and then bringing it to them, that's been core from our perspective. And then the only other piece of it that I would say is it's all relationships. These are all people on the regulatory side.

30:25They've got a job to do. We've got a job to do. And so we spend a lot of time basically working together to make sure we understand what they're trying to do. They spend a lot of time working to understand what we're trying to do, be really open-minded, make sure we're hearing each other out. like in any relationship and try to make progress every day, even if it's not as fast as you might like. How would you frame, this isn't an exact one-to-one when I say this, but how would you frame test coverage, you know, in terms of a coverage level, confidence level, when you ship something and it's in a regulation standpoint or your specification you got to adhere to, is there like 95%, 98%, like what do you strive towards when it comes to adherence to a spec or defining the spec?

31:04What is confidence to you and your team? Oh, great question. So there's a model for thinking about this that was called the Swiss cheese model. I did not come up with that name, although I'm known for naming things with really dumb names. It was actually a NASA name, the Swiss cheese model. Conceptually, basically what it is, it's like every approach to testing is like one slice of your Swiss cheese. It's going to have good coverage of some aspects, but if you look at it, it's going to have holes. for every type of testing you do. You obviously have many different layers of testing that you do.

31:41And the intent is if you stack up those layers of Swiss cheese and look at them through the stack, then you're not going to see through holes if you're doing it well. And we spend a lot of – so examples of our layers of Swiss cheese, some of them are going to be familiar with things like unit testing and static testing and things like that that are commonly done in software, software-in-the-loop testing. So this is where you have full-on physics-level simulation running of your full system, even your full fleet. And you're able to then write test scenarios on top of that. We talked a little about the hardware in the loop testing, where you take the hardware itself, the brain of the aircraft, plug it into a simulator and fool it into thinking it's in the real world.

32:17That's great for testing, you know, like the actual like hardware level components. Like if you short out a certain bus, does the actual full software stack detect that short and do the right thing? And then a lot of things in flight tests. We do all kinds of different flight tests. We do it on the order of 10 ,000 plus flights a week at this point across our flight testing. And almost every single one of those flights is pushing the system to an extreme in some way. We actually have this fully automated software system we call Chaos Monkey, where basically what it's doing as we're dispatching all these flights is Chaos Monkey is literally throwing chaos into the system as those missions are happening.

32:57that either will cause the airplane to do something extraordinarily dynamic against the physics limits of the aircraft or turning off subsystems of the software, killing various rotors, things like that, and making sure that the system, despite in this given flight having dozens of weird things happen, that off-nominal events, as we call them, happen in that flight, does the whole system handle it gracefully and safely? And this is just on the software side, right? When we think about hardware, a very similar thing. at many layers of testing of the hardware to make sure we understand deeply its performance and its reliability.

33:32Yeah, and there are entire teams of folks who look at all that testing and basically are constantly questioning, hey, okay, in this release, we have a new capability we're releasing. Did we add enough testing to all the different layers of the Swiss cheese such that that's going to be responsible to ship? And there are times we think we're ready and we do that review of everything we did and we're like, nope, this isn't ready. that we're pulling this from this release. It's got more work to do just on the testing side to get that confidence. These releases, are they, share as much as you like.

34:05You seem to be very sharing. I like this. I think about releases and I think about, wow, if I released my test level from idea to new capability, as you said, into production, there's probably several layers that are in that middle ground, which is like a true release, but not in a real world scenario, so to speak. How do you plan releases? How do you graduate releases from tests, delivery, observability, test again, potentially? You said you do 10 ,000 flights a week. That's seven days in a week. That's 1 ,400 flights a day, roughly. Gosh, that's a lot of flights. But how do you release and do that in a way that's like, okay, it's in production, but it's not real production.

34:48I'm assuming these things. So walk me through that kind of idea. Yeah. Okay. So depending on the scale of a given feature, I'll talk about one of the biggest features we've ever released. We just released it recently. So we talked about the drone that flies up high. It has a wing and can hover. Until about a month ago in our releases, in our operations, we were only hovering. We were not flying on that wing. So we were flying it more like a traditional quadcopter. It's a, it's a pentacopter, but trying flying like a traditional quadcopter, just hover, which means the range is very limited. But while we were working on the, actually being able to fly on the wing and that's the, the, the, the software and systems to transition from hover onto fixed wing is what's referred to in the industry.

35:32And then back into hover, including when anything could be going wrong, like any of those rotors could not be working and that kind of thing is there's a lot to that. And we've been working on that for, we're working on those features, you know, at some level of development testing for two years to enable this. And so about two or three months before that release cuts. So when I talk about releasing every six weeks, we cut basically at the beginning of that six weeks. We will look at all the new features that folks think are ready for a release. And we'll look at how tested are they? How proven are they?

36:05How good is the test coverage and all these layers of Swiss cheese? And then we'll decide, okay, hey, yep, this feature is ready. This feature is ready. No, that feature is not ready. It's not going to go in this release. And then we'll cut a release. And then once we cut a release, there's a bit of a stabilization that takes us about a week to make sure that release is really stable. And then it goes into this very structured release campaign that takes about five weeks on average, which is, you know, depends on the release, but something like 50 ,000 flights, a whole bunch of hardware loop tests and software loop tests and all these other tests we will do again as well as part of that release process.

36:39And then all the teams are basically paying a ton of attention to all the data coming out of all that testing. And over the years, we've developed all kinds of tooling and dashboards that help all the teams look at that data and look for anything weird, anything that we consider a near miss. Obviously, the real miss is if we have a parachute landing during that testing, we're going to know exactly what goes on there. But we're looking for anything that's at all weird. We call it a near miss. And studying it and seeing if it needs to be addressed or if it's fine. Seeing if it sometimes when we do, you can kind of think of that flight test as like the last leg.

37:17If we find something in flight test, that's a red flag to us. Like, hey, the other layers of Swiss cheese aren't good enough. So if we find something in flight test during that campaign, we're immediately investing in those lower, those other levels of Swiss cheese to make them better and improve their coverage. And then only once the data says we've got a, you know, we've got a system that meets or exceeds the overall target level safety of our last release, do we then say this is ready to actually share. Do you have the idea of a black box in your thing? I mean, I know that airplanes do. So I imagine you got some version of a black box where if it does, I'm sorry, crash, that you've got some sort of thing in there.

37:54And then I've also think about like observability because you've got all this testing happening. And I think about like traditional software in the cloud, it's kind of easy to orchestrate observability. But you've got over the year, I'm sure you've got some version of Wi-Fi or access to the thing digitally via a network of sorts, telemetry going. How does a black box, is there a black box and how does observability happen in those scenarios? Yeah, so absolutely. So we log a ton of data on the vehicles at all times. And that's, of course, done in all this testing, but also in production. and that's the philosophy is that like anything that could possibly go wrong or be a near miss you got to study and and you mentioned crashes we i should mention we have a we have a parachute system for the aircraft that's kind of the backup to the redundancy so there's redundancy in the aircraft so it can nominally fly when things don't work but the backup to the backup is a parachute so it'll float to the ground like a it's actually made for us by a skydiving parachute company so it's like a skydiver coming to the ground and so yeah if we ever have a parachute landing in tests or operations or anything that's even close to a weird data.

