Podcast: Play in new window | Download | Embed
Today, most new cars on the road are built with features designed to assist with driving but fail to deliver. comma.ai’s open-source software, openpilot, enables your car to steer, accelerate, and brake automatically in its lane. In this episode, Cameron Herold is joined by comma.ai’s Senior Vice President of Operations, Viviane Ford, to talk about the company’s journey from having a start-up team of thirteen to creating an open-source self-driving development kit that allows automation across various car brands and becoming profitable in the world of open-source business. Viviane also lets us in on her role as the SVP of Operations—identifying comma.ai’s image and taking care of the team culture. Join her and Cameron in this conversation as they talk more about openpilot and open-source technology.
Viviane Ford is the Senior Vice President of Operations at Comma.ai based in San Diego. Most new cars on the road now are built with features designed to assist with driving but failed to deliver. Comma’s open-source software openpilot enables your car to steer, accelerate and brake automatically and its lane. It’s easy to install and trusted by thousands of drivers with over 10,000 miles. Viviane previously worked at Eastwood Communications for one year back when there was six people in a garage in San Francisco. Viviane, I’m looking forward to learning what you’re working on. Welcome to the show.
Thank you for having me.
Tell us a little bit more about Comma so we understand where these devices are being installed or how they’re being installed in cars. What kind of cars are we installing them in? I’m especially excited about autonomous vehicles.
Comma sells a hardware development kit that works on existing 2016 and newer Hondas, Toyotas, and Subarus. We have a whole range of supported cars. We sell this hardware that runs our open-source software. Back in 2017, we open-sourced this thing called openpilot which is autopilot for existing vehicles. A driver assistance system that takes you on the highway. On the highway, it will drive you 90% of the time and then when you’re doing backroad driving, you obviously take over. We have thousands of cars driving with both the Comma EON and openpilot. We have over 10 million miles of driving data collected and are working on upping that number, getting more compatible cars, and more people with the system in their car.
This is a third-party software so are you starting to license this software to any of the car companies or are you selling this as the after-market addition to vehicles?
The software is under MIT license. We open-source this on GitHub. If someone wanted to go and build a system that ran the software and then sell that system, they’d be completely entitled to. We sell to whoever wants to buy it. If Ford wanted to buy 100 Comma EONs hardware and install them in their vehicles, they could go ahead and do that. We don’t have a business development department. We have a website with an Order Now button and we will fulfill on Mondays and Thursdays.
Years ago, I rode in the first Google experimental car at TED. I was at the main TED conference sitting in the audience and someone from Google came out, talked about this vehicle and they showed a video. I’m sitting beside Jeff Bezos and Bill Gates is in the audience and they’re talking about this car that none of us had seen. I don’t think many have even heard about it years ago. The person who’s doing the demonstration said, “If you want to drive in it, it’s out this door on the parking lot.” One thousand eight hundred people stood up and went running. I got to drive it and I have a video on my laptop of me driving in this Prius with Matt Groening, the Founder of The Simpsons, was beside me in the vehicle. We drove on a test track at 45 miles an hour. I was going, “Holy shit, this thing is extraordinary.” I’ve since owned two Teslas. I’ve been driving them now for years. Do the MIT labs develop the software and then you guys are third-party licensing it?
It’s under MIT license but we did develop the software. We maintain the software. The beauty of having open-source software is that you get a community that can contribute to it. There are many makes and models out there that are a little different in how they run. When you ask a community member that wants this to be supportive for their car and if they go through the effort to send a full request to have their cars supported and merged with the next version of openpilot, that’s fantastic.
You’re building a community of people that are actively working to get to push forward the technology. A beautiful thing about the open-source code is if anyone’s concerned with the code, they can go and read through it. If there’s safety concerns or whatever, to some extent code is speech. They can go ahead and have a look at everything we do and decide for themselves if this is something they want to put in their car instead of getting into a Honda Civic with the Lane Keep Assist System and having little idea of how it works, trusting it, engaging it, and seeing it for yourself how it works on the road.
