Balancing Product & GTM | Stanley Tang (DoorDash) & Niilo Säämänen (Wolt)

All right. Hey everyone. Thanks for having us. Super excited to be here. So, my name is Stanley. I'm the co-founder at DoorDash. I lead DoorDash Labs, our robotics and autonomous group, and I'm joined by my counterpart Neilo over at Wolt.

I think both companies have very similar founding stories. The way we operate our company culture is very similar. So I thought for today it would be good to just have a conversation about how the two companies think about product and innovation, and specifically the topic we wanted to go over is this theme of innovation at scale. As companies get bigger, as we get more global, how do we think about innovation? And then I think the case study we really wanted to use was about autonomous and robotics, which is a big bet that the companies have started to make over the past few years. Just for some context, we started this group called DoorDash Labs a little over seven years ago, where we're looking at bringing autonomous and robotics deliveries into DoorDash and Wolt. We do it primarily through partnerships, so we work with a bunch of delivery robot companies, drone delivery companies, but we've also built a lot of this technology ourselves in-house. Actually, a couple weeks ago we announced our in-house delivery robot. It's called DoorDash Dot, and you can see a picture here on our slide. It's actually live doing deliveries autonomously in Phoenix now. We'll talk a little bit more about DoorDash Dot later. We'll have a little sneak peek at the end, but I think it would just be good to have a conversation on how does DoorDash and Wolt think about innovation? How do we set up these innovation groups, these moonshot labs? I would love to just hear your perspective and share the DoorDash perspective as well.

Yeah, so when we first started DoorDash back in 2013, we certainly weren't thinking about robots or things like that. I think it really started out as an experiment. DoorDash actually wasn't ever meant to be a company. It was literally a college project. It was a campus food delivery service at Stanford. It was me and my two co-founders. We were all college students doing deliveries ourselves. There was no technology, no app, no automation or anything. It was just us picking up the phone, doing the deliveries ourselves. The original name wasn't even called DoorDash. It was called Paltodely.com. That was the original name. Palto is a city where we went to school in California. That's really how DoorDash got started. I think that's going to be a common theme you'll hear over and again: the idea of innovation and experimentation are tied very closely together. We weren't really thinking about robotics and autonomous delivery. It wasn't until several years in, 2017, when I get this cold email in my inbox. It was a startup founder from Estonia. He told me he was in San Francisco visiting and he had some delivery robots to show me. That was the first time I ever heard about this idea of robots even being a thing for delivery. I met up with him and he showed me these amazing delivery robots. The company was called Starship. That was the first time where it gave me this window or inspiration into this world of actually one day deliveries could be automated, done by autonomous robots, done by drones. This needs to be something we should look into. I think Neilo, you guys had a similar story as well on the Wolt side.

Yeah. So after meeting Starship, we decided this is something I want to look into. We started this group called DoorDash Labs. In the beginning, the main thing was really just trying to understand what the world of autonomous delivery even looks like, what are we even trying to build here. I think that's the most important thing I learned from starting one of these moonshot bets or any new businesses or new products within an existing company: it's super important to understand what's the use case you're solving for. In the beginning, it's very easy to get distracted by 'oh this is a shiny toy' or 'this is a cool technology trend, let's just use this for the sake of it because it's a cool tech thing.' I see this in AI right now. Everyone's like 'oh let's just use AI,' but what does that even mean? So in the beginning, it was really a lot of just figuring out the use case, how do we actually use autonomous deliveries. Start with the problem, the use case, and then work your way backwards to figure out what the right solutions are. In our case, I'll show you some slides from the early days of DoorDash Labs and explain what's going on here. In the beginning, it was really just trying to understand what is the use case. It turns out delivery is actually pretty nuanced. There are no two deliveries that are the same. We've done over 10 billion deliveries. We have short distance delivery, long distance, urban, suburbs. What we realized was there's not a one-size-fits-all solution. You have the sidewalk robot, the Starships of the world, which is really good for deliveries that are less than a mile. We started experimenting with drone deliveries as well, which we found really good for rural areas where road infrastructure isn't as good. But then there was this big gap where a lot of DoorDash deliveries fall under. About 50% of deliveries fall under what I call the dense suburbs. Think of cities in the US like Phoenix, Arizona, Dallas, Texas, where it's much more spread out. Our average delivery is typically between 3 to 6 miles, yet the delivery still needed to be completed in under 15-20 minutes. There weren't any existing solutions for those. When we actually took a deeper look at understanding the use case, the problem for DoorDash, we worked our way backwards from first principles and asked ourselves: what's the ideal autonomous delivery solution that suits DoorDash? There wasn't anything out there, and that's when we decided maybe we need to invest in this ourselves. Only after that we decided this is the time to start our moonshot bet around autonomous deliveries. We went through a whole process of building our own delivery robot, which we'll show at the end, called DoorDash Dot. Here are some early days pictures. The cool thing about this is that the robot on the bottom left side of this slide pretty much was when we joined forces in 2022. We actually flew to SF and met Stan for the first time. One of the coolest things we wanted to see was what was happening on the DoorDash Labs side, because we've been dreaming of doing that on the Wolt side but we were so focused on expansion and had a lot less funding to focus on this. One of the coolest things of us getting together, outside of this kind of really having a similar culture of customer obsession and operational excellence, was the cool stuff we could build. I love cool robots. When I first saw Dot was pretty much in 2022, roughly in this shape at the labs, it was a really cool thing to see. Also, when you think of innovation at scale, what companies tend to often do is just kind of love them to death with your current organization. But when I came to your labs, it was a pretty scrappy operation, a handful of engineers in a warehouse, very startup style, just figuring out how to do something meaningful for your customers rather than trying to do cool tech for tech's sake. It was one of those really beautiful moments to see. We have yet to test this in Finnish in the icy weathers outside, so let's see how well that works. But looking forward to getting to that point.