38:59Like we go study that to basically deeply, deeply understand it because quite often it's a new insight. We're so far in the long tail of the problem that many of the problems we see now, they happen one in hundreds of thousands of flights. And so you basically, like if we didn't log it, think of it as like, oh no, like now we have to wait for this to happen again in hundreds of thousands of flights, which is a total shame. Total shame. Should happen more often. Exactly. No, no, that's not the shame. The shame is not the logging. So we log these things. Now, in addition to the logging to enable us to go through those logs and understand everything that happened, we use those logs not just for root-causing a specific problem.

39:38We use those logs to understand things in aggregate. And so we have tools where you can mine those logs longitudinally, like across thousands and thousands of flights, to deeply understand the statistics of a certain thing. That's really important as well. And then, of course, during live operations, yeah, the aircraft are reporting over radios, like live what's going on. We have a remote operations center that's monitoring all of that. And so during live operations, we have certain things that basically, our pilots, I think of them more like an air traffic controller where they can be like, okay, you go back and dock and that kind of thing and understand what's going on at a very sort of fleet level.

40:16Have you ever logged a tornado? We've definitely, yeah. So the closest thing we've logged, we've logged some tornadoes, some very small, I don't think you'd call them a tornado. I think they're like basically the things that are too small to be considered tornadoes. And some of our testing, we're actually chasing some tornadoes right now as we look for hail and more extreme weather events with our mobile test rigs. The craziest things we have a lot of log data on is what's called a microburst. So it's the beginning of a thunderhead formation, about the first 90 seconds of a thunderhead. Yeah. You have the vertical winds at like 50, 60, 70 miles an hour vertically in the middle of the microbursts.

40:52On the sides of the microbursts, the same thing goes down. Those are some of the most extreme weather events that are hard for the drones to handle. What do they do in those circumstances? Do they turn around and leave or do they land immediately? Or what's their protocol? So it really depends how big it is. This is actually one of the first things we developed AI to forecast. It's because if this is really big, right, if it's kilometers across and really strong, then we can't, you know, if we're in it, we're going to end up parachute landing. And so we know this because in a lot of our operations around the world, when we're doing emergency deliveries for, you know, patient on the table delivering blood situations, we will take, we will basically ignore our weather limits and fly in very extreme weather.

41:33So we go through these things, but then for non-emergency deliveries, we rely on our forecasting to keep us out of this stuff in those really extreme events. If they're small enough, yeah, the control system will fight its way through it. But of course, you know, the, the, the, the, the aircraft has its physics limits, right? It can only climb so fast. They can only do so much. And so if in certain situations, it can't overcome them. And yeah, that's when the parachute kicks in. I assume there's some sort of wind limit that there's just no chance for anything to fly through successfully. It's got to be around a hundred miles per hour or so, maybe a little bit less.

42:05I don't know, but there's certain storms where it's just like, you better just parachute out because there's no, they're there. There's no success as a possibility regardless of the size of this thing like you know 30 seconds at 150 miles per hour is knocking down massive trees here and so i'm assuming it's taken a drone off its course quite a bit there's a certain mountain pass in rwanda i'm very familiar with because we have to fly through this mountain pass to serve about a third of the country there and in this mountain pass it is the winds get whipping like the winds can be higher than we you know we fly we fly that aircraft at like, what is it?

42:4160, 70 miles an hour. And the winds, yeah, get up to a hundred miles an hour. And you'll see our drones flying backwards as it tries to fight through it. Developing the control system and logic to like, because most of those winds are gusty, right? So if you can kind of hang, if you can, if you can stay in the air long enough and stay out and then punch through it at the right moment, you can get to the mountain pass. And yeah, it was a big breakthrough when we figured out the, like how to make that logic robust so that you see the drones just flying backwards for a while, flying backwards for a while, flying backwards for a while, and then floored at the right second.

43:14You know, it's estimating the winds dying down online, and then it punches through the mountain pass and gets through. Do you have any video feeds? Because you're going to come up with some great drama in those moments if you could capture video just when it's really gnarly, you know? I got a great video of the logs of that. It's not funny. You can see the position on the map going backwards and things. I don't think Hollywood's going to call you for the log file. No. Let's talk more about that wing. So did you guys just put the wing on there because you knew eventually you were going to use it?

43:43And so you just had it in manufacturing for a long time, but couldn't use it? Yeah, absolutely. We knew we needed the range. You get about 10x more range flying on a wing for the same energy than hovering. And our customers need that range. So many of our use cases need that range. And so, yeah, we always knew we needed the range of flying on a wing. And we decided to launch without having developed that system yet. Really, just that's how we're wired. We want to get out into the real world fast because we just know when you're in the real world serving customers, you're going to learn a lot of things you just can't learn any other way.

44:13And so we want to get out there earlier. Yeah. And that hardware was there. And the teams that spent all this time developing that hardware were like, we want to see it fly on the wing. And, of course, we've been doing the test sites for a long time as we've been developing that capability. It is one of those things where you kind of develop it and then wait. And then once you can get, once you can, yeah. One of the things, you know, you're talking about regulations and safety, like that is also part of our ethos. And a difficult thing in a startup is we hire people who have like inherent urgency.

44:41Because if you set a deadline on safety, you can end up in a really bad situation. Right? If you're like, cool, we must ship this on this date. Some of the stories you hear about of AV companies that have hurt people in the past, it's that, they take that mindset. And they try to, and then they'll fool themselves into thinking they're ready. as opposed to saying, no, no, no, we set a safety bar and we decide how we're going to measure ourselves against that safety bar. And that's when we ship. It means your dates aren't quite as predictable. And so if you, but you still need to move fast. And so you need people with a very inherent sense of urgency to work through those problems and get there as opposed to like a more traditional, like, you know, we have a date and we're going for it management approach.

45:18So what was the range prior to the wing and then of a typical drone, like assume good conditions, Like how far do they go and how long a time? And then you said 10 X, so I can do the math, but that's a huge win. Yeah, great. Yeah. So, so we're doing about a mile and a half hovering and we'll be able to do right now. We're at about five miles on the wing and we'll be able to take that to about 10 miles over time. So again, as we continue to validate and sort of work our way up to it, once we feel confident in it. And I assume each drone launches from a location and then has to return to the same location, at least until you get multiple hubs in a city or something.

