Autonomous Delivery Robots with Coco CEO Zach Rash

So, we built the Cocoa Robots to be, you know, the best autonomous vehicle for transporting goods. So, it's smaller

transporting goods. So, it's smaller than a car. It's more compact, doesn't take up a lot of parking spaces. It

helps get cars off the road and reduce our traffic and congestion and emissions in our cities, but it's big enough to fit four grocery bags. It can fit eight extra-L large pizzas. So, it can do all the types of deliveries that people need

in a city, but you're doing it in a much more compact, energy efficient way. So,

we always figured, you know, the the autonomous vehicle of the future of of delivery was going to look quite different than a car.

All right. So, today we're thrilled to welcome Zack Rash, co-founder and CEO of Coco, the autonomous delivery robotics company, transforming last mile delivery in cities around the world. Since 2020,

KCO has grown from a scrappy idea into a category defining robotics company completing over 500,000 zero emission deliveries across multiple major US cities and abroad. Under Zach's

leadership, they recently raised 100 up to $110 million in backing from Sam Alman, Founders Fund, Pelleon, and others. Koko is building the world's

others. Koko is building the world's largest urban autonomous delivery fleet and redefining the future of sustainable logistics. Zach, how are you doing

logistics. Zach, how are you doing tonight?

Great. Welcome to Coco After Dark.

So, I always have the same opening line.

How are you doing today? But I think this is the first podcast I've ever done at night. So, thanks for inspiring me,

at night. So, thanks for inspiring me, you us.

Yeah. Yeah. Well, we we were doing an event here and Harry and I were joking that this looks like a podcast studio.

This is our lobby in our office. And so

now it's a podcast studio.

All right. And so usually, you know, we have people listening to the podcast and sometimes watching on YouTube. So if

they want to see our faces and the cool lobby vibe Coco After Dark that we're going for, we spent about 20 minutes figuring out the lighting. So I think it might be worth popping over to YouTube and watching this.

You definitely have to look at it. And

let me know about this art piece. There

was some hot takes on it. There's

actually we found this at a local art studio in Santa Monica. Someone who

works at the company like walked through this art gallery and all these pieces had cocoos painted in them and so we bought a few of them and that little pink thing in the in the prairie here is

a is a cocoa robot.

All right. Well, if one is missing tomorrow morning, don't blame me. It

might be up in my I do have a spot for it in my office. So,

perfect. I last interviewed you on the Ride Share Guy podcast back in 2022, so it's been a few years. And you guys, you know, we were just talking off air. You

guys have made a ton of progress. You've

landed a bunch of partnerships, fundraising. I feel like you're really

fundraising. I feel like you're really on a roll. I'm excited to dig into the future of sidewalk delivery, the future of autonomous delivery, and everything that you're building. But first, give us, you know, us. I think I feel like

everyone's probably seen a Coco robot at this point, but tell us about the state of the business, where you guys are at, and you know, kind of how you're doing.

Yeah, so we built the Cocoa Robots to be, you know, the best way to transport, the best autonomous vehicle for transporting goods.

So, it's smaller than a car. It's more

compact, doesn't take up a lot of parking spaces. It helps get cars off

parking spaces. It helps get cars off the road and reduce our traffic and congestion and emissions in our cities.

But, it's big enough to fit four grocery bags. It can fit eight extra large

bags. It can fit eight extra large pizzas. Yeah. So, it can do all the

pizzas. Yeah. So, it can do all the types of deliveries that people need in a city, but you're doing it in a much more compact, energy efficient way. So,

we always figured, you know, the the autonomous vehicle of the future of of delivery was going to look quite different than a car. So, it's a, you know, it's a big tub on wheels. It can

go on the sidewalks and then in some areas it can actually go on the road and bike lanes to to go at higher speeds.

So, you can think of it as we're attempting to basically build a, you know, eventually it should approach the performance of a of an ebike courier, right? where they can kind of weave

right? where they can kind of weave through traffic, they can go right up to the door of the restaurant to get the food and then they can really, you know, nail that handoff experience with with the customer. So, we started this in

the customer. So, we started this in 2020. We're partnered with, you know,

2020. We're partnered with, you know, most the major delivery platforms like Uber and Door Dash here in the US and we're live across all of the LA area and

Chicago Miami Helsinki Finland and have a number of cities both in the US and Europe coming up soon.

Very cool. Awesome. And I feel like, you know, I saw recently you guys had a big fundraising announcement. Well, I know

fundraising announcement. Well, I know it's been a few months now, but I feel like there's almost like a second wave coming now of sidewalk delivery robots.

Do you feel the same way? And if so, what's behind that?

I think I mean there's two things, you know, I think there's two things at a macro level that have been happening.

One is I think the need for this sort of solution has has become, you know, much larger in the in the last few years. you have in in major cities,

few years. you have in in major cities, you have a lot of pressures on cost of living and affordability um that are headwinds to continuing to grow delivery, right? It's just really

delivery, right? It's just really expensive and and it's super it's such an awesome service. It's super

convenient. Door Dash, you know, Door Dash and Uber Eats are are awesome services, but if we want to keep growing this or accelerate the growth rate of of these services and make it more affordable, more accessible to people, the cost needs to come down.

Yeah. And the you know one of the obvious tools to do this is to is to complement and supplement human drivers with autonomous delivered vehicles. So I

think there's been an increased need for this sort of technology. At the same time the technology is clearly becoming capable and there's clearly been this step function change in AI and the sort

of technology that powers all this. You

know first it was chachbt clearly showed there was a big step function change in in the tech. Then you had Whimos go from, you know, this R&D project that's been going on for over a decade to all of a sudden you see Whimos everywhere.

