Delivering the future in drones with Keller Rinaudo Cliffton of Zipline

The more we saw smartphone scaling, the more we felt like, hey, a lot of the components and supply chains behind these phones are going to enable a new kind of robotics company to be built. And so we were asking ourselves, what would the Apple of robotics look like? Is there an opportunity here to build really the first global robotics company? And so we didn't have much money. My co-founders and I, we were all unemployed, which is a great motivation for entrepreneurship. And we started building these little, we called them smartphone robots. And they literally used an iPhone as the brain of the robot. You would connect an iPhone to the robot and it could run around on the floor and do telepresence. And that was the first thing that we started selling as a company.

Yeah. Yeah. Exactly. We started selling them. We didn't have product-market fit figured out, I wouldn't say. It was half a toy, half a telepresence solution, half a way of teaching kids to program. I think we were circling around this intuition that it was a good time to be building a robotics company, that new things were possible, and also we needed something that we could sell in a month or two because we had no credibility, no investor was going to give us money, and so we needed to be able to build and sell a very simple product very fast if we wanted to keep the company alive. And so that was the very first thing that we ever built.

Tony was someone who super inspired me. When I graduated from college, I didn't know what I wanted to do with my life. School didn't teach you about entrepreneurship. It was never clear to me that that was an option. When I read Delivering Happiness, Tony's book about building Zappos, it totally blew my mind. It was also inspiring because Tony had gone to the same school and had lived in the same dorm that I lived in. He was just 10 years ahead of me. When I was in Las Vegas, I emailed him and said, 'Hey, I'm here. I am deeply inspired by your book and we're building this robotics company and I want to meet.' I think this is one of the things that's special about Tony and about certain circles of technology and entrepreneurs. Someone like Tony was happy to meet me even though I was a nobody. He was so impressed with these robot iPhones, these iPhones mounted on these robotic platforms, instantly when you showed it to him. I don't know if anybody could have been that impressed by what we were building. It wasn't very impressive on its face. I think maybe he believed more in us. We were very nerdy, very hopeful, certainly very naive. So I think he was interested in the bigger vision of trying to build a new kind of robotics company and he was willing to take a bet on three nerds who had no credibility and no money.

That exact same company is what became Zipline. We didn't necessarily wind it down. Over the course of a year, we realized that building a new kind of robotics company meant we had to focus on a really big bold vision. I also realized we did not want to focus on operating in the home. We wanted to focus on a more controlled environment where automation would be a more natural fit. That's the reason we became obsessed with logistics. The crazy thing is that same company that Tony invested in 12 years later has now become Zipline. Yeah. So, all right.

I think there were a couple of aha moments. One is that I met a guy named Mick who had founded a company called Kiva. Probably most people won't know what Kiva Robotics is, but they had just been acquired by Amazon, and they built these amazing orange robots that could run around inside a warehouse, go to a shelf, pick up the shelf, and bring it to a human packer to pack something into a box. If you saw one of those warehouses, you would see all of these autonomous orange robots running all over the place moving things. I remember seeing that thinking, wow, what an incredible company, and somebody is going to build that for outside the warehouse. Whoever builds that, I just remember thinking, is going to be a world-changing company.

Our backgrounds when we started building Zipline were already in automation, robotics, and autonomy. We had this sense that it was probably going to be possible to build an automated instant logistics system for the planet. But there were so many things standing in our way. To be honest, every expert we spoke to told us this idea was stupid, it was never going to work, we were wasting our time.

We had this sense that if you wanted to deliver something very quickly over a distance of anywhere from 10 to 50 miles, you wanted to do that with a vehicle that was 10 times as fast as traditional logistics, half the cost, and zero emissions. Reasoning from first principles, we thought what you really needed was a vehicle that was electric, autonomous, weighed more like 50 pounds rather than 4,000 pounds. If we wanted to go really fast, we knew we probably needed to fly rather than drive.

I had grown up reading all of this great, hopeful, optimistic science fiction. The funny thing is, this is totally normal in most science fiction books. I think I had this sense that if technology had advanced to a point where we thought it was possible to build this product, we were pretty confident that it was something that people would want.

It's totally fair. When we started building this, our backgrounds were not in healthcare, they were not in aerospace, and we didn't even know that much about logistics. A lot of investors or experts were saying, 'What do you guys know about this space? You're totally clueless.' The reality is, the one thing I had experienced is that after graduating from school, I had been rock climbing full-time for nine months and got to travel to a lot of different countries. I had this sense that the logistics we take for granted in the United States is not broadly available globally. I think that was an important insight. Logistics does a good job of serving the golden billion people on Earth, but for the seven billion people who don't fall into that category, your access either sucks or is non-existent. We've been pretending for a hundred years like this problem is somehow excusable or unavoidable. We felt like it was not either of those things.