45:56Yeah. Yeah. That's actually, that's a software feature that's coming in and released next year, is that ability to go from one dock, go make a delivery and then dock somewhere else. But yeah, you're exactly right. We don't have the density of docking locations yet to need that. Right. So that cuts your range in half because you have to get out there and you got to get back. Exactly. Okay. So when you say five miles, is that if I live five miles from your facility, do I get deliveries or I got to be two and a half miles? No, you can live five miles. Yeah, so once we have that, we can dock at a different location nearby the delivery site.

46:30Yeah, we can do more than 10 miles. Right. That's actually an interesting thing. We've learned over the years it's way better when you're working with these partners to talk about a conservative range than where you're going to go in the future. Because it's so complicated to talk about that when you get into the details of your terrain and your physics and your local weather and all that stuff. It's like, no, no, no. We just talk about this little circle. And then they're thrilled that we can do more than that. Right. I was just flying my drone the other day. I had a little DJI flip and I was flying it too far away, like slightly too far away to where it was telling me to come back.

47:05And I'm like, yeah, I'm coming back. So I start flying it back. And then the wind picks up going against it on the way back. And I'm sitting there thinking like, oh, man, it's just not going to make it back, you know, because the battery percentage continues to go down and it's still 1 ,200 meters away. And it was very dramatic. It did make it back to the house, thankfully, but just barely. I got to imagine that's a stressful, you know, we're talking about range anxiety with our new EVs. I'm sure you probably have extreme forms of range anxiety with your, because these are expensive machines, right?

47:34What is, you know, what's an all-in cost on one if you can share that to build and manufacture? What are we talking about? I can't share that specifically, but it's lower than you would think. And this is one of the reasons we develop our own drones, both for this range problem you're talking about and for the cost. But for the range, take our fixed-wing platforms. This is what we do these very long-range deliveries with. There's no hover capability. It just flies like a plane. We advertise basically a 50-mile service radius. But if it wasn't windy and there wasn't a mountain in the way, it could easily do a 100-mile service radius.

48:07So we literally have twice as much energy as you would need nominally so that we can make it back on those windy days reliably. And that's just really important because once you start serving a hospital delivering blood, you can't be like, oh, it's too windy today. Sorry, guys, we don't serve you. We only reach the people closer. It's like, no, we turn you on and we want to serve you all the time. Does anybody get underwhelmed by the range when you say five miles? I'm not throwing stones here, but I'm just saying, like, I know you're innovating, but does anyone ever get like, Like, that kind of seems small.

48:38It all just depends on the use case, right? Like this is like that. The 10 miles was developed going really deep with our health partners and exactly on all the use cases that they want to use the drones in metro areas. And of course, when you're going out of metro areas, that's where the 50 mile range drone comes in handy. And all of this has been developed just, you know, basically technology second, customer first. So deep with the customers. You know, we even have like for this precision delivery platform, we had contracts with partners, you know, in the while we were still in what we would call like the conceptual phase of that project.

49:16Right. Nothing was locked in yet. And having those contracts with the partners gets them to give you like that gets them invested enough in us to give us the feedback and the data to make sure that we make something that's going to be useful for them. And so, yeah, some people get overwhelmed, but we just have that confidence that we've had so much time in these use cases, analyzing them, going deep with our partners to just have confidence. We're making a platform that'll that'll serve their use case as well.

49:44Well, friends, I'm here with my good friend, Chris Kelly over at Augment Code. Chris, I'm a fan. I use Augie on the daily. It's one of my daily drivers. I use Cloud Code. I use Augment, Augie, and I also use AmpCode and others. But Augie, I keep going back to it. And here's where I'm at. I feel like not enough of our audience knows about Augment Code, not enough about Augie, the CLI. It's amazing. I love it. What can you share? Yeah, we often say Augment is the best coding assistant you've never heard of. And that's both frustrating as someone that works there and is like very proud of the work we've done, but also like inspiring.

50:20Like we want to go and sort of punch above our weight because like we aren't anthropic and we aren't open AI. And so the quality of the product itself, you know, with our context engine, once you do touch it, people are like just blown away by that. And so like that keeps me going every day. So not to bear the lead here, but this is a paid spot. You are sponsoring this show to get this awareness. Now, at the same time, we're selective and I love to use your tool, but there is in the world. So a lot of developers look at the space and they say, OK, well, how long can this work? How long is this sustainable in the case of Cursor or Windsurf?

50:57Or you pick the name and you think discounted tokens. Help me shape a lens for our audience. I think it's a lot of awareness, right? Like Cursor got a lot of publicity early on for like fast revenue growth, which well deserved. I think, you know, frankly, some of the media gets the story wrong and that like if I gave you$1.50 for every dollar you sent me, I'd be the fastest growing startup in the valley. And so when you're selling discounted tokens, yes, of course, you're going to grow very fast. But all that money plus more goes to the model providers. So I think the real story is the story of Anthropic and, you know, being an API provider.

51:36I think the market has just moved so fast and there's so many pieces of competition out there that it's just hard to get noticed. So, friends, I love Augment Code and I love using Augie and I highly recommend you use it. I love using Augie. I can hand Augie a well-defined specification, a well-defined PEP, as I call them in my world, an agent flow, and it executes flawlessly. So the cool thing about Augie that I love most really is that context engine. And I can hand it a task. And I can just churn away on my well-defined plan and just never bother me and accomplish the mission. It is so cool, leveraging the latest models, the context engine, and all the fun things behind the scenes in that awesome CLI.

52:21So, yes, go try it out, augmentcode.com. Right in the top there is a CLI icon, a terminal icon. Click that, install it, and change your world. It's going to be awesome. Augmentcode.com. And also by our friends at Framer. You know, most design tools, they lock you behind a paywall. Well, Framer flips that script. It is a free, full-feature design tool that does something most site builders cannot. It's actually designed for designers. Framer has already built the fastest way to publish beautiful, production-ready websites, but with design pages, they've defined and redefined what it means to design for the web.

53:00This isn't a Webflow clone or a WordPress competitor. It is a true design platform, vectors, 3D transforms, gradients, wireframes, all the tools you actually use. And they're all free. Unlimited projects, unlimited pages, unlimited collaborators. And here's the kicker. You design, you iterate, and publish all in one place. There's no figma handoff. There's no messy HTML imports. There's no tool switching. So if you're ready to design and publish in one tool, start creating for free at framer.com slash design and use the promo code changelog for a free month of Framer Pro. Once again, framer.com slash design.

53:48Could you walk us through the user experience that either exists now or will be existing as you guys roll out domestically in the United States? from the end user perspective, but then also as we go, maybe explain the software systems that have to support that because I'm fascinated where software meets the real world. And there are all sorts of weird situations that you have to account for. I know there's an app. I assume it's similar to a DoorDash kind of a thing where I'm going to order, let's just say, I'm going to get a Chipotle burrito because I already cracked that earlier. And I'm a fan, not a sponsor, but I'm a fan of the old Chipotle burrito.

