Interview: Vignesh Ganapathy, Serve Robotics VP of Public Policy
On building institutions for new technology
Serve Robotics builds autonomous vehicles (AVs) for last mile delivery. Their robots navigate city sidewalks and deliver orders on platforms like UberEats. While the technology is interesting, the way that regulatory regimes have begun emerging around autonomous delivery robots is important to understand on its own. To that end, I sat down with an expert.
The following is a conversation with Vignesh Ganapathy, Vice President of Public Policy at Serve. Vignesh and I met years ago when he joined the Public Policy team at Postmates and spearheaded the project of getting permission for robots to operate on public sidewalks. The work he did runs counter to the mantra of asking for forgiveness instead of permission and his experiences have added nuance to how I think about navigating government as part of rolling out new technology.
Public Policy for Robots
Jeff Fong: To set the stage, how long have you been working on public policy for AVs and what does that entail?
Vignesh Ganapathy: I started working on public policy for delivery AVs back in 2018 when I joined Postmates. It was a new area for me, but I've always been an urban planning and transit enthusiast, so it played to those interests and forced me to get more knowledgeable about things quickly.
In terms of what it entails, we’re supporting the advancement of new technology in a way that fosters social acceptance and brings other people along. I think of it as a public affairs question, “How are we building support for this aside from just notching individual wins?”.
JF: When you say acceptance, you mean rolling out technology and getting members of the public, policymakers, whoever, enthusiastic about it?
VG: Yeah. Tech companies and businesses see a lot of pushback these days. But there are many technologies that have greatly improved people's lives, like…Mapquest was revolutionary, and I think if you went back and suddenly banned MapQuest or Google Maps, you’d have a lot of really angry people. You want to make sure that people actually understand the technology’s value. And that means designing and building the technology with that value in the first place.
JF: When Serve got going, what was the initial launch city and why?
VG: Serve started out as the special projects division of Postmates back in 2017. Short distance deliveries tend to be the most difficult ones. They're the least lucrative for drivers (they end up having the most idle time between jobs), and they're the most challenging from the perspective of consumer satisfaction. You have drivers circling the block several times looking for parking and other issues come up like folks seeing a delivery driver stopped in the bike lane waiting for an order. So we were interested in seeing how we could augment our delivery fleet to include new technologies. We initially partnered with outside vendors but eventually brought the technology in house.
The tech was initially developed in San Francisco where we were located , so it made the most sense to start there
When Permission > Forgiveness
JF: Starting in San Francisco, how did you make the decision to preemptively engage with local authorities rather than just yolo’ing and dealing with the consequences?
VG: Honestly, with San Francisco, it was a little bit of building the ship as we sailed it. We were doing all of this at the very beginning of a broader tech backlash. Uber and Lyft had significant pushback from communities both in terms of a broader labor conversation and increased congestion on streets. Austin, for example, had banned rideshare temporarily. We also saw the pushback against Airbnb given its potential effect on housing values. And we’d also just seen the proliferation of electric scooters that occupied many parts of our sidewalks and streets without any sort of consistent expectation among folks in terms of how they should operate. Many cities had banned scooters all together. We realized that for a technology that most people had not seen in the wild, we might need to be a little bit more thoughtful. And be prepared for the same sort of backlash that those other technologies had been facing.
JF: During that process, y'all were engaging directly with municipalities, correct?
VG: Yeah
JF: How did y'all make the decision to engage at the municipal level (versus trying to go to a state level authority like the ridesharing platforms did)?
VG: So, there are a couple of things at play here.
One is the regulatory regime for sidewalks is (where we operate) versus streets (where ridesharing lives). Streets are governed by state departments of transportation and sidewalks are largely municipal. In order for us to roll out, you could go to a municipality and say “I want to put a robot on the sidewalk” and you could just put one out there. But putting something on a roadway requires licensing, a designation in the vehicular code, and a whole regime in place. So I think the incentives for the rideshare companies were a little bit different.