With the open-source idea, why is it scaring the hell out of me, that somebody can download this and put it into their car? I drove my Tesla Model 3 from Scottsdale to Vancouver and let it drive 2,000 to 2,200 miles on its own but this scares me for some reason, what am I missing?
There are a couple of reasons that this is scary. The first one is that anytime there’s an autonomous vehicle accident, it’s publicized until people can get every single last click on their story possible. We have many ideas in our head that these systems aren’t safe or, “Did you see that? This autopilot accident caused this issue.” Whereas along with that article, what if we also publicize how many people died by driving a regular car that day? There’s that level that’s a little alarming. The open-source code also scares people because responsibility is a hardware to place in this situation but it’s like, “Take a look at our code. If you like our code, then you’ll want to explore and experiment with it and see how it runs with your Honda Civic.” There’s an active exploration phase that the driver is using because they’re downloading this code from the internet and they’re trying to get in their vehicle as opposed to a black box delivery of Tesla autopilot, “Here you are. Take a drive.”
Where are they testing this? They’re downloading the software, they’re throwing it in their car, and they’re going for a drive around Scottsdale.
We have millions of miles of driving data from this. We have hundreds of YouTube videos of people driving around with the system. Where are they testing it? It’s better tested almost than any other self-driving car company’s product because they don’t even have a product. Part of the big issue too is there are so many edge cases with driving that it’s impossible to hand-code all of the edge cases of driving. Elon Musk had a line when we first came out with this system and he was like, “Ninety-nine percent of this is easy. It’s the 1% that’s hard.”
When you try to capture all of the edge cases like Waymo, I don’t know what their fleet is. Let’s say their fleet is 100 cars or whatever, but it’s not going to get you all of the edge cases. When you’re relying and depending on all of the data from your own cars as opposed to decentralizing the data consumption and getting data from people all over the world, all of a sudden you have a much more fully packaged data exploration to go on.
Were there a lot of legal hurdles in getting launched or getting the product out the door?
We got a letter from the National Highway Transportation Safety Authority back in 2016. Regardless, they were concerned. We had announced at TechCrunch that we were selling this product with openpilot on the product and this is going to drive your car. There was a lot of hype. Nothing can drive your car. You still can’t sit in a fully self-driving car. Everything and the driver assistance system is still level two. Similarly, when you ask why am I scared of this? There’s a lack of information around a lot of this stuff. A big thing with open-sourcing is how do we show the information? How do we explain what’s going on and publish more of the information and data? We tried to do this with more Medium posts.
We’ve tried to publish more like this is how we’re using the data, this is how we’re being compliant with NHTSA’s requests. They sent a list of seventeen questions. They were reasonable questions asking us like, “What is this product? What is it doing? How can you guarantee certain aspects about the product?” Apart from that, the trend of the general self-driving space is moving more towards, not even away from regulation. Waymo comes out and announces that they hyped it up a little too much for the first X years that they’ve been working on self-driving cars or you have all these companies that are slowly realizing that this problem is difficult. Not that regulation catches up but if the problem slows down that the pace will think it’ll be solved and that reassures people automatically.
People are getting more comfortable with it now. We’re not hearing about the accidents constantly. We have heard the arguments that there are other cars that are crushing. Years ago, it was weird hearing a Tesla creep up behind you and now, we could never hear them before and now we can hear them which is strange. How did you get involved in the company?
We lived in this house called the Crypto Castle in San Francisco. It was a three-story townhouse in downtown San Francisco with a bunch of people who were interested in Bitcoin and cryptocurrencies. George, the founder, moved into the basement of this house. He was looking to start his own self-driving car startup and needed a basement where he could hack away a car. I met him and we became friends. I lived and saw him develop this company for the first 7 or 8 months. Several years later, he was looking for an operations person. I went up to him and I said, “You should hire me.” It took some convincing. About two weeks later, I was offered the job and I took it with very little operations experience.
You guys got involved. You saw it at the earliest stages. Did you launch it in New York?