Yeah.

And so I think there's a time when you you you need a solution and the technology is clearly there and the time is now to to invest in this. So I think that's causing a big robot wave in general. You know, humanoids,

general. You know, humanoids, self-driving cars, delivery vehicles, I think all types of robotics. I think

that's going to be a huge wave. There's

been I mean robots in general are hot and all the rage right now. How do you guys fit into these different categories? Do you

consider yourself autonomous vehicles, robotics, or does it kind of depend on who or what or where you're talking?

Yeah, you know, people have different definitions. It's I think generally

definitions. It's I think generally there's like there's the there's a thesis that people are building towards of the humanoid robot which is this general purpose robot that like can do a lot of things

and it's not necessarily the best in the world at any given thing but it is like human level you know aspirationally human level capability at a lot of things so that you as a you know person

living in your house could could buy one and it gives enough utility for you to be worth the 20,000 30,000 however much it costs. We're in the other category

it costs. We're in the other category where we're, you know, very specialized at saying, how do we build the best type of robot to do logistics? Yeah. Right.

And today that's a lot of groceries, food, a little bit of retail items, but in the future, I mean, that's anything and everything moving around our cities.

And, you know, you obviously want something that's super power efficient, can travel longer distances on on, you know, very low power, have a can carry a lot of weight, and is ultra low cost, right? So, you want something that's a

right? So, you want something that's a few thousand dollars, not tens of thousands of dollars.

Got it. If you're moving pizzas and you want to be extremely energy efficient, so you don't want to have, you know, a bunch of actuators where you run down the street. We'd like to roll down the

the street. We'd like to roll down the street. So, I think you'll see that

street. So, I think you'll see that trend in a bunch of different areas.

You'll have the more general purpose and then you'll have more specialized robots. And I think both will grow very

robots. And I think both will grow very quickly.

Yeah. So, you guys are in that specialized robot category, it sounds like, you know, doing last mile delivery. And I mean, when I walked into

delivery. And I mean, when I walked into your office just now, I'm always I mean, I I've seen them many times before, but I feel like every time I see them, oh, it's a lot bigger than I realize. Have

they gotten bigger over time too? How

has the product evolved, you know, maybe from, you know, kind of an operations from a product standpoint, you know, over the last few years?

Yeah. So, the the design constraint, you know, how we designed the vehicle was you want to be able to fit, you know, it doesn't have to fit every type of order possible, but you want to probably fit at least 90% of the types

of orders that that happen, right? So,

that has to deliver for for for for I'd say for like on demand use cases, right?

And so that's typically food, groceries, you know, things that you're ordering more frequently. So you're not going to

more frequently. So you're not going to do a big order, you know, once a month or something like that on.

I mean, you said four grocery bags can fit in. I mean, that's pretty hefty,

fit in. I mean, that's pretty hefty, right? That's most orders, I would say,

right? That's most orders, I would say, right? So, so you want to make it enough

right? So, so you want to make it enough to fit most the types of things. Um, but

you also want it to be about human shoulder width. Yeah.

shoulder width. Yeah.

So you're not taking up too much space on the sidewalk. You need to let people pass. You don't want to be an

pass. You don't want to be an obstruction. And it's important to be

obstruction. And it's important to be able to use the sidewalks because the merchant is not going to want to have to walk into the street and look for your robot somewhere, right? The the kind of

current full size self-driving not always somewhere to park in the street. Sidewalk's a little easier to to

street. Sidewalk's a little easier to to pull right up.

So the, you know, there's a lot of a lot of the minutes of a delivery, you know, are finding parking and then walking to go pick up food from from the merchant, right? So we don't, you know, we don't

right? So we don't, you know, we don't want to like put all that time onto the restaurant. They do not have time to do

restaurant. They do not have time to do this, right? They're never going to

this, right? They're never going to adopt technology like this if it makes their lives a lot harder. It's already a very difficult job.

So you we designed the system to say, how do we make this actually a good experience for the merchant and and something where they can feel like they take back a little bit more control of the front of the house or the restaurant and who's coming into the restaurant and

they can kind of create a little bit more order, but isn't a lot of extra work. So most

of our merchants will prep the food. If

they're really busy, they might take out multiple orders at once and they just drop them into a couple robots. they go

back inside and the robots take off.

So, it's really important to have a vehicle that can appropriately and safely fit on the sidewalks. So, so

yeah, you're about human shoulder width.

You want it to be about the the, you know, the weight of a of a human. So,

you around 100 pounds, right? You don't

want hundreds of pounds and you don't want it to be so large that it's not really safe or appropriate to be on the sidewalks.

Got it. And the robots, are they driven autonomously today or does it depend on the city, the order, or the destination?

So the the the way we always started was we wanted to be very when we started the company we want to be very good at the teley operations side which are the remote drivers driving these and the reason that was important is because the

clear trend in AI is to train these systems what's called end to end.

So instead of creating a map a map of the world and using a bunch of LAR and creating a 3D map and then saying what is the the what is the least likely to hit something path that I can draw through this map and let me just do that

you know 20 times a second. The current

state-of-the-art approach that that most people are are doing now is you take all the raw sensor data in. So you take all the camera data, you take the acceleration, you take the how fast the, you know, the the wheels are moving, you

take all that information in and you directly output what would a good driver do in this situation. And we have one of the largest data sets ever of robots interacting in the real world, especially on the sidewalk where you

have a very unstructured, very dynamic environment that has a lot of like social cues and important things, right?