Additionally, the CPUC1 had been chosen as a regulatory authority, essentially because rideshare had decided they would cede some of the advantages of municipal control where they could talk to local lawmakers more easily, in exchange for preemption. They wanted consistency across cities, and especially when they have drivers driving incredibly long distances. One of the major shifts in California politics was that preemption became far less powerful in terms of any regulatory regime, especially anything tech. The dynamics around tech accountability groups, organized labor, disability rights groups, affordable housing advocates and others reached fever pitch, because people didn’t feel like they were consulted in the process. And without a process in place, we all wanted to be able to petition our local governments on how these technologies would be deployed. If Serve had gone to the state level, we probably would have seen more regulation without getting the preemption part, and I think that would not necessarily have been an exchange we were willing to make. So local governments made a lot more sense.
JF: So this is a story of proprietary platforms, like Airbnb, the electric scooter platforms, etc, creating backlash and changing the politics?
VG: Yeah
Political Economy for Entrepreneurs
JF: At the beginning, y’all were engaging with city level administrators. How did you figure out who to even talk to?
VG: Good question. First off, it's been different in each place. When we first tried to engage in San Francisco, Postmates/Serve wasn't the only company trying to receive permission to operate sidewalk delivery robots.
There were companies such as Marble (RIP), and Starship, who all had an interest in rolling out these technologies. A lot of the outreach happened at the Board of Supervisors2 level, and in some ways it was more of a political conversation than it was a policy conversation. A lot of supervisors had cold feet. They were already starting to receive backlash about other technologies in the space and they weren't ready to stick up for us, understandably so. We had just come out of the period where Airbnb regulations had been put in place as well as a broad based electric scooter ban in the city. They ended up saying, “we're going to ban it, but we'll create a path for you all to work with the city to develop a regime.” This is how San Francisco first developed its Emerging Technology office.
Los Angeles was quite a bit different, because LA is not a unified government. First off, there's a group for supervisors and there's a separate City Council. But additionally, LADOT (Los Angeles Department of Transportation) has long been a very thoughtful governmental agency on emerging technologies and receives some deference from lawmakers. They're very receptive to figuring out regulatory regimes and playing with pilots to see how we can get from here to there. At the time, they had just implemented a system that they called the Mobility Data Specification, which is a way for them to monitor how many scooters are currently operating in LA and whether companies were complying with their permits. Since then, that program has been expanded quite a bit, and to protect proprietary data, it's been spun off into a nonprofit that they take part in. But they've always been very thoughtful about building regimes for tech regulation.
So there were two different avenues there. We ended up going first to the Bureau of Engineering and the Department of Public Works. And then we spoke with the individual city council members representing the districts in which we wanted to operate. We also reached out to a legislator in West Hollywood who was interested in how they could convey that they weren’t hostile to all new tech, “we're not just hostile …we do have pilot programs, we do invest in the space”. So we were able to pitch them at the right time to develop a positive use case.
JF: Well, I think that’s a great jumping off point into my next question.
There’s a standard narrative within the tech industry – particularly in the period that y’all were rolling out Serve – that asking for forgiveness is better than asking for permission. And that if you ask for the go-ahead from government, you’ll die of old age and your coffin will be wrapped in red tape. To some degree, it sounds like y’all ran into that in San Francisco, but not in Los Angeles. What was the difference in LA?
VG: There's two parts as an answer. I think one is…LA's cultural politics around urban planning are very different from San Francisco's. Namely, San Francisco has a lot of entities that are competing for a limited amount of public space. And that results in a scarcity mindset among everybody. Everyone's fighting for their piece of the pie. Anytime anyone thinks you're trying to do X, Y, and Z, they're worried that you're encroaching on their right to do A, B and C.
JF: And by that you mean something like Serve taking up space on the sidewalk and, say, Disability Rights Advocates viewing that as zero sum?
VG: Yeah. Ideally speaking, you would be able to talk to the Disability Rights Advocates, talk to the Pedestrian Advocates, the Bicycle Coalition, and community members, and say, “Hey, we need wider sidewalks in this part of town.” Or “let's work together to get cars off the road because most accidents that impact pedestrians and disabled people or those with ambulatory challenges are cars that collide with them.” And I think because of that limited public space, the political culture of San Francisco makes it difficult to do because everyone's fighting for their piece. And most importantly, they’ve been burned by the history of other companies deploying.