54:24And I know it's under eight pounds, so it's going to fit into my zip. Unless I get the double. I don't know. Can you get a 10-pound burrito from Chipotle? Maybe not yet. Someday. Someday we can dream. I open up the Zipline app, I assume. Or do I open up the Chipotle app? Maybe that's where we start. Do I start in the Chipotle app? Do I start in the Zipline app? Where do I start? Yeah. So soon you'll be able to start in the Chipotle app. Today you start in the Zipline app. And depending on where you live, if you live in the best place in the Dallas-Fort Worth area, you open that app and you're going to see a bunch of different storefronts.

54:54You'll see Walmart, very soon Walmart Pharmacy, you'll see Blue Bell Ice Cream, you'll see Chipotle, you'll see, well, just a whole bunch of great restaurants, Buffalo Wild Wings and others. And yeah, you pick what you want. That part of the experience feels a lot like DoorDash or someone driving the thing to you. Then they'll start prepping it. our partners will prep the food or prep the order. And then we put it in this thing that we call a zipping point, which is this fun little kiosk you put the package into. And that little, similar to how we deliver, the drone stays up high and the delivery zip comes down 100 meters, grabs the package out of the kiosk and brings it to you.

55:37And this is where things are very different than like delivery by car. We actually have to warn people, you know, hey, your food may be too hot to eat. um so we deliver so fast that like we've had people be like oh yeah like i just assume my my my hot wings would be soggy and they just stuck them in their mouth and burnt their mouth because they just like they're so used to how long it takes for deliveries to happen the old fashion way that when it takes only a few minutes it's it's uh and it's well insulated and controlled the way we do it like the food's actually fresh uh and sometimes very hot still yeah that's a great problem to have so show me the chipotle side like from your integrators your restaurants and i'm just picking on them because i know they're a partner of yours obviously there's going to be a bunch of them but imagine that i do order a chipotle burrito out of zip line i assume an api call kicks off over to the chipotle and makes the order and then they have to like somehow tell the zip when it's ready like how do they do that yeah so they do they enter with chipotle they use their software system the way they use it for any order right if you're ordering over the counter or you place an order to the Chipotle app and it shows up on their, they call their make line.

56:45That's, that's, that's the people actually like put the thing, put the order together. So they're going to make this like any, any other food they're making. Our system gets hit over the API. So we know when they started that and we know when they're done. And then when they're done, you know, we'll put it in that kiosk, dispatch the, the, the zip, the drone over to come get it. It'll come pick it up. That's usually about a minute or two after it's ready. And then we deliver a minute, anywhere from one to five minutes later, depending how far away you are. We deliver it in your yard. We have a fun little app that shows you the drone coming and gives you that really super precise ETA.

57:23I think one of the things I didn't appreciate about this kind of delivery that's really cool is just how human-centered it is. As opposed to whenever that driver is going to show up and sometimes it's like, wait, why did they take that crazy route? My kids are hungry. we tell you like no no within a minute of when we're going to get to actually show up and you can walk outside and pick it up you know it's just there now can you hold your hand out and it'll put it in your hand don't recommend it that could be a cool feature there can you come on put your hand out you know we'll put you can have a third little so you have like the big zip and the little zip and then like a hand that comes down you know and just drops a little bit lower Yeah, this is, you know, the, the, the, I literally test the way we, the way we talk about this is we don't want people to do stuff like that, but we test in case people do that kind of thing.

58:13So I've, it's delivered into my arms and I love it. It felt good, didn't it? It felt very, it felt special. Do you like pause the delivery then when you detect like anomalies you don't want to be present in? I assume that's probably a phrase or terminology. Like a child standing there underneath your target. It's actually kind of funny that originally we thought we, so we do a little bit of avoidance. So like if we're coming down and we see the person standing right there, we'll, you know, we'll scoot over and deliver next to them. They keep scooting closer. You keep scooting further. Exactly.

58:41If you do that, like, like you can make a really fun game for kids that we don't want to be a fun game. You know what I mean? Yeah. Like we don't want to make this to be a game of like, Oh, it turns out if you chase this thing, it'll, it'll become a game. And then everybody wants to chase these things and we can't deliver. And so basically what we found to work really well is just slow down. One of the things we like about that little delivery zip is it's very benign, right? If someone touches it or something, it's not a safety concern. And so, yeah, we just slow down and keep going and make that delivery and get out of there.

59:10And basically, the way I think to talk about it is like we'd be boring so that we're not creating a fun game for every kid. as a parent of young kids. Like I know, you know, I brought them to the test site too and I can see, like I know exactly what, if you did that avoidance with my, you know, my three-year-old, she would have so much fun and we would never make deliveries. Right. Well, it might be a consideration for down the road of getting like even more human-centric is like allow the orderer to set like a mode. Like, do I want to have fun with it? Do I just want my food? Because, not that you always, I know you guys want to, save energy and battery and all that.

59:50But like, you're already pretty happy to have your food delivered from the sky. Like I can imagine like that's a moment already, but like to have a little bit of a personality with the thing, I might take it over the top to where it's like, it's the only way we're ever going to get food is to just give us, you know, that's why I think the handoff would be actually amazing, but yeah, too dangerous perhaps to for sure today. Yeah. Oh, well, I love that idea. And there's this, your head is exactly where my head is. And it's, to me, it's not a question of if it's a question of when. So I look forward to it.

1:00:19Well, some will always mess around too in those scenarios, but did you mention at all when time of day you can deliver? Is there like danger zones where you can't deliver in the evening or after dusk or is there light concerns? Do you have lights on these things? Yeah. Yeah. So we can, we can deliver anytime. Yeah. We deliver 24 seven. Yep. Yep. Wow. Right now that's just limited by our partner hours. And most of the places we operate in the world, we deliver 24 seven with the health systems and then dependent, like, and it's always interesting to see like the, you'll work with a restaurant partner and all of a sudden they'll realize just how much volume we're adding to their normal business.

1:00:57And they're kind of like, Oh, wait a minute. Like if we operate with you all like, and you can go 24 seven, you're already operating 24 seven for other people. So we could do be open at night. And you know, there's a, there's a, there's a huge part of our economy that, you know, works at night, right. Night shift workers and things like that. There's, there's a, There's, well, we know that because all the health people we work with, they work, you know, 24 hours a day. But obviously, lots of manufacturing operates 24 hours a day. And a lot of our partners are seeing it like, wow, there's a whole part of commerce they can address and a big part of the demand they can address by, you know, operating more hours.

1:01:30And obviously, partnering with us, we can do that in a way that people really like. And so, yeah, it's really cool. You'll get whatever you need 24 hours a day. You know, as a parent, like just simple things, right? Like kid with spiking with a fever at night and you've ran out of the medication, right? To bring that fever down, right? Like dragging, the only thing worse than having a kid with a fever is taking your kid to like a 24 hour pharmacy at night to go grab that stuff you don't have. Yeah. Croup is a big deal for, for parents. I know that we've battled, you know, several late night ER visits for croup and generally, you know, I don't want to go into like the medical scenario here, but like, basically you have to be seen by some sort of medical professional to get a special medicine that is for croup that sort of helps your vocal cords and things, you know, reduce or whatever.