We're driving by storefronts. We're

we're driving in residential areas. You

have a lot of What kind of volume have you guys done over the past years in terms of number of robots or deliveries or what have you guys talked about?

So So we we have we have you know we have about a thousand vehicles say over a thousand vehicles today. We're we're

kicking off mass production of our next vehicle which we'll we'll unveil soon.

That one will make about 10,000 for next year. So you know in terms of data scale

year. So you know in terms of data scale we're looking at you know many millions of miles of hundreds of thousands of deliveries.

Yeah. I mean probably around a million deliveries now and millions of miles of trip data.

So that data set not just the video data set but like how does you know a very high quality driver you know human in the loop how do they handle that situation? What's the right way to

situation? What's the right way to navigate that and understanding what data to train on and how to train the system to handle those situations autonomously? That's been the key. So

autonomously? That's been the key. So

right now where we're at is we we we have some hybrid system where you can have the system drive autonomously, but we're really good at context switching in a human to make sure that the the the

delivery goes without delays.

And how does that context switching happen? Does if it runs into a

happen? Does if it runs into a situation, I mean, it's only going a few miles an hour, it can just pull over to the side because I think that's one of the challenges when you start increasing the speed, right? I mean, I think one of the reasons why we sell Whimo took so

long to launch on the freeways is because you can't just pull over on the side of the road if something goes wrong. Even if you're going 20, 30, 15

wrong. Even if you're going 20, 30, 15 miles an hour, you have some flexibility. When you're going 70,

flexibility. When you're going 70, things get I guess 65 because they follow the speed limit here in California.

It gets a little dicey.

Yeah. The And this is kind of what I what I go back to of like we didn't start by saying let's make this thing fully autonomous and then humans can take over when it fails and it's like someone in the office who can jump on

it, right? We are very good as a at a at

it, right? We are very good as a at a at a product level. How do we present information to the human so that they can solve whatever the task is? They can

intervene. They know what city they're in. They knew the rules of the road in

in. They knew the rules of the road in that city. They understand why the

that city. They understand why the autonomy system failed. They understand

how to recover it.

And that happens in seconds.

Got it.

If that doesn't take seconds, this this doesn't really solve anything. If the

food's cold, there's no amount of autonomy that like, you know, no one's going to use the product if it's like fully autonomous, but the food's always cold. So that needs to be done really in

cold. So that needs to be done really in a really deliberate fashion. And how we control the ratio of vehicles on the roads to the human operators is based on,

you know, the the average average delay caused by autonomy interruptions as well as the 90th percentile, the 99th percentile. So we look at kind of all

percentile. So we look at kind of all the worst cases and we say, let's make sure the food's still on time within a within a reasonable bound.

So is it fair to say that you guys do a hybrid operation or how do you describe it?

Yeah, I think it'll always be a hybrid.

you just eventually want, you know, if you have millions of robots, you're going to want pretty high leverage humans in the loop who are doing a lot more high level problem solving.

I one thing I've noticed in the autonomous vehicle space, there seems to be this I don't know about fear or apprehension. People don't want to say

apprehension. People don't want to say that things are teaoperated, right?

Because it's like, oh, then it's not autonomous. Seems like you're open to

autonomous. Seems like you're open to that I mean Whimo for example they say that you know if the car gets stuck they're very deliberate about saying that you know they're not tea operated but a tea operator will come on and it

will tell the AV where to go you know draw a line or you know kind of on the path and I don't know I I don't think it's that big of a deal like to me it seems like you know as long as it can safely you know if it's in a parking lot doesn't know how to get in what's the

big deal about having a teller operator help it out like some situations are humans are great at and you know maybe AVs will be great at that in the future but right now there's obviously a lot of areas they can help.

I have no idea why that's like such a contrarian thing to talk about. I I have I that was a unique part about our pitch, you know, and I I've had I had like in the early days I had investors tell me like, "Oh, cool. So So you're

doing it in the way where it's like the super low tech way. So you're not really a tech company. You're more like a like an ops like a like a traditional operations business."

operations business." And that to me was just so crazy because I'm like, "Okay, we're still building like a robot that's streaming data globally around the world." and and you know we have to build a whole routing

and dis dispatch system and map system.

I mean you know it's at least as technically complicated as building like a human courier tech tech company in in many ways you have to recreate a lot of that stuff. So but but to me that was

that stuff. So but but to me that was always the right sequencing right because again with hot food you are so sensitive to time and delays and we're operating in the most chaotic parts of the world. were operating in the

the world. were operating in the downtowns of cities and the city centers and these are the most lively places we have in the world and they're extremely unstructured especially on the sidewalks and so but I still need to get you your food in

15 minutes or you or you will not use my service and and and so that can be a lot of that driving can be solved very reliably with AI today but having that human in the

loop is both economical and good from a quality of service standpoint and you can work both of those in. I just think people have this like feels like you're faking it if there's like a human involved whatsoever. I see this on the

involved whatsoever. I see this on the humanoid side too where it's like, you know, you'll see a robot like running or something or doing some task and people like, "Oh, but was it teleyoperated?"

And it's like teleyops has nothing to do with like how impressive the motor control of that humanoid doing that running task or like hill climbing task or something is. I'm on your side because I mean I think there's a lot of these situations where you know I did an

interview recently in the driverless trucking space and you know I mean I think AVs are great at driving down the road 70 miles an hour and then when you get to the you know where the drop off point like have a human there they get in there they back it up. I think it's

called a hitch and a toe or something like that. Hitch in a post and you know

like that. Hitch in a post and you know when they have to do these very complex you know tight turns navigations in out like that seems like the perfect hybrid.

