Zagat logo


The Food Delivery Robots Are Coming

Zagat Stories presents Restaurants 21/22, a collection of interviews with leading voices in dining, hospitality, food, tech, politics and more. Each story takes the turning of the calendar as an inflection point to consider what happened in 2021, or what’s likely to happen in 2022, in the world of restaurants and food. See all stories here. And feel free to check out last year’s collection as well.

Zach Rash is a 24-year old surfer, environmentalist, robot geek, UCLA graduate, and CEO of Coco, a remotely piloted robot food delivery service that launched in Santa Monica in October 2020. Coco currently has hundreds of robots roaming the streets of Los Angeles, and plans to expand nationwide with an army of new robots in 2022. Next stop: Miami.

Food delivery is a big business, and it’s what got a lot of restaurants through the pandemic. The demand for delivery has been outpacing driver supply, so delivery quality and the experience for customers has gone down. It doesn’t seem like it’s a problem that will ever get better as delivery continues to grow.

With Coco, we want to help businesses grow delivery in a more sustainable way—sustainable for the environment, and sustainable for their business, keeping dollars in their store. We are helping local businesses connect to local customers. That narrative has really been powerful for us, especially during the pandemic when restaurants have struggled to reach their customers.

Right now our robots are fully operated by human drivers employed and trained by the company. Our drivers aren’t contractors. They’re W2 employees, and the job is much more accessible and less taxing than being a delivery driver. You’re working from home, and you don’t need a car—just a computer and internet connection.

Restaurant staff loading up a delivery robot. Photo: Courtesy Coco.

Anyone can do this once they pass pilot academy, which is about 30 hours of training. Once they graduate, we start our drivers with return trips on deliveries, which aren’t as time sensitive. They are simply returning the bot to the closest merchant. The most senior pilots will do locations that are the most complex or that are doing the biggest volume. Head pilots manage other pilots, create shifts and schedules, and help with training.

Our pilot community is awesome. We grow it pretty much off of referrals, with existing pilots referring their friends and family. They have their own channels in our company Slack, and they hang out together virtually. A lot of them play video games. That’s been one of the traits we’ve found that make you a good pilot.

Coco advertises our job listings using many of the same search terms used for other delivery roles. Many of our employees had previously worked in the delivery space as gig workers for car-based delivery platforms, and came to Coco to escape the vagaries of gig work and avoid the need to put additional wear and tear on their own vehicles. They work as pilots, field operations technicians, and across a range of other roles. These are better jobs with greater security, benefits, and opportunity for growth.

We launched Coco in Santa Monica because the city was starting a zero-emission delivery zone program and incentivizing companies to get cars off the road. It was a great fit because they were already thinking about this and putting resources into launching this sort of program. On Main Street and throughout downtown Santa Monica, there’s a lot of good food, and we built the product by putting it in front of merchants on day one. We built a robot very quickly and had some of my friends from college driving them from their computers.

We had some ideas about how we could solve problems with delivery, but we wanted to learn and build it with our customers. Santa Monica was a great place to do that because there were so many merchants willing to adopt a new technology. Alfalfa was one of our first clients, and they helped us refine the process of how staff work with robots. We started working with SBE and their Santa Monica locations of Umami Burger and Krispy Rice, so we worked to standardize the process across a couple of stores.

Coco CEO and cofounder Zach Rash. Photo: Courtesy Coco.

It’s been a pretty crazy ride. A year ago it was just me and Brad Squicciarini, my cofounder. We used to trade off driving the deliveries. And if anything happened, we would play rock, paper, scissors to see who would have to spring out and pick it up and make sure the order got there on time. We still have that same attention to detail on every single order, even though we have a couple hundred employees now.

We’re hiring like crazy right now, and I spend 90% of my time recruiting. We’ve raised $43 million, and that’s allowing us to scale quickly. We’re all across LA, and we’re preparing for the end of the year to start expanding across the country. We’ve been preparing our operations team to make sure the vehicles can handle a diverse set of environments and thrive in Miami conditions and other states with harsher weather. The vehicle is designed and tested to operate in temperatures between -20 and 60 degrees Celsius, the battery pack is waterproof, and the cargo module is sealed. Plus, it’s splash-proof and designed to operate through six inches of standing water.