1:02:18And it's kind of a pain in the butt because you kind of know as a parent, you're like, you know what? I'm Dr. Mom, I'm Dr. Dad. Okay. I know I've been down this road before with my child. I just need that particular prescription. So if you had a relationship where you could do telehealth, which is super popular, you We would not have to leave the house at all if that were a scenario. We can make a phone call in our telehealth, and maybe 10 or 15 minutes later, we've got a delivery from Zip that says, hey, here's your recruit medicine. You'd have to leave the house. You can do it efficiently. You'd have to be on the roads.

1:02:49I mean, you talk about the efficiencies and environmental effects. We'd have to drive. We'd have to lose sleep and maybe ruin our day the next day more so we got the medicine. And there's so many, you know, unintended consequences in that scenario that just make it so interesting to think about how you can reorient the world around this being a thing. Oh, totally. I mean, I'm so excited about this. And we have we have we have we have partnerships now with about a dozen regional U.S. health systems. And and all of them have this use case in in that partnership. They have some other ones as well.

1:03:24but this is, they call it the physical side of telehealth. It's, yeah, telehealth is obviously becoming a really, you know, predominant thing here. And like you said, for so many things, like there's a physical side too. And so if you can get that inhaler delivered right away or that, you know, that crude medication delivered right away, it's huge for the family. It's huge for, and you're touching on something too that I didn't appreciate when starting Zipline that now I appreciate a lot is just how much health affects basically the economy, right? Like you mentioned, like just, yeah, the parents not being exhausted the next day.

1:03:54They're going to do, you know, their work is going to be better and like, and they're going to, they're going to be more productive in the economy and just keeping people healthy in that sort of preventative way is something that I've, I've become extremely passionate about because I've just seen this like cause and effect in my own life and in all the, in all the patients we serve around the world, just how powerful that is when, you know, a parent doesn't have to stay home or a parent doesn't have to, you know, basically be affected in their professional world by what could be a very preventable sort of health.

1:04:22You know, some of these things kind of feel like inconveniences even, right? Blood, obviously, very acute, big life-threatening. But like a lot of this stuff, it's just a little bit of convenience just goes a long way for people staying healthy and productive. Have you, I imagine you probably have, but have you imagined a world where zip is at scale, where zip line is at scale, and you don't have a couple hundred in production? You have, I don't know. Half a million. I don't know a lot just a an astronomical number that you don't even consider today how the world changes maybe even how we see the world.

1:04:57Cause like, I can't imagine having it imagined or experienced a zip line delivering something to me. I'm sure it'll be cool once I have it in my own life. But once this kind of technology, whether it's you or just generally like flights isn't only by only owned by United or Southwest, for example, you know, when this kind of delivery or scenario becomes ubiquitous, have you imagined that world and how that changes the world? Yeah. I mean, I'm super excited about it. This is part of the reason we spend so much time on things like making the system silent is, you know, one of the things we know is like, like, I don't want to notice these things, right?

1:05:34Like, you know, just like a bird flying over your house. How many birds flew over your house today? You have no idea. Why? Because we don't hear them. Yeah. And we want, like, we spend a lot of time on things like that because we just know that, like, as these things scale, there's certain things that we've learned from our experience that just matter a lot, like it being quiet. So there's certain things like that that really we spend a lot of time on to make sure that this is when this scales is it's a world we actually want to live in. And then, of course, you have the net effects of this kind of instant delivery.

1:06:00I live in a small town here in California. And yeah, like there's just two bakeries. They're just like like they're so excited for when they get their little kiosk and they can just, you know, send out fresh bread in the morning to people. Yes, I was just thinking about that. Timely deliveries. Exactly. You know, and there's so many things on the personal life side like that, like getting fresh bread delivered. There's so many things on the work side. There's so many jobs that just require stuff. Obviously, doctors and, you know, the medicines, it's so obvious. And, you know, auto mechanics, it's so obvious.

1:06:30Yesterday, just yesterday, my HVAC system in my house went out. A little blower had failed. And the guy came out and he's like, I'll be back in two days with the part. And it's like, no, no, no, you'll deliver the part right while he's there. And I looked it up the part he needed. It weighs less than eight pounds. we can deliver it. Did you deliver it? No, we don't serve that area yet with, uh, with HVAC parts. That would have been so cool. If you're like, Hey, I got a solution for this. Hang on. There's no delay. I got the part. Yeah. Even just talking to him about his business of like, you know, this is an independent guy who does a lot of these kinds of HVAC repairs and, you know, fixing heaters and stuff.

1:07:02And he's, and he said he, his customers, they hate it when he charges for these visits just to figure out what he needs. Right. Like, you know, he's, if he could just fix it there on the spot, like not only would he be able to serve so many more customers, but those customers would be so much happier. And he has somebody who like hates that part of the customer, you know, conversation of like, I got to charge you for the 200 bucks for just to come see, come by and see this thing. He'd love that to go away because he hates that part of his job. And anyway, there's so many pieces like that. I just can't wait to see what people imagine and how they will use it.

1:07:34Because every country we operate in, people are using it in ways I never expected. And I will tell you one fun story from Dallas. So a couple this summer, we were looking at our usage data. And most people would either order one address or order to maybe two addresses. But we saw this one person who was ordering to a different address almost every single day. What is this? Like, what's going on here? And so we send this user an email saying, would you share like how you're using the service? And so this is a postal delivery carrier and he, his New Year's resolution for this year was to eat healthier.

1:08:12And he was getting a salad delivered to him along his route every day for lunch. And I just like, anyway, I just love that. These like, I would have never have guessed. I would have never guessed. We call him the Johnny Appleseed of Zipline. What about gifting? It's a great way to gift. It's like, just, Hey, call somebody up, go outside. Why? It's like, I got, there's a delivery coming for you. And like this thing just drops a gift out of the sky. Like, Oh, we did this. So we did a really fun thing for, for Easter this year, where we basically loaded up the, the, the little delivery zip with a bunch of Easter eggs, no packaging, just Easter eggs.

1:08:48By the way, one cool thing about what we do is it doesn't require packaging. That's another part of the future I'm excited about. None of this, you know, cardboard jungle we all have in our garage from amazon boxes oh my gosh um save me so we put a bunch of easter eggs in there uh and it was really interesting to see how many we had grandparents in other states outside of dallas who were just delivering in dallas uh place orders for their grandkids uh to deliver like this you know pile of easter eggs into the yard and the kids loved it and the grandparents loved it and uh we had just sort of this out-of-state gifting it was so cool um and there's so many things like this that that i can't wait to try there's so many places you can go with eight pounds, but at a certain point that becomes your limit.