Maybe AV could do that in the future but I don't think you need to solve that piece of it now. Yeah, and everyone is heavily using teley ops. It's just it's just like no one really wants to talk about it. Um I don't know if it ruins

about it. Um I don't know if it ruins the magic or or what not, but it's like a I mean that operations in the background is like a material part of the operations business and it's something you need to be really good at.

Like that matters a lot, you know, and like if you've ever had a human intervention in a Whimo like they're really good at it. Just don't talk about it. Are are your tell operators are they

it. Are are your tell operators are they based you know here in your office or do you think it's possible to use folks you know in other parts of the country or other parts of the world when it comes to tele operations?

We we've had a lot of different thoughts on this. So it started as me and my

on this. So it started as me and my co-founder Brad did all the deliveries.

Actually before you came we were just doing a lot of driving.

I think I saw a couple robots moving around when I came out.

Just making sure I still got it. But

then it turned into all of our college friends. So, you know, we were doing we

friends. So, you know, we were doing we started this at UCLA, like, you know, in Westwood, and we kind of yelled out the window to all of our neighbors being like, "Does anyone This literally happened. Brad yelled out the window.

happened. Brad yelled out the window.

He's like, "Does anyone want to help me drive a robot?" And our neighbors were like "Sure."

like "Sure." And they recruit their college friends.

We had kind of this US like college network.

Cool.

And now that we're operating in multiple countries, we have a system where any operator in any country in any city can actually operate any robot in any city.

that gives you obvious like staffing advantages because you have like a global supply supply base, but you do lose some context. Like it's a little disorienting to go from like, you know, Chicago in the winter to then Miami in the summer to then like, you know,

you're in a Finnish forest and it's like dark out, you know, like this and and and you know, Finland's got different driving rules like the bike lanes in Finland are on the sidewalk in Oh, really?

So, you actually have a designated lane that you have to stay in on the sidewalk before you even get to the road. So you

know we do a lot of work on the product to present that information to the teley operator so they have all that context but you know you are like switching them around a lot. So we might become more localized over time but from a you know

like labor efficiency perspective you you do kind of want a global pool.

So taking a step back more broadly autonomous delivery you mentioned you know we've talked about Whimo talked about Door Dash. I mean, I guess Whimo and Door Dash are a partner. I think

they're live now in Phoenix doing these Dash Mart deliveries, which is their kind of ghost store, right, where you can kind of get more convenience type items delivered. What do you think about

items delivered. What do you think about the autonomous delivery space and, you know, maybe in the context of verse rid share too?

Yeah. So, I I think they're they're very different problems. You know, it's like a similar technology, but a very different product challenge and a very different customer experience you're trying to solve for. So obviously on the delivery side you have you know two

different kind of customers right you have the merchant and you have that end customer getting the getting the food ride share is just that one that one user and you know one thing that I've found interesting is in the ride share

side of things it's been you know I think surprising in my own preferences and I think from people I talked to where the experience of sitting in a Whimo is so awesome.

Um it's a new car it's clean you kind of have the whole space to yourself if you're doing a meeting or you're going some with your friends. So, it's such an awesome experience in the car and that's such an overwhelming number of the minutes of the experience is sitting in

the car. So, even if it takes a

the car. So, even if it takes a different route or if it's following the speed limit and going slower or you have a maybe inconvenient pickup or drop off and and and that's all getting a lot better. But even with that,

better. But even with that, a lot of people still have a preference for for a Whimo ride. In the delivery space, it's very different because those minutes of travel, the user is not

engaged.

And so they do not like, you know, they don't care if the ride of the food inside the robot is really pleasant.

They care a lot that the food arrives in the condition they expect. It's warm.

It's correct.

And they care that it arrives on time.

Yeah.

You know, and and and they care that that handoff experience is really good.

So, you basically just have to like make no mistakes, but you can't. There's not

that many opportunities to add all these like wow experiences like you could in the 20 minutes you're sitting in a Whimo.

You know what's funny? I've never

actually thought about that, but you're right. When you're in the ride share,

right. When you're in the ride share, you know, when you're in an autonomous vehicle like a Whimo, for example, I mean, it's kind of a good experience.

Like, obviously, it goes a little slower. it follows the speed limits and

slower. it follows the speed limits and it's not, you know, maybe it wouldn't do exactly what you want, but it's such a great experience that you still and you also still end up going from A to B,

right? So, it's like if it takes longer,

right? So, it's like if it takes longer, not a big deal. Kind of like I joke when you know if you're in a really nice hotel or you take a business class flight to Japan or something like that, right? It's like it's a nice flight.

right? It's like it's a nice flight.

Like, does it really, you know, I don't care if it's a long one, right? Because

I don't really want it to end, right? If

it's something really nice or fancy. I

haven't actually thought about that before. You're right. Yeah. So we don't

before. You're right. Yeah. So we don't have that opportunity. So I think the the the bar to set a good experience for the customer is a lot higher which makes it challenging and and then there's all these other parts that you know that you

still in ride share you have the pickup and the drop off right and those are hard and and autonomous vehicles that can be difficult.

But for food, I think it's even more sensitive, particularly on the merchant side. Like, you know, I was saying

side. Like, you know, I was saying earlier, if you don't pick up in exactly the right spot, I mean, the merchants's going to like hate using the service, right?

So, the precision of the pickup, the precision of the timing of the pickup, because we're not like running into the store and and sitting there and waiting for them and grabbing the food. We do

need to have that relationship with the merchant and we need to make sure that's a good experience. So that handoff needs to be done really well.