For most of the restaurants we work with, delivery is a core part of their business. Staff at these restaurants really appreciate being able to manage and organize the way delivery is being done. Delivery can get a little chaotic during the dinner rush. Having an organized line of robots instead of couriers coming in was one of the biggest selling points. I thought that was going to be a downside, but it’s pretty uniformly been the favorite part of the service for restaurant staff. It makes delivery more efficient. Delivery doesn’t always bring your food in the best condition. We typically get it there 30% faster than other services, and it is preserved much better since it’s in the robot the whole time.

Normally when you order food for delivery, the hand-off with the courier is just logistical. It’s not an experience that anyone loves. With our first robots, it was a whole event. Recurring customers will still come out and follow the robot around, especially kids. One family kept ordering every day, and the kid would be outside waiting for it. And all the neighborhood kids would come out and it was like a block party. The parents loved ordering just to have the kids come out and play with the robot.

Somebody recently sent me a video on LinkedIn with their kid sitting on the curb waiting for the Coco robot, and she runs out and gives the robot a big hug. That kind of stuff is really cool. We’re trying to save businesses money and trying to improve the overall food delivery experience. But it’s cool to see communities react this way to it.

We’re working with the City of Santa Monica to understand our emissions impact, and in Q1 of next year, we’re on track to be reducing 27,000 pounds of carbon every single day. That’s a pretty sizable impact in such a short term in one small geography. If you multiply that forward as we keep growing and keep adopting cleaner and more efficient technologies—that’s one of the things I’m most proud of.

I ran an environmental club in high school, grew up a surfer, and have been hosting beach cleanups my entire life. So quantifying this type of impact is really special.

We have a zero percent cancellation policy. We refuse to have any order go undelivered. What happens if you have a catastrophic failure and the robots are down? Merchants don’t want to sign up for a new technology and be part of a test run if it’s potentially compromising the customer experience. We always have a human presence available with our field operations team, in case they need to physically intervene.

It used to be pretty common when we launched. You build a robot in one night and throw it out on the streets. Sometimes they break down or run out of battery. Brad and I were running around to save the deliveries. It’s a much smaller portion of our deliveries now. But even if it’s just a couple deliveries a night out of a few thousand, it’s still important. We want some way of catching it and not just having it ruin somebody’s night.

We’re still improving our robots all the time though. We have an entirely new design coming out in early December that is much larger than the current one. We can’t fit pizzas in our current robot. The new fleet can fit 18-inch pizzas and up to four grocery bags, so it’s designed for all merchants. We put in a pretty advanced suspension system on it, so it’s much smoother and can handle uneven terrain a little better. It’s built from the ground up for our use case. With the original robots, our goal was just to build it from off-the-shelf parts and get it into merchants’ hands.

Coco robots are basically like scooters with a cargo area. It’s the same motors, wheels, and batteries. There’s nothing that complicated about the device itself. Thanks to the popularity of electric scooters, we didn’t have to develop that supply chain from scratch, and it accelerated our ability to get Coco to merchants quickly. We have a couple manufacturing partners, and we’ll get various sub-assemblies sent to us, but we do the final assembly here in California.

What’s complicated is designing the best robot for merchant workflows. For example, in our new fleet, we added QR codes to the top, and you can put any order in any robot. We used to assign an order to a robot. Now, you can scan any robot and it will unlock, then you put the order in, and tell us what order you put in. We added a keypad because sometimes that’s faster for some merchants, and you can punch in the order number instead of scanning. Plus, we made the robots taller because we work with a lot of ghost kitchens and drive-thrus. So it’s easier for them to access it through a window.

We have thousands of robots currently being deployed. We want to replace the majority of trips that cars are making when it’s just a two- or three-mile radius. How many cars are required to support the top 10 cities in the US, and how many robots would it take to replace them? That order of magnitude is what we’re setting up for today. That’s likely to be close to 100,000 robots.