a16z Podcast - a16z Podcast: Eyes in the Sky

Episode Date: March 25, 2017

In this episode of the a16z Podcast recorded at our inaugural Summit, Jonathan Downey, CEO of Airware, Grant Jordan, CEO of Skysafe, and Kyle Russell, partner at a16z, discuss our future with “eyes in the sky.” How do you balance experimentation and following the rules in a space where people have fears about what a future with drones might look like? This conversation covers the most interesting enterprise use cases for commercial drones, where we are in the introduction of drones into the consumer and commercial space (including the most interesting enterprise use cases for commercial drones), and how the industry will scale. Downey, Jordan, and Russell parse out what the new FAA regulation means big picture for drones and airspace, and what’s been overlooked. Regulation, says Grant, is just one element. What do we want our drone future to look like — where we want them flying and where we don't — and how will our responses to consumer and commercial drones affect each other? What are the privacy and safety implications, and how do we navigate them?

Welcome to the A16Z podcast. In this episode, recorded as part of the A16Z summit, Jonathan Downey from Airwear and Grant Jordan from Skysafe, discuss with A16Z partner Kyle Russell, all things drones. They take a close look at the FAA regulation for drones released this past summer, where we are now, and where we're going in the drone market. Thanks for joining today. So drones. At A16C and Z, we're incredibly excited to see today, cameras, tomorrow, flying computers. to see them take humans out of dangerous jobs like climbing towers or oil rigs. But even with all the excitement about what they could do, there's this kind of dark cloud hanging over the ecosystem regarding, you know, what is government going to do about

flying computers over cities or over workplaces? And so Jonathan, just for those of us who haven't been following that story, what was the regulation introduced this summer? What were the kind of key takeaways that people should know about regarding like how that's going to shape the ecosystem. Yes, this was definitely one of the biggest concerns when we went out and raised our first round of financing. There was a big question of, will this ever even be legal to do? And finally, as of the summer, it's called Part 107. It's essentially the rules of the road for using drones for commercial applications. And what it says is that commercial companies can do this. They can do it up to 400, 500 feet. They can do it when there is an operator who's responsible for the vehicle, who can see the

vehicle, although they don't have to actually be flying it. And in fact, in most cases, the drones are usually flying pretty autonomously. But that's kind of some of the basics of it. So there's also some limitations. The converse of that, it means you cannot fly aircraft beyond your visual line of sight, although there is a waiver process now where the FAA is going to grant kind of case-by-case permissions to companies that are going to allow them to do that in the future. You've been doing this for five years now. I imagine that the conversation has kind of evolved once the groundwork was starting to be laid by regulators. How have the conversations changed with the people you were talking about, you know, whether they knew what they wanted to do with drones or maybe

if they were just kind of speaking to you in an exploratory way? How have those conversations changed? So I've actually been in the drone space for about 12 years going back to 2004. And I remember at 2004, it was, this was unquestionably only a military thing. And around six or seven years ago started to see a lot of large enterprises go in many cases to military companies and say, hey, is there a version of this, can we use this for power line inspection and monitoring vegetation encroachment and monitoring, you know, oil and gas and infrastructure applications and things like that? And so the conversation has definitely changed. And I think six years ago, when some enterprises were looking for this, there were a lot of missing pieces. There was a, you know, missing

commercial version of the technology. A lot of the software was missing to actually make the images in the video taken by drones, useful business intelligence. And of course, there was the missing regulatory framework. So a lot of those things now exist and have been accelerated by the introduction also of a consumer market, which has further driven down the price of the technology, gotten into smaller and smaller form factors, and made it really ubiquitous. And so now what Silicon Valley is best at, which is building a lot of software that makes this really useful for a variety of different applications is making this available to just about every industry. And so it's not obvious for every business how drones might fit into their workflows or transform their

workflows. And given that, Grant, I'm curious to ask you, what are things that businesses should be considering as unmanned aerial systems become more pervasive? Because we're kind of on the security and drone protection side, we talk to a lot of customers that their biggest focus right now is on thinking about protecting themselves from other people's drones. But I think along with that conversation about, you know, airspace enforcement and what rules are and how the regulations are changing. Part of that is also a conversation of how can they actually use drones themselves in the future, right? So prisons can use drones for all sorts of things, like inspecting, you know, they have really large physical perimeters. There can be all