And that that's one of the reasons we designed the robot to I mean the first day of our business like Brad and I took a prototype robot we built in our living room and brought it to a restaurant.

Mhm.

Um it was Alpha Alpha on Main Street in Santa Monica and then we had LNK Market which is the the kind of corner store, convenience store, liquor store on Main Street and we built the product with them and

they gave us a ton of feedback early on.

the product barely worked, but we like we knew what they actually wanted and and how to build that experience. And

the tiny details of that matter a lot.

Mh.

Because if somebody, you know, we have high volume restaurants that might load 200 orders a day into a robot, you know, that experience matters a lot. And then

do they load it in? Do they type a code in or what do they do?

So we've created different experiences for different types of users, right? If

you do 200 a day, you know, you're a kind of a power user at that point. And

we let you pick and assign robots. You

can actually do the first order I have ready can go into any robot and will give you like a taxi line. It's kind of like you know how Uber would build like an airport model is different than your typical pickup experience.

So that like enterprise restaurants or ghost kitchens that's typically the experience.

But then if you have like a mom and pop that might do an order every few days with us like that needs to be a very different experience. We're not optimizing for you

experience. We're not optimizing for you know high frequency use. We're

optimizing for it to be reliable and very intuitive and require the kind of the least work possible. So in that experience, yeah,

possible. So in that experience, yeah, there's a code, you punch in the code, you open it up, and and you don't you don't really do anything. You have, you know, we're assigning the robot, we're putting it there, and your tablet will tell you to go to go load it.

Those things need to be done really well. Every small error on that

well. Every small error on that experience is just such a bad experience for the merchant. And so you need to be really excellent there to have the merchants love you and want to to continue using this.

Yeah. So, one thing we've seen with delivery, I think you touched on it. I

mean, it's expensive to get a human to go to a restaurant, pick up your food and then drop it off at your house, right? Like, it's kind of a lot of work

right? Like, it's kind of a lot of work for that human. And then obviously they're sitting in between and obviously a lot of people complain about the high cost of food delivery. They're still

using it. I think Door Dash is doing quite well, but as a company, but people still use it. And I think the first thing that stood out to me because I I kind of know all the stats, you know, I've done Door Dash, I've done Uber

Eats, I've delivered food, right? We we

cover it a lot on the ride share guy.

You know, over I think over 50% of a courier's pay is actually tips. Yeah.

When you do autonomous delivery, you don't have to tip. I think that is I haven't really seen many people talking about that.

Yeah. I think there's a few reasons for that. I mean, in in our European

that. I mean, in in our European business, this is also different.

There's just not a tipping. There's not

an exp of tipping. So, that's like less of a benefit that you can kind of market from an affordability perspective. I

think the other challenge is right now the experience if you would order on Uber Eats or Door Dash and you get assigned to, you know, Cocoa Robot for that delivery, you're still checking out with a higher price, right?

And then if you get matched, your tip will be refunded, right? So from like a you know elasticity perspective, you're not actually producing more demand because it's like it's like maybe delightfully maybe you'll be delighted

that it's that it's a little bit cheaper. But you but you've like you've

cheaper. But you but you've like you've already you've already like committed to that amount a couple other fees.

Yeah, you've already clicked order and said this is how much I'm spending and you're already you know feeling the pain of that and then and then the refund is kind of like happens later if you look at your credit card statement. So over

time though I think we want to make sure that we can create an extremely reliable service with our partners where you know if if we say we can do a delivery in X minutes like we can always

I mean that's like a big a big lever to pull in the future to you're right right now it's kind of a surprise and people don't bake that in so they're not ordering more because of that right the elasticity but in the future if you know it was like hey you are going to get

this refund right even if it was based on like an average of let's say six out of 10 orders and then six out of the 10 customers we give that the other four, you know, we don't, right? It might

still work.

Yeah. And and and this is ultimately up to our our partners and how you know, their marketplace dynamics and different partners are going to have different philosophies on this, but our job is to deliver an incredibly reliable service

and and and do so at a price point that gives them the leverage to do these sort of things, whether it's on the merchant side, the customer side, their own margins to run to run a better service, whatever it might be.

But today, we can profitably do our deliveries at a much lower cost than a human courier. I mean, particularly in

human courier. I mean, particularly in like a Los Angeles, right, where, you know, Cal California the cost for, you know, cost of living here is just so high that and and and you have Prop 22 and you have some of these these, you

know, regulatory frameworks where the the the amount of pay going to occur per delivery is is is from a consumer standpoint is quite high. So, we're

already much lower than that. And so, we are our job is just to give reliable service and give the kind of affordability levers to the to our partners to then work their marketplace magic for their customers.

Yeah. I I I I thought delivery has always been very interesting, right?

Because you think about ride share, you've got people delivering in 20, 30, 40, or in Whimo's case, $75,000 vehicles plus all the hardware and sensors. And

obviously the cost, you know, the average order value is higher on ride share, maybe 2, 3x delivery, but you know, when you're delivering, right, you might be on an ebike, for example, right? Or you might be, you know, I'm

right? Or you might be, you know, I'm guessing your robots are a lot cheaper.

So I have to imagine, you know, the unit economics, I don't know what they look like, but it sounds like like can you talk a little like what are the margins on some of these deliveries or how many deliveries you guys need to do to how does how do the financial piece work

out?