That was in 2016, the very early stages. He had done a Bloomberg article in 2015 which was the first unofficial official launch of Comma as a company. People were intrigued to see what he was up to and what he was doing. A tech country announced this whole shebang of a product trying to fall into the hype of self-driving cars before you quickly realized that all of that is dangerous to fall into. We open-source the software in 2016. We open-sourced the hardware required to run in the software but we weren’t selling it. We had a community of people building out this hardware and working on the product. We launched the Comma EON, the hardware that supports openpilot if people want it. When we launched it, it was for people with a dashcam development kit. We launched that in 2017 and have consistently been updating with new versions of both EON and openpilot.
How are you funded?
Through E16(c). We are still Series C. E16(c) led our two rounds. We did two series seeds.
How much did you raise so far?
About $8.1 million.
That’s substantial for a seed.
It was two seeds. The first seed was $3.1 million. The second seed was $5 million. We have revenue and we keep operations very lean. We’re a team of 12 or 13. The real goal has been, how do we become profitable, which is not a question that startups ask themselves. It was important that if this is going to be sustainable, how can we make money along the way, as opposed to blowing through millions and millions of dollars through mobile technology that you can’t even ship?
I was brought in as a coach for the team at Hootsuite years ago and I led their strategic planning retreat and I pushed them to consider profitability as a way to scale instead of more equity and debt financing. They decided to go for it. I’m like, “I don’t understand why people don’t. It’s simple to build a real company if you decide to and you have way more control.” Is that your focus then?
What you said was to build a real company. That’s exactly what we want to do. It a way to have more control. It’s also the easiest way to know that you’re adding value to the world is are you making money?
If people are buying it, you’re making money off of it. It keeps you lean, smart and scrappy. We think when WhatsApp sold for $1.4 billion, they only had 50 employees. You guys are on the same trajectory. I love that he was working with getting some early-stage free PR as well. I wrote a book called Free PR. It talks about how to leverage free publicity for your brand and how to generate it in-house instead of using a PR firm. You guys have got some huge upside clearly with PR, I’m sure.
George, our President, the Founder, he comes from a hacking background. People have followed what he’s been up to and then what his projects have been throughout. That was helpful too. It’s an underdog story that Silicon Valley loves and all the people in San Francisco love where this one guy puts together some self-driving car that competes against Teslas. Whether or not it’s true or how much hype you want to have, if you get a story like that to start off your company, people are going to be excited to see where it’s at.
I missed out on the seed round at Tesla by not putting a $5,000 deposit down on it. The original Roadster years ago. That was a bad call. I also told the founder of Uber before he launched, I took him to Burning Man and he was explaining it to me and this was before he hired Travis. You’re onto something. If I told you it’s a bad idea, go all-in though as well.
I’m not worried about it. It’s like, “Can you tell me exactly about that?”
I know how you got involved in the business. When you started dealing with the whole open-source, what can you teach us about that? What are the hassles or pitfalls and the struggles with the whole open-source side of the business?
The beauty of being an open-source business is that you’re not going to open-source your software and then be secretive about everything else you do. You’re going to be an open company. It’s going to approach what you do with an open mindset. A default mindset of Comma is publicized or be honest about what’s happening. We don’t want to have things to hide because if you have secrecy, you don’t have anything at all, is the idea. That’s a very positive reinforcement of what open-source does. For me, from a personal perspective, I joined Comma from a tech PR firm that was very much, “Embargoed until this date. Don’t you dare break that.”
I entered into this world of Comma where all of these people deeply admire the open-source community. You can’t help but get stressed about potential competitors that decide to use this or get a leg up on whatever they’re doing, but then you realize that that’s exactly the point where if you do open-source this and you do give, even people who aren’t necessarily direct competitors. If this information is out there in the world that everyone already has a plus one to having this information so we help. If a competitor comes along and builds a better hardware system that’s going to run this system better, then you better work your ass off to make sure that that’s not the case and that your system is always promising a better experience for users. For me initially, it was a jolt of a different way of thinking.