sorts of inspection, kind of constant, you know, perimeter security applications. Stadiums would love to use drones in the future to do their filming and to do all kinds of other stuff. But their big concern is, you know, they don't want to just roll drones into their infrastructure if they can't control what drones are there and what drones aren't. I think we've all probably seen viral videos or tweets where someone uses a net gun to shoot a drone out of the air or sends a falcon after one. How's the Falcon as a service business going? Or is that not what you do? Sorry. It's not what we do. Unfortunately, it sounds very exciting and awesome. We don't have a falconry. But, no, our focus is more on the communication side, on looking at the drone protocols,

on identifying what drones are flying in an area so that that can lead to actual airspace management of authorizing drones to be in an area and knowing when there are drones that aren't authorized to be there. Instead of just saying, oh, there are drones here. We don't want drones at all. We're all about facilitating actual proper use of drones and making sure that they can control those spaces. So, Jonathan, back to you,

back when Airwheres started, you had more of a horizontal approach. And you since switched to some more specific verticals with big problems to tackle. Could you want me through kind of the evolution of that kind of set of approaches? you know, why initially go horizontal if we kind of didn't know what to do with drones? So back in 2010 and 2011 was when we started to hear a lot of large enterprises asking about drone technology and trying to figure out how could they incorporate this into their

business. And we didn't have great visibility into where was this going to be adopted earlier and where was it going to be adopted later. But we had a sense of, for any of these different applications, there's kind of a lot of the fundamental building blocks that are just missing from the ecosystem entirely. And so we really focused on building out a lot of that. So think about, you know, software for operators in the field so that you don't have to be a pilot or an engineer to operate the drones. And back in 2011, everyone who was doing any software for drones

assumed the operator was some military pilot or some incredibly experienced, you know, software and electrical engineer. And so we hire a bunch of game developers and build a UI that's just really easy to use. We knew that that was going to be required. Another thing was our cloud back end and kind of a, web UI for managing all of the aspects of collecting data across multiple drone operations, either at individual site over a time history or over a lot of different geographic locations. And we knew whatever the use case was, you needed to be able to manage the data,

you needed to be able to manage who was flying, what drones at what locations were, and you needed to make it really easy to use for the operator in the field. So we started to really get kind of pulled into the verticals where there was the most urgency see from large enterprises where the enterprises had a real felt business problem and where drones could be a big part of that solution. And I was actually surprised, you know, one of the verticals is the insurance vertical. We're working with some of the largest carriers in the United States to transform the residential claims inspection process and also the commercial underwriting of buildings. So that might not be obvious when I hear insurance and drones, what that actually

means. Is there going to be a drone following me, making sure I'm safe as I drive? Or how is that actually going to be applied in their model, their businesses? So today, if you have wind or hail damage and you call your insurance company, it means most of the time there's probably someone coming out to your house and they're climbing up on your roof. If you have a one-story house, it's probably with a ladder. If you have a two-story house or with steep roof tufts, it's often with ropes and harnesses. It can take as much as half a day to do the inspection. In many cases, the roof actually, actually can get damaged as a result of the inspection. So it's really a process that I think in five years we're going to look back on and think it was totally archaic that somebody would

climb up on your rooftop because you have hail damage when a lot of this damage can clearly be seen from an aerial perspective. And instead of filling out a paper report describing the damage, all of this can be collected digitally, stored in the cloud, is available to be viewed over time and as much more accurate. I'm going to ask maybe an unorthodox question. because, you know, obviously it's important to have certain groundwork laid, certain guidelines about how things should work, but, you know, we want to have freedom to experiment. And so maybe counter to that tendency, I'm going to ask, what in the existing regulation or, you know, the new rules that kicked in the summer do you feel is missing? What do you think isn't addressed? I mean, I think there's still a lot of things that we haven't quite worked out, not just kind of from a straight regulatory we need to make laws about it, but I don't think we've quite had the conversation about what we want that drone future to look like. where we want drones flying, where we don't want drones flying, you know, right now it's been kind of an ad hoc thing because it's worked, because for the most part, we don't have drones

everywhere. You know, we don't have 10,000 drones over our heads in the sky. But as we move to the point where it becomes commonplace for, you know, every house to be inspected by drones, for drone delivery to happen, medical supplies, all that stuff, we need to actually kind of sit down and think through, like, what are the privacy implications, what are the safety implications, and how do we navigate that? I think there's a real danger that we kind of move ahead, like, here's the current area, we can just kind of fly wherever, and we end up having kind of major backlash because we have major public incidents that happen