Yeah. So I mean you can think about our costs are in two categories, right? You

have your your your you know and these are loose categorizations, but you have your fixed costs and you have your variable cost. The fixed cost

variable cost. The fixed cost is the cost of the vehicle that you're depreciating and then it's all the things that go into the total cost of ownership. charging, cleaning,

ownership. charging, cleaning, maintaining, repairs. You got to store

maintaining, repairs. You got to store parts in each market, right? You have to manage a supply chain and and then the reverse logistics on those parts.

So, you have all the day-to-day and the month-to-month management of the fleet.

And then you have the cell data, which is material, right? We have we run three SIM cards on every vehicle. We have

we're always streaming data on two carriers to make sure that the signals signal is very reliable. It's a lot of data. So, yeah. So got the hardware

data. So, yeah. So got the hardware cost, all the fleet management costs and then the the cell data costs. Then you

have other stuff like insurance and and you know some of these other fees, property taxes so on and then you have your variable costs which are you know part of the part of the you know repairs are variable right if you drive more your tires going to wear down more but

largely the variable costs would be your teley operators or any any human in the loop right and so as the vehicles get more and more autonomous that cost dries to zero very close to zero pretty quickly. Yeah, the variable side

pretty quickly. Yeah, the variable side and then the fixed side is what you're left with, right? And that is a that is a matter of how do I make this at the right price and then drive a ton of utilization through it, right?

And that's where I think the human ride share and you talk a lot about this, but the human the human courier side of the marketplace and the autonomous vehicle side of the marketplace are very different, have very different incentives. So when you put an

incentives. So when you put an autonomous vehicle on the network, you kind of want it to do base like the kind of the base load capacity where it is kind of just doing trips

all day every day, rain or shine, all weather conditions, but we're we're not as good at flexing up to like meet 7 p.m. demand, you know, for dinner,

p.m. demand, you know, for dinner, right? And food deliver food delivery.

right? And food deliver food delivery.

Yeah, I was food deliver has always had much stronger peaks, right? Door Dash

and Uber Eatats in Door Dash especially, right, has always had more of a scheduling system, right, for dashers.

like they want you to schedule their hours because they kind of need to control the supply a little bit more than you know like ride share. I mean

Uber Eats does this too, but you know you can kind of log on whenever and wherever you want, right?

Yeah. So, so I think that's why, you know, these hybrid networks are are really are really powerful of of you know, human gig workers that can come on and flex up and earn a lot during

dinner. But the robots provide that base

dinner. But the robots provide that base that base capacity. And I think those two things combined make the ecosystem cost overall a lot lower. I think that grows the pie a lot. I think that's like great for robots and I think that's also

great for driver earnings. And I think it will continue to grow the overall marketplace. But yeah, but you know, and

marketplace. But yeah, but you know, and we've gotten to the point now we have thousands of merchants using our our vehicles across, you know, these different platforms. So, we're now getting to the point where we we will

have vehicles just non-stop driving the whole day throughout the city and they'll kind of go between these different neighborhoods. Like we have,

different neighborhoods. Like we have, you know, one of our densest deployments is, you know, the Westwood Sautel Brentwood kind of that West LA pocket.

There's a lot of people there. We have a lot of restaurants there. And so like all those vehicles we deploy there are driving all day.

So we get really good utilization. And

so then if you get really good utilization, the thing you need to be very good at is you need to make sure this is not like you know materially more expensive than than a car or than even than an ebike, right? So it needs to be a few thousand.

right? So it needs to be a few thousand.

If it's tens of thousands of dollars and you're trying to transport 10 or $15 of food, it's really hard to make the math work even if the utilization is really high. Passengers are a little bit easier

high. Passengers are a little bit easier and the vehicle should be more expensive for passengers. You have you're

for passengers. You have you're traveling at higher speeds. you have a lot more safety requirements and redundancy requirements because of the safety risks involved. But the lower speed vehicles that are much lighter weight, your planning horizon is a lot

shorter, your stopping distance is a lot shorter.

So all of that should fundamentally make the the vehicle much cheaper to make and that ultimately gives us a much lower cost.

So the vehicle cost is an order of magnitude lower and you're using a hybrid autonomous and you know human system but over time maybe the human system you know goes towards zero

hopefully. It seems like you know those

hopefully. It seems like you know those you know either one would be you know more more you know efficient from a unit economics basis right so I guess the combination I mean is there a you know

100 deliveries it needs to hit and how do you think about that utilization like we want our robots doing one delivery an hour you know 5 to 10 a day and then once they get to 100 we're we're looking good or how do you think

it's it's probably about right I mean you know one an hour is pretty like would be pretty good uh with no batching so today we We don't do any batching. So

when we pick up an order, we go to the customer and we go pick up the next order. We will never pick up two orders

order. We will never pick up two orders and then do two drops.

Yeah.

And that's not like a religious thing.

We, you know, if we could do that and make it a great experience, we will.

That is like a necessary evil of food delivery because the economics just do not work if you don't do batching.

So we might introduce hardware eventually that does batching for the right types of orders if we can really guarantee a good customer experience because right now our economics are good enough where we don't need to do

batching. So we're we are not. So, you

batching. So we're we are not. So, you

know, I think that without batching, you know, one to one and a half is kind of the most efficient deal.

I'm not a fan of batching as a customer.