1:09:26I don't think you guys are anywhere near that point. And we all know that, but at some point, have you thought, okay, what's the next, like, where's the zip line jumbo? Like, what's the weight that we need to go to, like, expand to a new set of things? And then what do we need to manufacture in terms of drones to actually get that done? I know there's a long range one. Is there like a heavy weight one, or is that just too far down the line to care about? It's a great question. You know, eight eight pounds gets us really far as you're alluding to something like 80 % plus of Amazon packages are less than eight pounds.

1:09:57So many of the use cases that we are really focused on, like meal for family of four from almost any restaurant is less than eight pounds as an example. So, and then, yeah, if you need more, we just send two deliveries, I mean, back to back and yeah, it's two little bags of food and it's no boxes, none of this other stuff. So I'll be interesting to see how it evolves. If I was a betting person, I think we will increase our payload over time beyond eight pounds, there's a good chance it won't go a ton past that, maybe 10, maybe 12 pounds. And really what it comes down to is like what doesn't fit, right?

1:10:28You know, once we get to that scale, we can deliver a very full bag of groceries at that scale. And the advantage of having one platform as opposed to like different variants is we can drive the cost down further and further just from scale. And so as the cost per delivery goes down, there's a chance that we'll never be able to compete with like, if we made a drone that could deliver like four bags of groceries at once, that will always be more expensive than four of a smaller drone delivering four bags just 30 seconds apart onto your picnic table in your backyard. Yeah, and then there's stuff left out, right?

1:11:03Like the flat screen TVs, right? Like it obviously can't deliver those. I was thinking about a TV on a cell. I was like, can you bring the TV to me? I'm just kidding. I was not thinking that. That's cool. Yeah. And so, yeah, as we've been looking at the car, the partner we'll be launching within Dallas, you know, we can deliver something like 90 % of their car parts, but obviously 10 % we can't do. And, yeah, we think about that. And ZipLine has always just kind of been like, you know, customer first. So once we get the data, once we get the – once enough customers ask us for the same thing over and over again and we go deep with them, if we have conviction that there's a thing that will work for them that will make sense for the business, like, yeah, that's how this precision delivery, you know, drone inside of a drone.

1:11:43system came from. So yeah, we'll do more things in the future. Is anybody else doing this? Like who's your, who's the Southwest to your United? I know that Amazon announced a thing, like you said, years ago about drone deliveries and then that got canceled or sidelined or whatever, maybe it's happening, but I hope there's other people that are tackling the same problem because we want diverse and unique solutions and learning from each other. So who else is working in this space that maybe is either your competition or even that you admire what they're up to? Yeah, so Amazon has started doing some deliveries, quite limited as far as I understand.

1:12:17Google has a project called Wing. I admire a lot of how they're thinking about the problem, and we collaborate with them a lot on some of the projects with the FA. I really like how they think about some aspects of safety, for example, and things like that. I admire that a lot. There's a fair bit out there. I think the, you know, I'll be honest, I don't think a lot about the competition. For us, it's just about the customer and just like, what can we do? how can we do better every day and really measure ourself on our own velocity towards allotting those customers. Yeah. What's 2026 going to be about?

1:12:48Scale? Other things? Yeah. What's it going to look like? It's a lot about scale. So it's our long range platform is scaling around the world. So that will continue in 2026. And yeah, this precision delivery platform is scaling. Yeah, it is scaling really fast. And so a lot of what 2026 is about is building up the supply chain capability to manufacture faster, building all the internal muscles to launch the charging sites and things like that to enable scale. Yeah. And just getting better and better at that user journey you were asking about of like, how do people order it and how does that all work for all the different use cases?

1:13:26We found you can't be super generic there. Different use cases need a tailored user experience for it to really be magical. And so we spend a lot of time there as well. How do you pick your next city and why is it going to be Omaha.

1:13:43It's going to be Omaha because the, yeah, the, it's really about the testing and validation story. It's so, you know, I can't announce the next city, but we'll be launching our next cities in the first half of 26. Part of how we choose those cities is they're very similar to the Dallas Fort Worth area. Similar in terms of weather, similar in terms of terrain, that kind of thing. And that's because that's what we validated the system for. In parallel, we're validating the system for, you know, expanded weather, more complex terrain, more complex, you know, smaller yards, all those kinds of things.

1:14:20And so, you know, basically once we validated the system to go into metros that aren't like, you know, we refer to like the Sunbelt metros, then we'll move into these other metros as well. Yeah, it's really fun. We have a test site up in the Cascades where we're doing a lot of testing to validate the system for, you know, for metros that aren't at all like Dallas and Fort Worth. But yeah, it'll be, it'll be Sunbelt metros first and then, and then we'll, we'll keep going from there. Yeah. You've solved a ton of technical problems to get here and now you're trying to take it in lots of places. What are the big technical problems that are ahead of you?

1:14:54Like if you had a checklist as CTO, like here's the things we need to solve, like what's unsolved for Zipline? Almost every front, there's more work to do. I've been thinking a lot recently about weather forecasting. So we spend a lot of time to validate the system for certain weather. But then when you operate, you've got to know how do you not go in that weather you can't go in. And that's harder than it sounds to do with scale. There's some easy things like wind limits. Forecasting that is fairly straightforward. But there's much more complicated things like icing conditions. Right now, we're testing to understand exactly what level of icing conditions we can handle and how to handle that well.

1:15:31But there will be a limit there that we're going to need to forecast. Forecasting icing conditions is notoriously very hard to do. This is like when you have fog and clouds and cold weather. Oftentimes, you fly through that and just cover your aircraft in ice. And another good example is lightning conditions. We are currently chasing lightning with our mobile test rigs right now to really understand how does lightning interact with the system. And we think it won't based on analysis. But Lightning is one of those things where you don't trust the analysis. You got to go in the real world and get real amounts of data to get confidence to that analysis.

1:16:03So there's a lot of work both on the capability side to expand the kind of weather we can fly in and on the weather forecasting side. There's a lot of work there. There's a lot of work in the airspace side of just how do you manage this fleet at scale? You were talking about going from one charging point to a delivery, then landing a different charging point. there are some things there that are just way harder than they sound where you can end up with your fleet kind of like migrating organically in a direction where away from where the supply is and you got it that can't happen. Like a beehive.

1:16:36They're swarming. They're going somewhere else. Oh, no. Exactly. So you got to manage that carefully to make sure you can keep delivering for people. So there's a lot of almost every front of what we do at Zipline. There's a reimagining happening for the scale, for new capabilities. capabilities. We've talked about some of these healthcare use cases. There's a lot of new capabilities we need to build both into the, how we operate the drones and into the tools that the humans use to, on the healthcare side and system side and on the patient side to use the system. So there's all kinds of exciting things ahead of us.