You said you came in and you didn’t have any operations experience. How did you learn it?
The thing I quickly realized was that you don’t want glory in operations. You want to be as behind the scenes as possible because it means the company is being run and people aren’t realizing that something is an issue. You don’t want there to be any issues, which was a different approach of like, “How do I go about this? How do I make sure that the things that are happening are happening in a way that people can do their job?” Thankfully, when I joined Comma, we had 10 or 11 people because we had interns and those interns left. We were down to six at one point. It’s a six-person company, if you’ve never done operations before, you can learn and you can manage quickly. Because Comma has scaled a very appropriate rate, it’s been a good way of learning. When you’re in San Francisco, every other person does operations to some extent.
The Bay Area has got to be one of the most competitive areas for tech talent even for the entrepreneurial scene. How are you attracting talent?
Working at Comma is different than working at almost any other company. With the open-source factor, you’re going to attract people that are interested in the open-source community and that was key. The number one way for us to attract talent was through pouring over people’s GitHub making sure that people were already contributing to the open-source community. People who are excited about those projects are going to be excited about open-sourced driving assistance systems. That was a helpful way. It is hard when Google is going to pay a machine learning engineer hundreds of thousands of dollars and you’re trying to keep your startup very lean and you’re not going to offer that.
What you get on the flip side is people that are in it because they do believe that this company is going to succeed, which is always the goal. People who come in, you give a sliding scale of equity and salary, and people who take this lowest salary and the highest equity is always a positive reinforcement of where this person thinks the company is going. It’s hard to recruit. It’s very hard to attract talent and going to San Diego, one of the ideas was like, “We might be able to attract talent more easily. Qualcomm is here. We’ll be different than the thousands of startups in the Bay Area that are all trying to attract the same people.”
How did that play out?
Ironically, a decent amount of our employees have come from our community. That’s proven to be more and more the best way to get people is people that find themselves in the Comma community are already interested. If they contributed positively floating the idea of joining the company, it’s an exciting next step. Realizing that San Diego is also a more pleasant place to live in the Bay Area. Bay Area is amazing. It’s easier to attract people from wherever in the US to move to San Diego than to move to San Francisco.
San Diego has got a good technology sector and entrepreneurial sector.
There’s a lot of biotech but also that helps because we’re not biotech so we’re different.
I’ve coached a lot of companies and I’ve got a number of different coaching clients that are in San Diego and a couple of our members of our COO Alliance are in San Diego as well. It’s a great market. What are you guys struggling with right now? What are the pain points?
The main pain point is, what do you want your company to look like? We’ve built this company and you’re working towards profitability. For now, this is the ultimate goal in the immediate future, but what company do you scale into? What kind of company do you become? How do you decentralize the responsibilities of the company so you don’t have one single point of failure being your CEO or whatever? At the core of it, it’s our product and the company. We talked about a lot of stability. Stability on the product and stability in the company. How do you build something that will outlast everyone that is currently at this company? Those kinds of thoughts that you have. If you’re profitable, you can self-sustain for hopefully a very long time and then what does that look like in the future.
How are you giving the vision to your team right now? How are you articulating it or getting them aligned and inspired? How do you also keep a group of people excited when they’re coming on for equity and the company is seemingly moving towards profitability focused? Does that distract them at all? By all rights, they should be focused on profitability anyway, but if it’s a build or bust, that’s a different goal.
How you keep people incentivized is an interesting one.
Less incentivized but more aligned and inspired with the vision so they’re not trying to focus on an exit.
What helps is having an actual user base is we have a community discord. A lot of our users are on that discord. A lot of our employees are on that discord talking to users like having beta programs and stuff like that. That helps because you see what you are doing and affecting people that you were talking about. Being a thirteen-person company, what you work on has a big impact. You’re not working on a project with ten people and then have that be scrapped. If you don’t do something, it’s obvious and it hurts the company.