and the public kind of sours on the idea. It's really easy for us to say, well, we have all these great applications, but if we don't think through the privacy, the noise, the whatever, I think there's a good chance that we could kind of set ourselves back. To follow up on that, you know, it's one thing when you're deploying drones as a business

and you don't want to annoy customers. You don't want to cause an incidences because it would lead to bad press. At the same time, drones are also in the hands of consumers who are going to push rules. They're going to use it for just getting footage of a wedding or, you know, random events that they just thought I wanted a camera with a unique perspective. Yeah. And so given that tension, you know, how do you think about kind of regulation versus like normatively deciding, you know, where we should land on things like... Well, I think some of that is kind of tough partly because, you know, and from like a kind of safety and accountability perspective, I have very few concerns about the commercial drone space, right?

Like they're the ones who have all the incentives to be a good actor to, you know, think about these issues and not try to upset people. But I think one of the dangers right now is the consumer drone market and the major companies that are pumping out all these consumer drones. They're doing so in a way where they currently just don't want any regulations. They restrict their ability to sell drones. You know, they want as little friction as possible. and that's good, but the danger is that we have bad events that happen before that regulation

quite worked out before we figure out what we want the restrictions to be, and just the whole market kind of sours on the idea of just having drones and bad actors hurt the potential for that commercial use. Another kind of tendency that I've noticed quite a bit is wanting to connect new trends emerging to historical analogs, looking at desktop computing, for instance, or even mobile we see that things tend to start with something built for work. It's made to accomplish specific sets of tasks. And then eventually something emerges typically maybe in the consumer space

or just lower end where it's a little bit more approachable but not as powerful for getting work done. But eventually there tends to be kind of a convergence. Where do you feel we are in that process? Or is drones maybe kind of separate from that? Different things are going on. I think what you said is often true for hardware and for the devices where if you're using a phone at your office now,

it's probably the consumer phone that you're using with your family and stuff as well, but maybe with some different software loaded on it or accessible to you. When it comes to the software, there's usually companies that are really successful targeting the consumer market or the kind of pro-sumer market who are much more focused on volume transactions. And then there's the companies who are focused on the enterprises and putting all of the tools in place that enterprises need to be successful, including the ability to manage users, enterprise security, permissions, rules, and approval

workflows and things like that. But the hardware in a lot of cases can be the same. So in the drone space, what we're seeing is some of the companies who made the most kind of ubiquitous consumer drones, which in the beginning weren't that great and crashed a lot and people got them under their Christmas tree and they probably put them aside after a few weeks or crashed it into a tree, now in years kind of two and three have just gotten better and better at a really rapid pace. And those companies have started releasing models that are, you know, more geared towards pursuer and even enterprise use cases, but are lacking the software that enterprises need to make them successful. And so now you have this kind of stack of different software

companies, some of them focused on consumers and an app that can help you plan a trajectory that takes an incredible, you know, video of your friend, you know, as their skateboarding. And and you have companies who are focused on kind of the enterprise data collection and business analytics of it and how you integrate that data back into, you know, all the other things that your company is doing. The number of companies deploying drones for highly valuable use cases where they say, you know, the DGI drones, even at the highest end, just don't quite cut it. So where what is the set of, I guess, capability cutoffs where you leap from maybe spending

a couple of thousands of dollars per vehicle to moving up to something that's tens of thousands of dollars? What are the requirements where you kind of need to make that lead today? So actually, I would say the DGI vehicles are great for all kinds of applications. And three years ago, we were tracking about 700 manufacturers of commercial drones around the world. And most of you will have never heard of most any of those seven companies because they've all been passed now by the company who made the best consumer drone, who's now making some of the best drones, regardless of use case consumer or enterprise. and the only differentiating factor really is what software you're using along with the drone

and whether you're using the consumer software that in many cases provided by the drone manufacturer themselves or whether you're using the enterprise software that's provided by one of the companies focused on enterprise use cases. Got it. So when you think about, you know, there's companies now, like if we talk about cars, we don't really know is it going to be one or two companies selling autonomy capabilities to car makers or it's going to be five or six. We kind of just don't know how difficult it is. When you think about autonomy, as relates to aerial vehicles,

you already mentioned that they're increasingly autonomous, but we still have that line of sight requirement baked into the new regulations that came into effect. What use cases are unlocked as we maybe get rid of that requirement? And what's it going to take to make the federal government kind of feel comfortable with that transition? I assume that you're having more conversations with the FAA than most startups in the space.