In 2013, we were already thinking, if we're going to build a new kind of logistics system, we need to find a public healthcare system we can integrate with. That meant we needed to be looking outside the US. We were focused on countries where this technology would have an outsized impact in healthcare access. I was in Tanzania and I met a researcher who had built this really cool early warning alert system using cell phones. He had several hundred primary care facilities and hospitals using the system. Each had a cell phone, and they would text into a database when they had a patient having a medical emergency needing something specific to save their lives. He had this real-time database of all the medical emergencies across these 300 facilities. I remember seeing that database and realizing it was basically a database of death. Half the system had been built, but they were receiving all these emergency requests and it was like 'okay, thanks for letting us know, good luck.' There was no other half of the system that could deliver what was needed quickly. These network solutions had advanced to the point where we have the information we need, we know there is a medical emergency and exactly the product needed to save that person's life, but logistics has not advanced enough to actually solve the problem.

Exactly. That kind of infrastructure is often referred to as leapfrogging infrastructure. The first people who said they were going to build cell phone networks in Africa were widely called idiots. People thought it was totally unrealistic because landlines hadn't succeeded in taking off. Now, a lot of the countries we operate in have better cell service than the United States does. You've seen this incredible leapfrogging technology where you get the technology and use case right, and it can actually grow in these developing economies faster than anybody appreciated.

When we initially had this idea, we started talking to a lot of different customers to try to figure out what exactly the problem was and what the specifications of the thing we were going to build needed to be. We started to understand that if you could deliver something really quickly over a range of say 80 to 100 kilometers, 50 plus miles, that could have a massive impact. We would be able to serve hundreds of hospitals and health facilities from a single distribution center. You could keep more medicine centralized, send things just when it was needed, and it would save a lot of lives. That's what we were hearing. We started building the first product in 2013. By 2015, we were flying non-stop at our test facility in Half Moon Bay. Around that time, we were about 15 people. I started spending a lot of time talking to governments trying to find our first customer. That was when we ended up having a conversation with the government of Rwanda. I remember in that conversation, we were saying, 'Hey, we have this cool technology. We can do teleportation to any hospital or health facility. We can deliver any medical product instantly.' I remember the Minister of Health looking at us like we were goofballs and said, 'Look, shut up. Just do blood.' She explained that blood transfusions, 50% of them go toward moms with postpartum hemorrhaging, 30% go toward kids. It's a total logistics nightmare because you have packed red blood cells, platelets, plasma, each with different shelf lives and storage requirements, plus all the different blood types. Getting the right thing to the right place at the right time to save someone's life was really hard, and there was a lot of waste. This product was crucial for family health. She ended up saying, 'Just deliver blood to these 21 hospitals and prove that it can be done.' We signed a $200,000 contract with the government of Rwanda to deliver blood to those 21 hospitals. That was how we got started.

A lot of people may find it surprising, but Rwanda has been very forward-thinking in adopting innovative technologies to solve healthcare challenges. They were willing to work with us and take a chance on a new approach. That was key to getting started.

But the reality is Rwanda really modeled itself on Singapore. It is a small, highly efficient, technocratic, fast-moving, innovative country in East Africa, run by a lot of very visionary leaders who believe in technology. They often talk about being a proof of concept country. They want to be the first adopter of new technologies and show how it works, then let other countries follow in their footsteps. They're unique for a lot of reasons. I also have to say, when we were signing that initial contract, when we were launching during that first year, we would be on constant calls with executives and government leaders Saturday, Sunday, early morning, late at night. I just remember thinking, no US government official would be talking to us at this time. I think that by virtue of what that country has been through over the last 30 years, it is extraordinarily entrepreneurial and willing to work hard to move the country forward. All right, so the initially the idea is let's figure out a way to get medical supplies to people in rural parts of Rwanda and eventually we'll expand beyond Rwanda. But really you start with blood supplies, right? Blood has to be refrigerated, kept at a certain temperature, delivered fairly quickly. You could do all of these things in the plane, right?