In LA it's almost like you’re working from a far less developed baseline. I don't think anyone looks at LA and thinks this is an urban planning paradise. It has a long way to go in pedestrian and transit infrastructure. But because the city has this history of being filled with smog prior to the 90s and they're really struggling with ways to upzone and increase transit infrastructure, there's far more of a willingness to try new things to mitigate these issues. Folks are less entrenched and they're less worried about losing their piece of the pie because there's more to fight over. And this impacts the way that governmental agencies work.
LADOT, LA Public Works, LA Bureau of Street Services have all had to spend time being creative in this space. They’ve also dealt with what happens when you don't act. There was a lawsuit in LA around sidewalk quality back in the mid 2010s. That resulted in a large settlement, but they still haven't improved all of the sidewalks in the city. Not all of them are ADA3 compliant. There aren't enough curb cuts. Many of the traffic signals may still rely on push-to-cross. The push requests might not last long enough for someone who is disabled to actually cross the street. All of these are huge challenges for them and they're trying to figure out ways to improve that. If a sidewalk is unsafe for a robot, it's probably unsafe for someone who's in a wheelchair; it's probably unsafe for someone who is was elderly and using a mobility aid or is hard of hearing and can't tell whether there's an emergency vehicle coming. We want to make sure these spaces are friendly to all those groups. So we were able to find good allies and LA Walks, the LA County Bicycle Coalition, the pedestrian advocates in LA. Everyone was more willing to speak with us out of the gate because they saw a shared interest. We were able to build a broader coalition to enact these new technologies and it’s a garden we have to regularly water. I think that is reflected in the way governmental agencies reacted: “Hey, let's figure out a way to do this”. There wasn't a “no” out the gate, but rather, “How can we make this happen?”.
JF: It sounds like, at least in the case of LA, there were policy changes that you could get behind with other coalition members and that short circuited the zero sum competition that y’all ran into in San Francisco. Is that fair?
VG: I think it is. It doesn't mean that LA doesn't also have issues that can still percolate up. In West Hollywood there's always going to be residents that don't like delivery robots. That's a fact of life. And we're not going to be able to minimize our incidents to zero just as we can’t do so with planes, trains and automobiles. But safety is our highest priority…it's the thing that we spend the most time on, both from a design, implementation, and advocacy perspective. And this policy work is about currying favor among the public. The way we do that is by proving it in action. The fact that we have the ability to do so month over month, year over year, right now, in the second largest city in America, is a testament to how our team works day in and day out to prove that a short distance delivery really shouldn’t be made in a car.
New Institutions for New Technology
JF: When y’all initially showed up in LA, there was no pre-existing regime for regulating sidewalk delivery robots. What did y’all end up creating with coalition partners? How does that system work and what does it allow you to do?
VG: It was different in West Hollywood and LA, I think for political culture reasons, as well as the systems of government.
West Hollywood had just banned electric scooters. LA had not. We decided for we wanted to make sure we had explicit permission before we operated in any jurisdiction that had explicitly banned electric scooters. We didn't want to touch any third rails. We didn't want to end up being associated with the scooter debates or with policy battles from ten years ago. We wanted to be able to show our safety record and community bona fides on our own.
LA had not banned scooters and, additionally, when we met with LA's regulatory agencies, they said that each city council member was able to govern their own fiefdom.
JF: So member deference / alderman’s privilege?
VG: Yeah. In LA, if a city council member doesn't like something in their district, they have a lot of authority to quash or change it. Which, in some ways, is good. We wanted to talk to the council members that represented Hollywood, Mid City, mostly areas adjacent to West Hollywood, and let them know that we wanted to operate and who they could reach out to if they had any concerns. We let them know how many robots we were operating, and asked how we could be better citizens.