even. Yeah, we've been one of the first companies to really lean in and work really closely with the FAA. When it comes to autonomy, I think it's important to understand autonomy for what, for what purpose? Autonomy, just to be autonomous, isn't necessarily inherently good. We think about autonomy in terms of automation and taking something that maybe you could do it manually. But if you go and you do it manually, it's often not repeatable. It may not be accurate enough. And it may just consume too much human capital and other capital to be able to do it repeatedly and all the time. And so we think about automation throughout the entire workflow, automation in how you transform the photos and the video into actionable business intelligence, but also, of course, autonomy for the

vehicles in the field. And whether you need that person there to operate it, whether the person is there just in case something goes wrong or whether the person is there, not at all. And so there's, you know, our main focus, and I think the industry's early focus is on a lot of the applications where the person has to show up no matter what to get a job done. Like in the case of the insurance claims inspections, the person's there to shake the hand of the customer. They're often doing an internal inspection of the house, but we're taking one of the most difficult aspects of their job

and we're automating it substantially by bringing in a new tool into their toolkit. There's other inspections and jobs that are similar to that. I think over time, as autonomy is enabled and as the regulations allow for flight operations that don't have a person there at all, we're going to see a lot of applications enabled where, you know, the person doesn't need to go out into the field at all. One of the other industries is the mining industry where right now people are flying drones weekly, in some cases, even daily, but you need that person there who's operating them.

I think that's a great application or market where as soon as the regulations enable it and the technology enables it, they'd prefer to not have a person operating the drone at all, and there's just the drones regularly flying day after day on their own and automatically. The other piece that that beyond line of sight and greater autonomy gives you over time is scalability of all these things. It means that you eventually break the one-to-one ratio of one pilot operating one drone. That's the thing that the military started thinking about a long time ago, you know, when they started thinking about, you know, drone operators in the field. And how do you build that autonomy to the point where you can have one operator operating essentially five drones or 10 drones? And that operator is really only called into service when there's something weird going on. or there's something that requires, you know, manual operation. But I think that's where, like, the real advantages start to come in.

You came to the drone space from working in the Air Force, correct? Yeah, yeah. I was an Air Force officer doing small UAV systems. Got it. And so in the course of your career, what from that side of things has kind of made its way to the consumer space? Well, honestly, a lot of it. I would almost say all of it.

A lot of the things that the military was thinking about back in kind of the early 2000s, about autonomy and, about building out mission plans and running these operations and things, that's all trickled down to the consumer side, you know, thinking about, I mean, even, you know, you buy an off-the-shelf drone and you can do what, you know, in 2005 was really advanced for military small UAVs of like planning out a mission, hitting go, it takes off, flies a thing, comes back. I think that's amazing. And even the sensor systems, you know, have trickled down drastically, like the fact that, you know, DJI sells a system with a, you know, high-end thermal

imager on it is huge. Jonathan mentioned that, for instance, that he had to hire video game designers in order to make a UI that was actually understandable and usable. So what, on the consumer side, you know, with billions of dollars of capital going into drones for companies like DGI and unique and 3D robotics, our portfolio company, Skydeo, what's making the opposite, kind of taking the opposite path? Where are things that initially they were, you know, done for the consumer space, are making the way over to the military? Oh, that's a good question. I think a lot of that thinking about ease of use and thinking about, you know, breaking some of those traditional assumptions on the military side of like, oh, we have a thing flying in

the air. There has to be a pilot. Like, that person has to have been traditional aviation. They know how, you know, piloting and aircraft works. Like, that's starting to break down on the DOD side. As, you know, the consumer in the commercial space proves out, like, no, you know, autonomy helps you, you know, handle that actual flight. A few years ago, if you had a drone, you also usually had this giant box that you wheeled around with you called a ground control station and it had joysticks and it had this screen and all of that is being replaced for most drone operations by an iOS or an Android device with software that you touch with your hand and that consumerization of this technology is also a big part of what's actually going to make it really scalable

because the military's approach to a lot of these things is not scalable. Right. Nature of way they establish contracts and such. Thank you both for your time. I really appreciate you catching us all up with where things are today and where they're going. Thanks.