1:17:08I think of us as like, this is more of the very beginning of this journey and there's a really exciting roadmap ahead. I love it, man. I'm excited. I think that I love that there's like one of your teams is like a group of storm chasers. And they're just like, our job is to find lightning in the world and then go fly these drones up into it and see what happens. What do you do for work? That's what I do. I'm a storm chaser for zipline. That'd be a rad job. Yeah, it's a badass team. And they're hardcore, too. Because like chasing storms, when you have basically a big old trailer, we have these docks mounted to a trailer that they use.

1:17:40And obviously, when the storms, it's just hard to get around in these places. And it's really cool how they work and their dedication. Is there a manual override on those things that you could just take over if you needed to or no? Yeah, but it's very high level, right? It's basically you basically have three buttons, a button to tell it to go back and dock, a button to tell it to loiter right where it is, like stay put, and a button to deploy that whole aircraft parachute and land where it is. That third button we've never pushed, but every regulator wants it. And then mostly it's the other two buttons.

1:18:11You can't jump in there and stick fly the thing. It'd be extraordinarily hard to like these aircraft fly themselves in a sort of superhuman way. And so like they're there, you know, how fast they make adjustments, how fast they can self-diagnose and handle faults and still fly nicely. Yeah. If you try to put a human in the loop of that, it would be extraordinarily hard to do. I was actually just watching a video from Mark Rober and it was a goalie. It was an AI goalie, essentially. and I forget the soccer player's name because I'm not a soccer fan to that level where I know the person's name, but I imagine it's probably like Messier, like somebody who's super famous.

1:18:49And they were going against the goalie and they could not beat the goalie because the reaction time was just so uncanny, like how fast they could react. And I take that kind of example, which is sort of a fun experiment, which Mark Rober is a scientist and he's super serious. But at the same time, it's meant to be fun and not really an experiment of seriousness on a large scale. And how what to what level your advancements may be well beyond that to to truly say what you say, because I'm sure there's a lot of people say, you know, can I hear you saying that? But, you know, I still don't believe it, you know, because I can't believe it until they see the truth of the superhuman ability to fly these aircrafts.

1:19:30you know i just uh think back to the maybe the mark rober thing that soccer goalie uh that was ai totally demolished the the human could not beat the the ai goalie yeah it it is surreal to see these things handle the when i saw the first um the i mean all this fault tolerance we test extensively right and you know the fault like so when i face a whole tolerance like one rotor fails and the system handles it and it goes back and docks. When that is working well, when you're watching it happen, if you don't know it happened, you won't even notice it, right? Like, you'll have people, like, with binoculars, like, watching that thing, because the only way you're going to see it is if you actually see the propeller blade, that one propeller stops spinning.

1:20:12But the aircraft, like, handles it so seamlessly, it barely moves. And if it's windy at all, like, it doesn't move more than it does normally in the wind. And I think there's there's a this is one of the exciting things about autonomous systems is they can they can get super robust, which means we can we can operate and serve customers and deliver to patients and hospitals in all kinds of really crazy conditions. Again, the conditions that, you know, flying low altitude in these storms, nobody does it like and I say nobody. I didn't even realize this when we started Zipline. I thought like at least medevac helicopters and this kind of things, Coast Guard must do it.

1:20:46And we ended up calling some of them because we were struggling to like figure this out. And we're like, cool, do you have any data? And they were just like, oh, no. We stopped flying those conditions in the 70s. Way too many people died. And, yeah, you can only do it if you use an autonomous system. And, yeah, it's quite wild to see these systems perform because they're just, like, they're so fast and so superhuman. And, yeah, and then they can get these deliveries through these conditions that, like, if you're on the road, you're – some of these conditions are so crazy. If you're on a good road, you're kind of feeling like, I should not be on this road.

1:21:20but the drones do all right yeah i used to listen to the bill burr podcast quite a bit don't anymore but he is a hobbyist but a trained helicopter pilot and so he would chronicle some of his you know it's just him talking for an hour so he needs a funny person to listen to talk so it's enjoyable but he would chronicle some of his efforts at just flying a helicopter and i knew it was like i mentally understood that it was difficult but like hearing somebody try to learn it and like the maneuvers they're going through just to get down the basics of flying it's so fraught and dangerous to fly helicopters that in adverse conditions i mean it's a death knell and we know that we've had you know famous people like kobe bryant dying helicopter crashes just because of fog or because of a bad systems and humans are struggle and have to have extreme training to be able to fly those things and computers can fly them so much better now that It's like, why even put a human behind the sticks, so to speak?

1:22:18Yeah, absolutely. And I think there's, yeah, I mean, what our system can do literally in the blink of a human eye is like basically handle, detect, decide, handle almost everything that can go wrong faster than a human blinks. And that's, you know, that's powerful for making autonomy scalable and safe. One last question for you, Keaton. What's your tech stack? So if you were to describe, I know you have lots of different techs. So like if you were to pull out all the tools in your CTO, you know, tool belt, programming languages, frameworks, you know, what's Zipline built on? Right. Yeah. I mean, for high performance stuff in the cloud, we like go for almost everything on the aircraft is moving to Rust.

1:22:59You know, there's various parts of the tech stack, right, for almost every part of the tech stack kind of has its own little world. a lot of we've built our own in-house simulation system we call Phoenix that was how we do a lot of this basically a lot of the simulation and testing yeah you know we the yeah boy the tech stack is broad you know I worked a lot on Ross the open source robotics platform back in the day and everybody always asked me like cool you must run on Ross and I'm like oh man no we don't if ROS2 had come along early enough, we would have. But unfortunately, we kind of went down that path before ROS2 was ready for that.

1:23:39Yeah, we use open source where we can and of course contribute to open source where we can. I'm a big proponent of open source. And one of the things we actually, on the TechStack Fund, we spent a lot of time on is basically the future of autonomous flight. Even today in Dallas, we share the airspace with some of these other drone operators we mentioned. And I think of that, no one really knows how autonomous flights going to scale in the airspace. And I can tell you what we do today in the airspace is not going to scale. It barely works for the scale we have today, which is very low compared to where the future is going.

1:24:12And there's a lot of need for folks in industry and academia to bring the best ideas to the table, make sure the best ideas win, so that we are both scalable in the airspace and safe in the airspace as we scale. And so ZipLine's working on a bunch of projects like that that we have or intend to make open to the community so we can kind of prove it internally as like, hey, this is how our drones don't collide at scale with each other and make those technologies available so everybody can share that. That's a big part of the tech stack that we think a lot about because, well, again, I just love, to me, the open source mindset comes from this thing of like, hey, we're all in this together.

1:24:56and as I think about the future of the airspace, we're all in this together. It's going to be great for Zipline if all companies that need to use the airspace and all users of the airspace can use the airspace safely. But the future of autonomous flight is still to be figured out and we're spending a lot of time on that. Do you see a world where there's some sort of open orchestration communication channel where you're not just detecting objects and avoiding them, but maybe you're actually publishing your future path. And so other systems can know like there's zips going this way and that. And so I can avoid those paths.

