Janine Cavicchia is retired. She lives in Macomb, Illinois, population 14,000. She is personally responsible for 62% of all Lyft rides taken in her city.

She knows which passengers need to catch the 6:20 train. She once drove four hours to Chicago so a regular rider could make a medical appointment. When she took a two-week vacation, one of her regulars lost his job due to lack of transportation.

Janine is extraordinary. She is also, in the most literal sense, a single point of failure.

This is the problem I keep coming back to. Not whether autonomous vehicles will work, but rather what's the model that actually gets us to the future fastest, reaches the most people, and builds something that lasts? My answer is the hybrid network, and Janine is exactly why.

The autonomous vehicle era is arriving in cities across the US where dense urban cores, predictable road networks, and favorable regulators create the conditions for new technology to prove itself at scale.

But most of America isn't on any AV's map yet.

Eight percent of American households own no vehicle at all, and a third have only one. In rural counties, the nearest hospital can be forty miles away with no option to get there except a car or a favor. The nurse on the 6am shift lives in a neighborhood no AV is cleared to operate. The rider who needs someone to walk them to the door isn't asking because the technology failed; human presence is sometimes the point. These tens of millions of Americans are not edge cases and the pure-AV model, as currently deployed, cannot reach them.

Autonomous vehicles and human drivers complement each other. AVs excel at the trips they're built for: geofenced corridors, repeatable routes, mapped environments. Human drivers excel everywhere else: the trip in a neighborhood that isn't mapped, the route that requires local knowledge no algorithm has captured, the moment a rider simply needs another human. A hybrid network deploys AVs at scale where they perform best, while relying on human drivers to fill gaps the AV supply and geofence can't cover. The result is a system more capable than either alone, reaching more people, more places, more of the time. A hybrid network gets AV technology in front of more riders faster, accelerating the trust-building that the entire industry depends on and enabling a transition that will take decades.

Consider what demand actually looks like at its most dynamic: a sold-out concert, a sudden storm, a rider who needs a hand getting to the door. These moments require flexible, responsive supply. An AV-only network is structurally ill-suited to them: almost always either under- or over-supplied, producing high costs and low utilization, or simply no ride when it matters most. A hybrid network solves this by design. Human drivers flex up when conditions shift. AVs anchor the baseline. Together, they match supply to demand in real time, maximizing utilization while making sure riders always get a fast and reliable ride.

Outside the urban core, rideshare has long faced a self-defeating cycle. Fewer drivers means longer waits and higher prices, so riders stop trying. Demand erodes, drivers stop showing up and the market never gets dense enough to sustain itself. An AV baseline breaks that doom loop. When autonomous vehicles guarantee consistent supply, riders begin to experience rideshare as reliable rather than aspirational. They ride more, some give up a second car (or even a first), and demand grows. Suddenly the economics work for human drivers too: there is now a market worth serving.

The AV doesn't replace the driver. It creates the conditions for drivers like Janine to thrive. 

Transit agencies have wrestled with suburban and rural coverage for decades. Fixed-route service requires density to justify its cost, and lower-density communities rarely clear that bar. Lyft has been working with transit partners to help, offering first- and last-mile connections, late-night service after buses stop running and on-demand options in lower-density suburbs. But even those partnerships can hit a ceiling: you can't reliably serve a market without drivers willing to drive. AVs remove that constraint. A vehicle that just runs, regardless of shift economics, regardless of whether it's 2am on a Tuesday, gives transit agencies what they have always needed: guaranteed baseline coverage in markets too sparse to sustain it any other way. The hybrid network turns a decades-old transportation policy problem into a solvable one.

Janine doesn't need to be replaced. She needs a foundation under her. An AV baseline doesn't displace her important skillset: local knowledge, human judgment, genuine care. It creates the infrastructure that makes those qualities the exceptional layer they always should have been, instead of the only thing standing between her neighbors and no transportation at all. No community's access to mobility should rest on one person's willingness to drive four hours to Chicago on a weekday. Janine is remarkable. The system she's propping up is broken.

Making AVs work at commercial scale requires more than a vehicle. It requires depot operations, charging infrastructure, fleet management, and the data from millions of real urban miles that tells you where the edge cases live before a passenger encounters one. Lyft has been building that infrastructure, quietly and relentlessly, longer than the current wave of AV headlines might suggest. Through Flexdrive, our independently managed subsidiary, we operate roughly 15,000 vehicles daily across dozens of markets. Our Nashville facility is the first Flexdrive location purpose-built for AV fleet operations, a proof point for what the hybrid future actually looks like in practice. And we have the network: hundreds of cities, rider relationships, payment rails, matching systems, and demand patterns that took years to build. That network is what gives an AV fleet somewhere to go. Real routes. Real riders. Ready to scale.

Our upcoming deployment in Nashville is the earliest version of what this looks like in practice. Our partner brings extraordinary technology. Lyft brings the network and the operational infrastructure. Together, we're building something neither could build as well alone. That's the thesis: the AV transition works through partnership between technology and network. Not by declaring one obsolete.

I previously wrote that autonomous vehicles may not need a driver, but they still need something deeply human to move forward: trust. I still believe that. But I'd add one more thing they need: a network. Not just an app and a geofence. A real network, one that reaches the rider who can't afford to be one minute late for work, the patient who needs dialysis three times a week, the person living in a community no AV company has mapped yet.

That is what the hybrid network delivers: autonomous vehicles providing the baseline, human drivers providing the reach, together covering the distance. The partnerships that define the next decade of transportation will be the ones that resist the temptation to declare victory only inside a geofence and at optimal times. The ones committed to reaching everyone.

That is the network worth building. And we are building it.

Jerry Golden is the Chief Policy Officer at Lyft.