Yeah, totally. And maybe this is a good segue to talk a little about operating philosophy. How do you actually, let's say you start off wanting to launch a new product, new business line, or a new moonshot bet, or a new innovation lab? How do you actually set that team up? I'll show some more slides. I think one of the mistakes a lot of these companies, especially big companies, make when they start a moonshot lab is they ironically overfund it. It's like 'oh, it's autonomous, it's robotics, it must be really expensive, so you have to give it a ton of capital.' When in reality, you actually want to do the opposite. You actually want to set constraints. Constraints turn out to be one of your biggest advantages because constraints force creativity and innovation. In the beginning, you don't really know what you're doing. So if you just give a team $100 million and say 'go do whatever you want,' guess what? They're going to blow all that money and just do whatever they want because they don't really know what they're trying to do. You've got to set constraints in the beginning. We actually kept the team really scrappy. The first two or three years, we probably funded this team like it was low single-digit millions. It was really just about understanding what the use case is, what is a product, let's do some prototypes first, let's validate the idea. In the beginning, it wasn't 'let's just go build a robot.' It was 'let's go build a prototype, let's build a mockup.' You can see here on the bottom right, initially the first version was literally a foam mockup of what we think this robot is. Then we brought that to the restaurants, to the customers. We put it on the street to see if this is even the right size, the right vehicle profile. Then the next step was okay, we have a phone mockup. Maybe let's build a prototype version of the vehicle. But in the beginning, there was no autonomy. It was a remote-controlled car. You can see on the top right side, it was literally someone on a steering wheel driving. That picture is actually Tony, our co-founder CEO, doing a kind of remote control driving of the robot. Again, you could start testing real deliveries without even building the full autonomy by remote driving. Then it's like okay, let's build a basic V1 autonomy, it goes on fixed routes. Then eventually, once we have certainty that this is the right form factor, we went through a couple of iterations, three or four or five iterations, before we really landed on the actual solution that we need. Again, it's very nuanced. Autonomy for delivery is different than autonomy for rideshare, let's say. You can't just take what Waymo's built, plop it into DoorDash, and then everything magically works because there's a lot of nuances. The vehicle is very different, the profile is very different, the speeds at which you drive are very different, the pickup and drop-off problem is very different. You have to go right up to a restaurant to a customer. So setting these constraints is really important. The team has to earn their way to unlocking budget. It's almost like we're a VC within DoorDash. You've got to get your series seed round, your series A, series B, and then prove things out. We have a mantra at DoorDash: dream big but start small. That's been our philosophy, frankly, with not just autonomy but most of our moonshot bets. They all start very, very small.

Yeah.

Yeah. I think the other thing I want to talk about is also you have to set this culture of shipping. I think that's super important. Again, it's very easy to fall into this trap of just because it's an R&D project, a moonshot bet, it's very easy to get stuck in what I call R&D hell. It's very tempting for engineers when they're working on a cool new technology problem. They just want to work on the cool engineering problems versus actually the end result is you have to ship a product. You have to commercialize. What does this technology mean in the context of real customers using it? It's very easy to optimize for something that doesn't really need to be optimized. Should this be something you want to build in the first place? That's something we emphasize a lot at DoorDash and Wolt. Yes, it's cool when you work on cool interesting new tech. Yes, it's cool, but you also have to realize this is not a research lab. You have to work towards a commercial product. You have to make contact with the real world, and the quicker you can get to that point, the better off your product is.