In West Hollywood, we went and said we'd like to create a pilot program, because clearly what happened with scooters was not working. They wanted to show that they were tech friendly and we wanted to notch a win by showing that we're receptive to the regulations. They came back to us and offered two consecutive six month permit programs where they’d allow us to operate three robots. And thankfully West Hollywood isn’t large enough to have 60 plus robots out there doing deliveries.
So we presented to different City Commissions and advisory groups. This included the Disability Advisory Group, the Public Safety Commission, Transportation, Senior Advisory, Public Facilities, and in each one of those instances, folks had the opportunity to ask questions and raise concerns that they had with the technology. And then the city worked with us to answer those questions. Over the course of the pilot, we wanted to show that we had a safe operating record, that we were taking cars off the road, and that we could create a reporting system to do so. That was our real trial run on it. We were able to get out of city council by a pretty thin margin, so we prioritized doing the approach of doing it right instead of moving fast. For the city council member who cast the deciding vote, I think it was very much because they didn't want to be seen as not tech friendly. While skeptical, they wanted to show what's possible.
After we had those two consecutive six months of operating terms we ended up going back to the city council and the city council gave us the green light. They had an RFP process and allowed many robotics operators to apply, but there’d be a 40 robot limit for the city of Hollywood. We were one of the recipients of a permit and now we're several months into the program.
In LA, we were able to just get permission from those individual city council members. After which we did continue to push for some sort of official regulatory scheme.
JF: So y’all had a political nod, but without a full blown regulatory system?
VG: Yes. And I think ultimately, we really did want a system because we didn't want to run afoul of any concerns anyone might have. And we’d get asked about it occasionally, right? Someone would say, “hey, are you allowed to do this?” and we’d say, “Well, yes, we have the permission from so-and-so lawmaker”, but that's not the same as having explicit written authority. Thankfully, we did receive a document from the city attorney that confirmed our right to operate.
But in 2020, Postmates was acquired by Uber and in 2021, Serve was spun out of the Uber apparatus. I left for a year, and during that time, the city created an official permit program which didn’t have a limit on the number of devices, but did require us to integrate with mobility data specification — reporting out how many robots we have functioning at any time — integrate with the city's 311 system, implement specific standards like braille marking on the device, and have other identifiers like a phone number to call if there are any issues. Funny enough, the ADA compliance piece had been suggested by us in our early outreach.
When I came back to the company a little over a year ago, we put ourselves in line to work together on those permits. We've submitted all of our documentation and we're just waiting on the city to issue its official regulations. It's currently going through a rule-making process, but in the meantime we have the authority to operate in all the districts that we've sought permission from, and to do so at scale.
JF: Are there other specific limitations on things like the speed at which the robots are allowed to travel down the sidewalk or the size of the units?
VG: Initially, LA's draft regulation (which never actually went into enforcement) had a 100 lbs weight limit. Our device started out at less than 100 lbs, but we added safety features that pushed it over. We added mechanical brakes, for example, which are much safer than electronic brakes because they're a failsafe braking mechanism. When we went back to the city, we let them know that we've made these changes and while the actual dimensions of the device haven't changed very much, when you're looking at weight as a consideration here, you should remember that there are safety mechanisms that might weigh more and also make the device safer.
It was a problem in the early days that a number of jurisdictions didn't know how to regulate this. They think about weight as being a major issue, when the real concern is how energy is transferred. A larger object moving slowly won’t have the same impact as a smaller object moving fast. Washington State and the District of Columbia both have a 100 lbs limit on robots. We would have to get that amended to operate in either of those jurisdictions.
Then on the speed limits, it depends on the jurisdiction. Each one of the devices goes an average of 3 miles an hour when it's on the sidewalk.
JF: Which is about walking speed, right?
VG: Yeah. But when we're crossing the street, we may want to go faster. When you're entering an intersection, there are emergency vehicles that are driving by, there are cyclists that might run the light. You want to be able to get out of the way of things pretty quickly or come to a stop. So, in those spaces we go faster. But broadly speaking, I think the regulations that we've started to see a national consensus around have been a traveling speed between 6 and 10 mile an hour and and then size / weight restrictions between 350 and 500 lbs limits, and no bigger or wider than a human being.