If I put my customer hat on, I I pay the three bucks on the Door Dash, you know, I think Door Dash only charges three bucks. Pretty good deal. I pay the three

bucks. Pretty good deal. I pay the three bucks on Door Dash every time. The, you

know, once in a while when I'm feeling a little cheap or, you know, I Oh, the restaurant's pretty close. I'll probably

get it first. You know, I think I had one the other day. They did two deliveries before they got to me. And I

think the unique thing about food delivery is, you know, you talked about ride share. You always get from A to B,

ride share. You always get from A to B, right? Whether it's an AV, even if it's

right? Whether it's an AV, even if it's a human, even if it's not a great driver, he doesn't drop you off around the corner. Usually, 99.9% of the time,

the corner. Usually, 99.9% of the time, you get to your destination. It may not be the exact experience you wanted or the price or whatever it might be. With

food delivery, it goes from A to B, but A to B matters plus the time. Right?

You've mentioned time a couple times. If

I'm ordering something hot, I mean, burger, fries are some of the most common orders, right, on a lot of these big platforms. Fast few, fast food, QSR, I mean, every minute the food quality is literally deteriorating, right?

Yeah. And and and and this is, you know, I think you you mentioned this, you know, you can pay like a priority fee, right? You can pay three or four dollars

right? You can pay three or four dollars more to get your food delivered to you with no batching. You can guarantee no batching by paying more.

This is kind of what I'm saying. By

giving a lower cost structure, like a fundamentally lower cost structure to our partners, they can choose how to use this lever. Like another way you could

this lever. Like another way you could do this is you could look at this as like this could become the premium option without that fee, right? Rather

than being a lower cost option that our partners could choose to say, hey, this we can control that the like the robot can can offer you the same at at the same price, we can offer you a guaranteed no batching experience,

right? So there's a lot of leverage you

right? So there's a lot of leverage you can do if you have a like a fundamentally lower cost structure on on these subset of deliveries. So I think there's a lot of opportunity to do that and yeah the the the time in the robots

like incredibly important to the to to the customer.

So I'd like to end on where you and Coco fit into the ecosystem of on demand and delivery. I mean you've mentioned a few

delivery. I mean you've mentioned a few times you mentioned your partners with Door Dash and Uber Eatats. You also

mentioned restaurants. Sounds like are you also partnered with restaurants or you know how do you guys fit into this ecosystem?

Yeah, so you know I mean ultimately we want to build a extremely reliable logistics system that can move anything and everything around around our cities.

So that might be restaurants, that could be retailers directly, that could be packages and and then that's the the the large marketplaces. And so we we we you

large marketplaces. And so we we we you know partner with anyone who has a delivery need and you know it's a it's a fundamentally pretty similar product, right?

Yeah.

We need to pick it up at a certain time.

You need to make sure it fits into the vehicle and then you need to drop it off at a certain time and if you can do that affordably and at a high quality people people will use it. We started really focusing on food just because the

perishable nature of food is what makes it so expensive to deliver. Got it.

Because you just you're so sensitive to the timing and batching and and you know the typical things you can do with a package or with mail or like an Amazon truck, right? Is you can do 100 pack

truck, right? Is you can do 100 pack deliveries at a time because it's time like it's time sensitive for the customer but it's not perishable.

I like that time sensitive but not perishable.

Perishable is just way more expensive and it's a way harder logistic system to set up. So we started there because it

set up. So we started there because it was the area where we felt like hey even when we have a high ratio of teley operators we probably you know with with software efficiencies and a super lowcost vehicle we can we can make a we can make a

difference on the on the on the cost of this and then as it's gotten more more and more autonomous I think there's more and more categories where we're price competitive and then over time as it trends to zero variable cost all fixed

cost you ultimately are aligned with the like the local like if you look at the local economy and like the local all the different retailers in in that in that area. You are aligned with them of

area. You are aligned with them of saying this is some asset we've put into your neighborhood. How do we make how do

your neighborhood. How do we make how do we like add as much value as possible to all of these different retailers and that will fundamentally lower our costs because you're advertising that same hardware over more and more trips. And

so we are aligned with all these retailers if the more they use it, the more use cases they can think of, our costs continue to come down.

Yeah.

And that's not true for human delivery because there are limits to how many people are willing to do that job. So in

the limit, if we make millions of robots and put them in all these markets, they can start doing more and more useful work that will continue to lower the cost because we can keep producing more and our costs get lower when we produce more. And I think ultimately that's

more. And I think ultimately that's going to make the cost of all the stuff extremely low and I think that's going to be I mean really good for the convenience and affordability of these services. These services are amazing and

services. These services are amazing and if more people have access to them more frequently, I think that's a very good thing and it's really good for our local economy. A lot of things get delivered

economy. A lot of things get delivered by Amazon or other other retailers that are in your neighborhood.

Yeah.

And so if you can keep a lot of those dollars more local by making it more you making it affordable and like cost competitive to ordering from one of the bigger retailers that are doing it in some warehouse outside of the city. I

think that's like really good for the local economy. It it keeps your you know

local economy. It it keeps your you know it keeps that flavor of culture in your community. And I I think that's like I

community. And I I think that's like I think that's probably what most people want in their in their cities and this technology should should make that happen.

Definitely. Yeah. No, I I think that all makes sense. And you know when it comes

makes sense. And you know when it comes to So are you partnering with some of these restaurants and retailers directly? I mean you can you know maybe

directly? I mean you can you know maybe get mashed on Door Dash or Uber Eats it sounds like through the restaurant right but it's through the partner and then others you partner directly. Do you have a preference or you'll take anything and everything? I mean, I know there's a big

everything? I mean, I know there's a big debate right now, you know, on the food delivery side, right? With restaurants,

obviously, I mean, they want to be serving orders themselves, but, you know, it's it's nice popping on a Door Dash or Uber Eats and you get a ton of demand and they charge a high commission, but they're obviously offering a lot of value through their

customers. We we we want to be great

customers. We we we want to be great partners for everyone and and so we're we're we're kind of agnostic and we just want to be available to as many people and and you know the more the the more partners we have the more efficiency we

get from the fleet the lower the cost gets the faster our pickup times get. So

this becomes a bit of a flywheel in a market where we can offer a higher quality service at lower price and that can keep going by the more people who use it. So you

know we we've you know been working with Uber for over a year now. We've been

working with Door Dash for a few years.