1:25:28Yeah, absolutely. So today there's a version of that that we use in Dallas and it's not live. It's done before you take off. So basically there's a, there's a federated cloud-based system where are the different operators talk, systems talk to each other. And basically it's very rudimentary today. Kind of think of it a little bit like a game of battleship of like, okay, I'm looking, I'm trying to share my intended path in the air, sort of a tube in time. And I share it with the others. And they're kind of like, nope, that collides with something I've already reserved. And then you try to find another tube.

1:26:02And then there's a, it finds a tube and then you're expected to stay in that tube about 95 % of the time. Not very scalable. It's not a very efficient way to use the airspace, which is plentiful. um we internally we we do we share our aircraft share uh information live with each other over a point-to-point radio link where they share their intent of where i'm going in the next 10 seconds or so and they do that also over the ip network uh i can't announce the partner yet but we have there's a commercial drone operator in dallas we're going to be announcing a partnership later this year where we'll be they'll be using that same protocol to so we can share the airspace without that those big reservations and that's the net to kind of explore hey is this more scalable Can we build up the data and then share with the rest of the regulators and industry to kind of bring everybody along and make the case for what we think is a better idea than what we have today?

1:26:51And then we'll move to that idea. So, yeah, I think what you're describing is going to be the beginning of where we go with this. I think there's going to end up being a fair bit of structure, a lot of real-time sharing to make sure we use the airspace really efficiently and scalably. It's like DNS in the cloud, basically, in the sky, literally the clouds. not the internet cloud as we know it but it's kind of like that it's like a phone system in a way or a route system or a flight path you now have to have predefined or known things and it's like well we have you know these coordinates and we know where we're at in the world so maybe they're logistically you know like that or something like that but yeah but constantly changing yeah exactly yeah we spend a lot of time talking to the original folks who created ican which created the dns system so like you know how did that actually come to be because in some ways it's kind of a miracle right how open the internet is and how undoubtedly it evolves and like it's like all right cool we want to learn all the lessons from that because you know we really want the future of the airspace to be similarly democratized and open and innovative and evolving right so you're studying dns to learn how to do things in the in the sky absolutely yep yeah and not just the technical layer of dns but also like if you're going to i can the international is a corporation of names and numbers, I think.

1:28:06Anyway, it's basically the people behind sort of the Internet that, like, agree internationally of how it's going to work. Just because, like, they figured out how to be a group of people that doesn't make bad decisions, you know. Right. And we want to learn from that as well because I can see if you don't get the right incentives and the right people and the right mindset, the right kind of structure, I think it's really clear how this kind of gets stuck in gridlock and doesn't evolve and innovate. Right. So when you operate on a 2D plane, everybody gets one place. So like my house address leads to one place.

1:28:44Now there's many paths to that place so you can alternate routes and that's all well and good. But on a 3D plane, if you have conflicting paths, how feasible is it to just change the Z-axis and say, well, we're going to fly five feet lower than we were going to and now we're just completely have free airspace. Is that like foolish thought? Is that legit thought? That's exactly how our drones don't hit each other. Okay. So they're just like, well, I'll go up or down a little bit. Exactly. They base this distributed algorithm. They share their 10 seconds out intent with each other. And there's a distributed algorithm based on what they hear from the people around them, that they'll adjust their course in a way where you get an emergent behavior you're talking about of like, okay, great.

1:29:22We're slow, you know, you end up with that nice passage very efficiently and dynamically. Yeah, I think you're describing exactly how this needs to happen. but the challenge is it has to happen in a very robust way, right? It can't be like, oh, if you can reach a server on the cloud and your cell connection is working and the server is working and everybody's talking at the same time and there's some central orchestration that doesn't have a problem, then it works. We have a lot of experience to learn the hard way that you just can't scale that way. It's got to be simpler, and the aircraft themselves have to have many layers that they can do without any radio connections to things that will still give you layers of safety.

1:30:02That's really important. When you bring up layers, it makes me think, is there an advantage in formalizing around some kind of layering, vertical layering of lanes, so to speak? So you're not just having the emergent behavior of height, but you're actually pre-planning, like, well, I fly at this altitude and this other company always flies at that, and so we're pretty much cool. I think that'll happen, and I think it'll be more than vertical. I think it'll be more like sort of highways or lanes. And the reason in a lot of these, there's a lot of things you do to optimize for safety. But one thing you'll do is you'll avoid certain areas.

1:30:37Right. So like, yeah, if there's a class examples, like a stadium, right? You're going to, if there's a game happening, you're just not going to fly over that stadium. Cause why would you like, it's not that you're going to have a parachute landing there, but like you might statistically speaking and flying around it, statistically speaking is great advertising. Like zip line. Yeah. Yeah, we'll make deliveries there. What is that thing? Not a crash. That's a parachute. And so everybody in the air is avoiding similar things. And so you'll end up with this congestion kind of going around these keep-outs or these soft keep-outs, as we call them, and sometimes hard keep-outs.

1:31:10So there's certain government places where the government is like, you cannot fly over this area. We consider that to be something you can't fly over. So these hard keep outs and soft keep outs end up funneling their traffic into these places where I think you'll have, you might, in the early days, vertical separation might be enough. But in the long term, it may be like, yeah, if you're going this way, stay to the right and up high. And if you're going the other way, stay down low and to the left. And you end up with this kind of literal kind of corridors through that create more and more efficiency, especially when you have these kind of congestion, forced congestion.

1:31:42um right so there's a lot of cool things that are that are going to be needed to really so cool yeah so cool well uh we've taken enough of your time we appreciate you answering all of our questions i so much i think this is one of the coolest things there is so i'm both happy with you and jealous of you that you get to work on zipline all the time because i just feel like there's so many hard problems there's so much benefit on the other side of it and so uh congrats and good luck, I guess, on scaling out, scaling up. It's a fun place to work. Just yesterday, I came to work a bit early. I was on a call in my car, and I was watching everybody walk into the office, and everybody kind of has like a skip in their step.

1:32:20And I was just like, just reminding myself how lucky I am to work in a place that's like that, where people are fun, the culture is fun, and the technical challenges are real, the customer challenges are real. And yeah, and everybody's skipping to work. It's like a good sign that life is good. So I'm very lucky. Yeah. Yeah, great to talk with both of you too. Really enjoyed this conversation. It was awesome.

1:32:45There you have it. Zipline. Pretty rad, right? I can't wait to try it out when they come to Omaha. And I'm stoked that Keenan confirmed Omaha is the company's next destination. Just kidding. He totally didn't confirm that, but he should have. And I hope they do. Thanks again to our partners at Fly.io, to our favorite Beatmaster in the entire verse, Breakmaster Cylinder, and to you for listening. We love it that you stick around all the way to the end. That's all for today. But we'll be back in your ear holes on Changelog and Friends on Friday. Bye, y 'all.

1:33:44Thank you.