JF: And as of today, how many different cities does Serve operate in?
VG: Currently we operate in two – West Hollywood and Los Angeles. But in the past we’ve also operated in San Jose, Redwood City, Bentonville, and briefly in Vancouver. And we do have plans to expand in the near term.
Deploying Tech in Public Spaces
JF: Looking back over your experiences, do you have any advice for technologists who want to roll out technology in public spaces? Specifically, how they should interact with government in a way that’s positive and can actually get something done in a reasonable timeframe.
VG: Yeah, a major one for technologists and entrepreneurs is to not come out the gate thinking about other interest groups as adversaries. You might hear something negative from an organization and it might be founded on concerns they have with technology that preceded you.
We've seen that senior systems, disabled folks, differently abled folks, many of them haven't necessarily been given the opportunity to have input into the ways new technologies are designed or deployed in their cities. So they are very reticent to – and understandably so – support these new technologies.
Similarly, there are urban planners trying to find ways to increase density and take cars off the road. There are cyclists who want safer bike lanes. There are pedestrian advocates. And all of these interest groups want to see more livable cities. That's something we're all interested in. Viewing this as a zero sum game results in a more hostile environment for all of us, because the worst actors won’t heed community concerns at all. The goal here is to bring our communities along and show what our value is. That means doing significant outreach to those various organizations. These robots are going to live and die based on the social acceptance of them, and social acceptance requires us to be thoughtful from a design perspective, from a planning perspective, etc.
There's one last piece I want to mention, which is that none of us know what the city of the future will look like. But I do think we all have an interest in making it a space that's hospitable to humans and to be human centered. There are many areas that have become pedestrianized during the pandemic and that received a positive reception among residents. We had a slow streets program in Oakland and San Francisco, LA had one. I’ve seen them in New York, Toronto, and cities across North America. Some of those continue to exist and in each one of these instances, city planners are really trying to figure out ways to take cars off the road and to increase transit usage. And all that coincides with our interest in seeing safer and broader sidewalks, protected bike lanes, having more alternative modalities to transport goods, etc.
At Serve we often say, “It doesn't make sense to transport a 2 lb burrito in a 2 ton vehicle”. But it's a real question whether we can right-size technology for the things that we're moving around our cities. It doesn't make sense to have cars delivering very, very tiny items from place to place, adding smog to our air, and increasing friction with pedestrians. When we think about what the future looks like, none of these technologies are going to be a silver bullet. We might see more of a hub and spoke model, more right sizing for the urban landscape, and the urban landscape itself adapting to that. You can very easily imagine a world where in the outer ring of the city, you see larger frames like what we already see with freight, but as you move into the inner rings, you go from transporting things in vans to transporting things on bicycles,drones, and small sidewalk delivery robots. Maybe even handoffs between each of those. That way we aren't adding congestion to those different parts of the city and we can actually design for the future — and that requires us to go back to the city planners and advocacy groups and say, we can build this future together. We're just one small piece of the pie and this isn't about the acceptance of delivery robots on its own. This is about building better cities. We're just a part of the equation.
Afterword
After this conversation, I think deploying fundamentally new technology might be different from productizing existing business models. Rolling out AV’s for delivery has been an ongoing, gradual process. Serve – and other delivery AV firms – were never going to be able to pursue ridesharing’s political strategy of rapidly scaling their customer base and using them as a political constituency to defend the business.
The other point that’s stuck with me was that engaging with government doesn’t just mean engaging with policymakers or electeds. It includes talking to potential coalition partners who might want the same things you want, but for different reasons. It also means preemptively deescalating issues with groups that might see new tech in public spaces as threatening. And while folks building products in the consumer software space are often able to sidestep this – for better or worse – this is a process that entrepreneurs building physical products need to remember and engage with.
California Public Utilities Commission, the state agency that regulates ridesharing platforms like Lyft and Uber in California.
The San Francisco Board of Supervisors is roughly San Francisco’s equivalent to a city council. Because San Francisco is both a city and a county, the Board of Supervisors is invested with the powers of both a city council and county board.
Americans with Disabilities Act