That alignment will come naturally by people seeing what their work has. In terms of the profitability equity thing, I don’t even think it’s directly related to Comma when people hear that. It’s more of a realization of the state of feces in the state of San Francisco and Silicon Valley was like, “The money that they’re handing out to companies will dry up.” If you do want to succeed, win that self-driving car, and sell self-driving cars, what better way than to create a profitable company that doesn’t buy on these investors? That mindset helps in general people get on board with putting profit first.
We’re open about that. Our highest sales month was around $160,000 in revenue. We spend at a rate of about $180,000 a month is our burn rate. We are close to getting those two numbers are aligned which would be great. Ideally, we don’t need to raise money again. We are profitable early in 2021 and can continue sailing.
How are you getting the word out right now? How are you guys getting the buzz? Is it PR or your community?
It’s the community. We don’t have anyone here that does marketing and isn’t an engineer apart from myself. I do operations. That’s engineering in some source in some way but we don’t have a biz dev department. If someone posts a YouTube video and then it gets on Reddit or something like that, how are those sales coming? It’s not the simplest thing when you purchase the EON dev kit. You’re like, “Is my car supported?” You have all these questions. There’s still a lot of low-hanging fruit of the people that come to our website. How can we make it so clear for them if they want to purchase it? What are they purchasing? What do they need to do? Getting more people to the site is a goal. We’re still focusing on the people that do come to the site on how can we make sure that it’s abundantly clear what they could be buying or not.
This is not expensive. This is $600. That’s ridiculous. The autonomous vehicles will be the cause of the next baby boom. You guys can quote me on that. You can put that in all of your PR. You think about people driving in LA traffic for an hour, what are you going to be doing? You’re not going to be hanging out in the front seat. You’re going to hop in the back. I’ll be like, “I’ll commute to work. I don’t want to work from home anymore. I want to commute.” Do you run on a remote team or do you all run in the office together?
We have one guy who is remote but he was with us for a couple of years. He knew very well how the company worked. We don’t have any meetings. We have one meeting a week on Monday mornings to sync up what’s happening. I joke that everything is a media in Comma in the sense that at lunch when you miss a conversation, you’re missing a lot. At all times, people are talking about their projects and what’s the holdup so it’s important to be on it.
It’s critical in the early stage of a business to have that cooler time and lunchtime and interacting. That beta transfer is amazing what happens there. How about yourself? What are you working on your skills as a second in command?
I run operations here. It’s everything non-technical. I do operations, recruiting, accounting, legal, HR and all of that stuff. We have some exciting things coming in 2021 that we’re working on and recruiting in this area or recruiting in San Diego, understanding how do we build out teams here, reaching out to people in the US and abroad, and all the things that come in operations. If I were to give you a given day, it would change drastically from one day to the next.
How about technology tools? What tools do you use or does the company use to scale, to manage?
I don’t know exactly if these are the tools you’re thinking but I’m such a big fan of these adorable startups in San Francisco like Gusto and Xero that are incredibly helpful and have fantastic customer support because they need to be and want to be. I’m almost pleasantly happy that the startup worlds is a thing even though it can be unbelievably frustrating. When you’re trying to get corporate credit cards for your employees from Wells Fargo and they require everyone’s Social Security Numbers and all of these things that are such a headache beyond belief, and that makes everyone feel so uncomfortable to have to give to some big bank. You stumble upon a company like Brex that you sign up online and corporate cards mailed to you in three days. To me, I’m happy with it because otherwise, it’s such a pain in the butt to have to deal with.
Gusto is the HR payroll software and Xero is the accounting software.
Brex is corporate cards so we use those.
Considering you started out in the Crypto Castle, do you accept any cryptocurrency as payment for your products?
We used to and then it was such a pain in the butt with taxes. It became so not worth it. Back when we had this open-sourced hardware called the Comma Neo, a terrible neon green color. Back in the early days, there was a company that was building them and selling them. They were building them and selling them for one Bitcoin which at the time was $700. In December 2017, people know Bitcoin rises to $17,000. Neodriven is their name. I hope they did well with their Bitcoin accept transactions.