Okay, but we were wrong. I mean, I joke because we were very naive when we were launching in 2016 in Rwanda. We thought we had a system that was going to be reliable. We thought we had it figured out. We thought we knew how to integrate with a national health care system. We thought we knew how to work with a national airspace regulator. We were basically wrong about all those things. We had no idea what we were doing. Honestly, it reminds me, I saw this flag in a gymnasium recently. It's like a big inspirational flag and the slogan on it is, 'We do this not because it is easy, but because we thought that it would be easy.' When I saw that, I was like, this is a perfect description of Zipline in 2017, full of optimism thinking that we had this figured out. We had signed this contract to deliver just blood, a very narrow part of the healthcare system, to just 21 hospitals. We thought, oh, that's great, we'll just knock that out and then we can expand to other things. But for the first nine months, we only served one hospital because everything that could go wrong did go wrong. We were really struggling to get the system to work reliably. The technology wasn't as reliable as we thought it would be. We didn't have a lot of the integrating software that we needed to really make the system integrate with the overall health system built. So we had to be building that. The team was pulling constant all-nighters working through the weekend. It was a really painful time and it was not clear that the company was going to survive those nine months. We were desperately trying to get it to work reliably for just one hospital. Meanwhile, we were lucky that the government was patient enough to say, 'Hey, we know this is hard. We know you're trying to do something for the first time ever. So figure it out for that one hospital and then you can expand to the other 20.' And how did you work with the hospitals? Because I mean these planes would be launched right from a central facility in Rwanda where the blood deliveries would arrive, and then they'd be launched and they could travel up to what, like round trip up to about 300 km.

Yeah. When we started building the first system, it was lucky that we were talking to a lot of customers because we were considering, oh, we need more complicated versions of the technology. We need a vertical takeoff and landing hybrid that can fly like an airplane but hover like a helicopter. We need different ways of delivering. But the more we spoke to customers, the more we understood, all they cared about was range and price. How do we make sure we can cover the maximum number of hospitals and health facilities by having a lot of range, and how do we make sure the cost of a delivery is as low as possible? Working with health systems in these parts of the world, they don't have a lot of extra money. So if you're going to solve the problem, it has to be really cost effective. That led us to build as simple a system as possible. We built a fixed wing system that really looks like a small airplane. It's a small electric airplane with a wingspan of about 10 ft. We started building distribution centers where we would launch and recover the aircraft, hold inventory, and load that inventory into a zip, which is what we call the vehicles. When we receive an order, launch it, the vehicle flies out. We guarantee a service radius of 100 kilometers around our distribution centers, and the vehicle can fly 300 kilometers on a single battery charge because we have to go there and back.

But from there, each distribution center was capable of serving anywhere from 200 to 400 primary care facilities and hospitals. It really was meant to be that other half of the system we were talking about earlier. We said, 'Hey, if telepresence is working, then teleportation is the other half. If we know that a patient needs something, we should be able to have a doctor or nurse at that hospital press a button on a phone and we can teleport the product needed to save that patient's life in just a few minutes to the GPS coordinates of that phone.

Over the first year, we ended up serving all 21 of those original hospitals that the government had asked us to serve, and we expanded from that to about 350 to 400 hospitals and primary care facilities throughout the country. Then we expanded from just blood to all medical products. Today we've delivered 2 million doses of COVID-19 vaccine, 5 million plus doses of traditional vaccine. We deliver 75% of the national blood supply of Rwanda outside of Kigali fully autonomously. And then we went from about 400 hospitals and health facilities in Rwanda to well over 3,000 today across seven different countries. The system grew a lot. We've now crossed 40 million commercial autonomous miles. It's become the largest commercial autonomous system of any kind, ground or air, on Earth. We definitely never expected it growing in that way. One of the most interesting things we've heard from customers again and again, both healthcare customers and governments, is that home delivery is really the holy grail. So many health systems want to be able to deliver directly to a patient at home because there are a lot of cases where you can keep someone at home. It's much better for the patient and more cost effective for the health system if the patient doesn't have to come into a hospital. A lot of the health systems we started working with in the US, for example, that are betting on Zipline are adopting our next generation technology, which is our home delivery service. We heard from so many customers that ultimately what's needed is teleportation from any hospital, store, or warehouse directly to a home.

Well, we launched in the US in 2021 and the pandemic was really the thing that drove a lot of the urgency. We were seeing a lot of healthcare logistics systems struggle in the US during that time. People weren't showing up to work or humans weren't allowed to show up to work. So a lot of logistics systems that were depending on large numbers of humans weren't working the way we hoped they would. We launched with Nant Healthcare in Charlotte under emergency circumstances. But once we started flying in Charlotte, the FAA started looking through a lot of the data that Zipline had. We had tens of millions of commercial autonomous miles with zero human safety incidents. I think that was really the moment where the US started to realize that this was a key area of economic and technology growth where we didn't want the country to fall fundamentally behind the rest of the world. We've worked closely with the FAA over the last four years. The neat thing is, today Zipline operates three distribution centers in the US, and that's expanding dramatically this year and next year. So the good news is, although the US has been a bit of a fast follower, the US isn't falling far behind. If you live in some of the cities we serve today, homes are already receiving deliveries day in and day out via this autonomous delivery service. I think it reminds me sometimes when people talk about science fiction, they say the future is here, it's just not evenly distributed. A lot of people who hear about this sort of thing say, 'Oh, well, it must not be working because it isn't serving me in San Francisco or in Manhattan.' And the reality is twofold. One is, it didn't start in the US. It started in rural parts of developing economies. And two, now that it's coming to the US, it's starting in places like Bentonville, not San Francisco.