We've announced like a few weeks ago, we've been working with Dashmart. We've

done tens of thousands of orders with them across LA, Chicago, and Miami.

I mean, yeah, I guess there's also the adjacent categories, but I mean, I guess my my real question is, I mean, it's kind of Uber and Door Dash are the main partners. I mean, I think there's been

partners. I mean, I think there's been over the past five plus years, there's been a big push, you know, to help restaurants do first party delivery and all these different tech tools and services. I don't know that they've

services. I don't know that they've really taken off. It sounds like that would be probably a good thing for them, but I mean it's tough running a restaurant and then it's tough doing delivery. It seems like, you know,

delivery. It seems like, you know, things are maybe gravitating towards Uber and Door Dash. Is that what you're saying?

Yeah, I mean, we're giving them the tools to to do if they want to invest in their own digital ecosystem and use our our our robots to do the deliveries.

They have access to it and they have access to this that same technology that Uber and Door Dash have.

But ultimately, like, you know, they need to create a compelling value value proposition. They need to want to do

proposition. They need to want to do this. Yeah.

this. Yeah.

And we do have some that do that. But

we've been working with Sweet Fin here in LA for years and we're integrated into the Sweet Fin app and they have their own loyalty program on there and they've invested in that. They've they

have a great marketing campaigns. They

have great food and that's been a great brand we've been working with and we did deeply integrate with their first party experience. That's something they cared

experience. That's something they cared about and they were able to drive like actually good demand and good retention through that. So we we'll we'll we'll

through that. So we we'll we'll we'll support that. But you know ultimately

support that. But you know ultimately like we'll make the service available to everyone and then and then you know the the the customer will ultimately decide where they want to order from and what experience is the best. Yeah.

And that's not for us to decide and we just want to we got to do our part from getting from A to B.

Definitely. All right. So last question, good one to end on. What verticals? All

right. So last question, good one to end on. What are you most excited for about

on. What are you most excited for about in the future of Coco? Is it new autonomous delivery verticals that you're you're going to be expanding into new products? What where do you think

new products? What where do you think you know your kind of real excitement lies in the sort of years ahead?

I I think there's a bunch of cool product stuff in the future that we're going to announce. A few, you know, a few of the areas that we're investing a lot into. I mean, the first one is just

lot into. I mean, the first one is just being able to increase our speeds. So,

driving faster creates a better customer experience. It has us gives us better

experience. It has us gives us better utilization because we can kind of pick up from a wider range, wider radius of of of different merchants with the same vehicle. Speed is always a good thing.

vehicle. Speed is always a good thing.

So, what does it take to go faster?

Well, you need to do more road driving.

You need to have a vehicle that's, you know, more capable in the roads, has better suspension system, has higher top speeds, and you need able to do all that very safely and reliably. So, that's an area that we're really excited about that I think is going to expand the kind

of total addressable market of these types of vehicles and improve the customer experience. Longer term, I

customer experience. Longer term, I think there's a lot of really interesting things happening in the world of both AI and the hardware side where you can make these vehicles a lot

more capable to solve some of these experiences. Think, you know,

experiences. Think, you know, you know, think if you could if you could actually pick up an order off the shelf yourself or you could open a door, you could go upstairs. I think there's a lot of things where the tech's actually getting there. It's not yet affordable,

getting there. It's not yet affordable, but I think it's getting to the point where we can economically make the vehicle a lot more capable to make the customer experience and the merchant experience a lot better. And I think that's inevitably the direction this

goes.

Very cool. When I did my orientation for Door Dash in Orange County, it was in person. That's how long ago I signed up

person. That's how long ago I signed up to be a Dasher about 10 11 years ago. I

remember someone in the group asked a question. Do we need to take the order

question. Do we need to take the order to the door? And they said, dude, it's in the name Door Dash. Yes, you need to deliver it to the door. So, I think that'll be cool if there's robots in the future that not only are picking up,

delivering, and you know, maybe even going upstairs or you know, we would love to take it to your door, navigating what they have to do to take it to your door and no tipping and that'll happen like we will be able to

get it to your door. It is a very hard engineering problem. But like ultimately

engineering problem. But like ultimately that's the beauty of robots is you can create these this this you can create this product that will work in all weather conditions all times of day all

times of year. It's ultra reliable and it will drop it off at the exact right place exactly when you want. And I think that's like aspirationally our whole product roadmap is to do that. And it's

a really hard problem. Ultimately it can it can be achieved and I think that's going to create a really great customer experience.

Excellent. Well, I'm excited to hear about the future and you know, we'll definitely have you on the podcast again. Appreciate first episode of Coco

again. Appreciate first episode of Coco After. Thanks for coming in the books

After. Thanks for coming in the books and folks want to learn more about what you're up to or Coco. Can they go to the website? What's a good spot for them?

website? What's a good spot for them?

Yeah, cocoely.com and then our socials are some combination of Coco Robotics and All right. I think I' I've been packing

All right. I think I' I've been packing a good Instagram post. So, whoever's

running the Instagram account, I think they're doing a good job. So,

trying to get our our our content frequency going.

All right, Zach. Appreciate it. Thank

you.

Thank you. Heat. Heat. [music]

[music]