Yeah. And maybe this is a good segue to talk about what Neilo mentioned: this idea of building systems, not silos or individual features. I think this is something that's really important at DoorDash and Wolt as well when we think about building product. Our business is somewhat unique in the sense that it's a delivery service. It's a network that you're operating at scale. It's not as simple as 'oh you just ship a product or you build a device, you plop it in, and then everything magically works.' There's a lot of nuances. You've got to think about what you're building in the context of an end-to-end system. For example, delivery is more than just the driving portion. You've got to think about the pickup, the drop-off. There are different kinds of deliveries. There are different restaurants you're interfacing with. Even with our robots, we have to think about things like fleet management, dispatching, charging, maintenance, all these things. One of the things I learned from the world of autonomy is that in order to build a successful autonomy business, building a successful autonomy business is more than just building autonomy. It's all the other stuff you have to do on top of just building the actual core technology that you end up having to also build in order to make this product work successfully deployed in the real world. Actually, that's what a lot of the team ended up working on. The majority of the effort wasn't just the core; it's all the stuff around it. We had to build out all this infrastructure, this network. We built out a thing called the Autonomous Delivery Platform, which essentially is like all the APIs and the dispatching and fleet management required to operate a fleet of robots both on DoorDash and on Wolt. By the way, this Autonomous Delivery Platform isn't just for our own DoorDash Dot robots. It has to work for our sidewalk robot partners, our drone delivery partners, etc. So you've got to think about things at an end-to-end system level. I think that's also been the other big unlock. You can't just think about features in little silos.

Yeah. And you mentioned drone deliveries, which reminds me of a really good operational example of why this system approach is really important. We started doing drone deliveries back in 2022. Google Wing is our primary partner. We've done it in Australia, the US, and some in Finland as well. One of the things we realized in the early days of drone deliveries was that weight capacity is actually a big issue. There's a weight constraint to how much these drones can carry. But the problem is we had no weight data in our system, in our menus. It's something we never had to think about before. So all of a sudden, how are we going to get this weight data? What we did was we had to go measure every single drone delivery that came into our system. We had to literally measure and weigh every item. So we gave a physical scale, a weighing device, to each of our restaurants. A drone order comes in, they'll weigh it, we'll record it in the database, and over time built that system out. We also productized the scale. We called it Smart Scale. We built an automation into it so it can connect in order to whatever item you're weighing. As we started weighing these items on the Smart Scale, all of a sudden a light bulb moment hit me. It's like, hold on a sec. These items that you're weighing don't just benefit drone deliveries. You can actually use it for all of our deliveries. Because one of the things we started catching was missing items. By looking at the weight data, we figured out, oh, this thing is overweight or underweight. Therefore, there's probably a missing item here. That's historically been a really hard metric to drive down, missing item defect rates at DoorDash. So we decided, well, you can use these Smart Scales for that. It doesn't just have to be for drone delivery. So we ended up building 20,000 of these Smart Scales and growing. We gave it to all of our restaurants, and overnight our missing item rates got cut down by half, which was truly incredible. Again, this is something we would have never thought of if I was just sitting in a conference room dreaming about what innovative ideas to work on. This is an example of you only come up with something like Smart Scale because you're thinking about making contact with the real world, thinking about the problem at the end-to-end system level, thinking about the use case, understanding the problem at a deep level. That's a really great example of something that came out of DoorDash Labs that didn't really have anything to do with autonomous but has benefited all of DoorDash.

Yeah, totally. I think it's a good time to start wrapping up. But maybe just in summary, for me, when I think about innovation at scale, how a scaled organization innovates, it's almost like a trick question. Innovating at scale is honestly not any different than innovating as a startup, and vice versa. It's not like startups innovate and then once you get big you stop innovating. It's just the way you think about innovation maybe changes, but the core philosophy of thinking from first principles, understanding your use case, understanding the problem at a deep level, setting constraints, maintaining a shipping culture, not a research culture—all these things stay true whether you're a larger company or a startup. It's just maybe the scale at which you're operating changes. Honestly, for me, company building is just one long journey of constantly reinventing yourself. At every stage of growth, you're going to have to reinvent, innovate, come up with new solutions. What got you from 0 to 1 is going to be different from what gets you from 1 to 10, 10 to 100, 100 to 1000. That journey is honestly a never-ending journey. That's been the big insight for me: innovation is a continuous journey. It never ends. I hope that day one mindset never goes away at DoorDash and Wolt.

Yeah. Awesome. Well, with that said, I think we're going to wrap up here. But before we go, we have a little teaser to show you what the DoorDash Dot robot looks like. Again, this is something we announced pretty recently, just a couple weeks ago. It's live doing deliveries in Phoenix already. I believe it's the first autonomous delivery robot that can travel on roads, bike lanes, and sidewalks. It's live in Phoenix, fully autonomous. Let's take a look.

Awesome.

Well, thanks everyone for having us. Have fun the rest of your time at Slush.