I got to go back and look at the dates, but it was 2009 that I was telling people on Facebook. It was when Bitcoin was $400. I was accepting Bitcoin as payment for speaking events and for coaching. People are like, “You’re an idiot. Why would you do that?” I ran a digital currency company years ago. I had Starwood Hotels, Bose Stereo, Avis Rent-A-Car, Hard Rock Cafe, all paying me with a digital currency instead of the US dollar. We sold the company in 2000. We had 30,000 businesses buying and selling using a digital currency. It was called UBarter.com. We sold for $64 million but it was a real business.
They were open to using digital currencies.
I had Starwood hotels. We had little swipe cards. You could use it. Some of them were using their merchant processing. It was a multilateral trade. It wasn’t back. We met with guys at Goldman Sachs and we’re like, “What’s backing your barter dollar.” We went, “Nothing.” They went, “You print money. You’re like a country.” We’re like, “Pretty much. We print our own currency.” If we needed more money, we would give ourselves a credit line. We’d go buy furniture for the office with our own cryptocurrency. It was crazy. I’ve always believed that. There’s a huge need for it. One day, it’ll come back around that you can start accepting it again.
It’s on the up and up despite how people might feel about it.
When you’re a lean startup in a smaller team and you’re focused on profitability and not burning through cash, what are you trying to say no to?
There’s a very fine line between being too strict with the cash. Where does productivity and cash spend meet perfectly? A big overhaul was random services that you sign up for that you forget that you’re signed up to and then you realize that at the end of the year, you spent $15,000-plus on services. We’re cutting down on that. Honestly, a good amount of PR can be ridiculously expensive either if you don’t know how to use ad spend correctly or you can hire a lot of consultants that a lot of times don’t add value and is ridiculously expensive.
What can I do? How much would it cost if someone else did it? What’s the value if someone else does it versus me doing it? Since it’s a small team, it’s faster for them to do most things on their own instead of using another service or whatever. A thirteen-person team in San Diego, we have competitive salaries now that we’re in San Diego but an SF, it’s definitely less. You start to almost naturally be more lean than when you’re in San Francisco.
Take a look at my book, Free PR. You guys might have a huge opportunity to leverage free publicity in-house and how to generate. When we built 1-800-GOT-JUNK, we landed 5,200 stories by ourselves with no PR firm. That was prior to Facebook launching. We had no social media platforms to leverage our PR on. That was all individual uniques including Oprah.
I should shout out 1-800-GOT-JUNK because you guys saved us so many times. We’ve been working out of a house in San Francisco and stuff piles up quickly.
That was our fifth franchisee, the guy who owns the Bay Area. He does $20 million in San Francisco on junk removal. He’s killing it. The last question we’ve got. If you were to go back to your younger self starting out in your career, what word of advice would you give yourself that now you know to be true but you wish you’d known then?
I would say at the end of the day, the responsibility falls on you which seems an obvious thing to say, but even if you were managing someone and they mess up, that inevitably falls on you. The second you think in that way, you start to think more like the company, you start to think in a much more productive way to getting stuff done, shipping out products and moving a company along because it’s no longer individually based. Thinking in the sense of what responsibility do I give, do I trust that that person will get it done and if they don’t, that’s on me, what do I do?
What did you do wrong that let that happen? Ben Horowitz in his book, The Hard Thing About Hard Things talked about that level of introspection. It’s important for leaders that often when we’re firing somebody, we’re like, “I’m glad they’re gone. They’re terrible at this. They’re terrible at that.” You hired them. What was wrong with the interviewing process? What was wrong with the recruiting process? What was wrong with the top rating, reference checks, the training program, or the leadership? How did you let this happen? When we take that level of ownership and introspection, that’s powerful.
On top of that, it doesn’t happen again because now, it traces back to this point when this was a mistake as opposed to letting it go as it’s their fault, it’s going to happen again.
Viviane Ford, thank you for joining us. I appreciate you sharing everything that you guys are doing with Comma and excited about what you’re doing for autonomous vehicles.
Thank you for having me. This was a lot of fun.
Thanks for sharing.