I think like a lot of fundamentally new and disruptive technology, the idea is easy and the execution is hard. Zipline was a much smaller team, not the best funded. These big companies had been investing billions of dollars in drone delivery trying to make it work. We never even necessarily thought of ourselves as the smartest team, but we were definitely by far the most practical. That was in our DNA. We were practical problem solvers, willing to go anywhere in the world where we could scale a service, get regulatory permission, and operate a service that would have a fundamentally positive impact. But the first four years of operating at scale were incredibly painful. The technology was way harder than we realized. A fundamental realization was that we were very obsessed with building a good drone, but it turns out the drone is 15% of the complexity. There is so much auxiliary software and systems that have to be built around it because none of our customers care about drones. All they care about is teleportation. They want something to go from point A to point B fast enough to save someone's life or create a really big economic opportunity. There's a lot of software and systems that have to be built in the background, whether that's communications architecture, data logging, air traffic control software, or even the software we use to run our mini fulfillment centers. Customer ordering interfaces. All of this had to be built out. It's a little easy to look at the problem and discount how challenging it is to build that whole new logistics system from scratch. That's what was really hard about the first four or five years. But luckily, we weren't sitting in an ivory tower. We assumed the first version of the product wasn't going to be very good, and that assumption was correct. The team spent four years desperately iterating, working nights and weekends to improve the product over time. Hopefully that gives other small groups of nerds without a lot of money and resources some hope. The reality is, if you think these big ultra-wealthy technology corporations are going to build the thing, had we been worried about what they were doing, we never would have started Zipline. Money is not the rate-limiting factor. Practical problem solving, unfanciness, and willingness to ship product quickly and learn from customers is actually the rate-limiting factor.

We sort of build on everything we've perfected over the last seven years with our enterprise service, which sometimes we call internally Platform One. The next generation product, we knew we needed to be able to deliver with dinner plate level accuracy and deliver silently to a home. That's really what we think is required to unlock automated delivery for homes across the world.

Yeah. It means if you give us a dinner plate, we can deliver a package onto that dinner plate every single time. That means we can deliver to a front doorstep, to a table on the side of your house, to a driveway. For apartment buildings, we can deliver to a little shared space right out the front door, or even for apartments with roof access, we can deliver onto the roof. We need to be hyper precise. Sound is incredibly important. Anybody who has heard a drone knows it sounds incredibly annoying. We think it's going to be important for the overall delivery experience to be essentially silent for you and your neighbors. We've invested a huge amount of time, effort, and money in designing new kinds of hardware that enable that kind of magical delivery experience.

Yeah. The funny thing is, even with that question, that's already happening in the US in certain towns. To give a little more context, one of the most interesting things we've found is that a lot of the earliest customers we work with are already going on this journey. If you look at Rwanda, we started delivering just blood, then expanded to all medical products to every hospital in the country. The government immediately started saying, 'What about our other national priorities? Agriculture? Can you deliver animal healthcare products? Artificial insemination for cattle?' We did that. Now Zipline delivers a lot of agricultural products. Then it was childhood malnutrition products. Now we do that. Then it was economic growth, e-commerce, building a new national postal service on top of the system. We're in the process of doing that. Similarly in the US, our early customers were big healthcare systems like Intermountain or MultiCare. But as we've grown, folks like Walmart have looked at the technology and said, 'This is going to be really strategic. Having the ability to build a magical portal in the wall of any store where we can pass products through and teleport them directly to a home in a way that is 10 times as fast, half the cost, and zero emission compared to how instant delivery works today is pretty strategic.' We launched with them about a year and a half ago. For the homes we serve day in and day out, the service went from sci-fi to totally boring in about seven days. People really love teleportation and once they have access, they use it a lot. The question of how long it will take to expand from the thousands of homes served today to the hundreds of millions we need to serve, I think that process will probably take about five years.

On my dinner plate.

So we think that we kind of convinced ourselves that we were innovating, while in fact we were using technology that's 100 years old, which is a human driving a car. The reality is, new kinds of technology can solve these problems in ways that are good for the planet, 10 times as fast, less expensive, and a way better customer experience to boot.