San José: Using AI-driven traffic signals to make buses 20% faster without new infrastructure

The problem

San José is the third most populous city in California, home to nearly one million people spread across 182 square miles.

For residents who depend on public transport, reliability matters as much as frequency. A bus that is scheduled to arrive every fifteen minutes but regularly runs late does not feel like a fifteen-minute service. It feels unpredictable. And when buses are unreliable, people with the option to drive tend to do so, which increases congestion and undermines the case for investing in public transport.

One of the most common reasons buses fall behind schedule is also one of the simplest: red lights. A bus travelling along a busy corridor can be stopped at intersection after intersection, each red light adding seconds or minutes to the journey. Over the course of a route, those delays add up. The bus arrives late, passengers wait longer, and the knock-on effects ripple through the rest of the schedule.

For city leaders, improving bus performance and public transit was a priority, with Mayor Mahan describing faster, more reliable buses as central to reducing congestion, improving air quality, and making transit “a more attractive alternative” to driving. Delivering on that ambition required finding ways to make bus services faster and more reliable without the cost and disruption of major infrastructure projects.

The solution

The city found an answer not by changing the buses or the routes, but by changing the traffic lights.

Working with LYT, a traffic technology company, San José introduced a system called transit signal prioritisation. The concept is straightforward: when a bus approaches an intersection, the traffic signal knows it is coming and adjusts to give it a green light, reducing the time the bus spends waiting at red lights.

The system works through a transponder fitted to each bus. The transponder communicates with the city's traffic signal network, which already knows where the bus should be based on its published schedule. LYT's machine learning software brings these pieces together: rather than following a fixed rule, it draws on real-time traffic data to determine which signal adjustment will move the bus through with the fewest unnecessary stops, and how to implement it. This is the part the AI performs: deciding, case by case, when and how to adjust each signal as conditions change.

As Stephen Caines, San José's Chief Innovation Officer and Budget Director, explained: “The traffic light already knows where the bus should be, based on the schedule. LYT runs the software that brings all these pieces together, and they are the ones who are effectively optimising the traffic signalling through their technology.”

The concept is similar to the signal pre-emption technology used by emergency vehicles, where traffic lights turn green to clear the way for an ambulance or fire engine. The difference is that transit signal prioritisation does not entirely override the signal. It makes a smaller, targeted adjustment as the bus approaches, extending a green light by a few seconds or shortening a red. Therefore, the disruption to other traffic is minimal.

The city began with a pilot in 2023, deploying the system along two bus routes. The results were strong enough to justify expansion. By early 2026, the system was operating across all 24 bus routes in the city, run by the Valley Transportation Authority, the transit provider for San José and surrounding communities across Silicon Valley.

Crucially, the project did not require closing roads, building new infrastructure, or making major changes to the bus network. As Caines described it: “This was the least-friction way of enhancing our transit system. It's modular, it's scalable. We didn't have to shut any roads down to introduce this technology.”

State and federal grants covered 90% of the project cost. As Caines noted: “If you are a city that wants to place an ambitious bet on technology, you're not just limited to what you currently have to offer. There's really innovative ways you can fund these projects.”

The impact

1. Bus speeds increased by 20% across all 24 routes
Since the citywide rollout, transit buses in San José are travelling 20% faster across all 24 routes. The improvement comes entirely from spending less time waiting at intersections. No changes were made to the buses, the routes, or the frequency of service. For passengers, 20% faster journeys translate directly into shorter travel times and more predictable arrivals. For a city trying to make public transport a genuine alternative to driving, speed and reliability are what determine whether people choose the bus or get in the car. As Mayor Matt Mahan described it, the system is “saving our commuters and working families time and proving that local government can deliver results where it matters most.”

2. Red-light waiting times fell by 50% during the pilot
During the 2023 pilot on two routes, the system reduced the time buses spent waiting at red lights by 50%. Red lights are among the most consistent and cumulative causes of delays in urban bus networks. A bus stopped at every other intersection along a busy corridor can lose minutes per journey, and those minutes compound across an entire day's schedule. Halving that delay on the pilot routes kept more buses running on time and provided the evidence base for expanding across the city.

3. Buses are now arriving ahead of schedule
The only operational challenge the city has encountered is that some buses are now running faster than their published timetables. For a public transit system where the normal problem is buses running late, arriving early is a sign of how significant the improvement has been. It also creates a practical issue: buses that arrive ahead of schedule can leave stops before passengers expecting to catch them arrive. The city and Valley Transportation Authority will need to adjust timetables to reflect the new, faster running times. As LYT CEO Tim Menard noted, "Transit signal priority isn't a spot treatment technology. It thrives on scale and, like compound interest, so do its benefits."

4. Real-time data gives the city visibility it did not have before
The signal system does more than speed up buses. It provides city officials with a live, route-by-route picture of how the bus network is performing at any given moment. If a bus is falling behind schedule, staff can see the problem as it develops rather than learning about it through passenger complaints. That early visibility allows the city to distinguish between technology issues, road conditions, and staffing problems and respond accordingly. In a related AI pilot, San José found that the city could identify nearly 70% of road maintenance issues before residents submitted a service request, demonstrating the broader value of real-time monitoring across city services.

Lessons learned

• The intervention targeted the actual source of delay. San José did not redesign its bus network, build dedicated lanes or add vehicles. It found that red lights were among the most consistent causes of buses falling behind schedule and addressed that specific problem. The 20% speed improvement across all routes came from changing when traffic lights turn green, not from changing anything about the buses or the roads they travel on. For transport teams considering where to invest, understanding where delays actually accumulate can point to more targeted, less expensive interventions than system-wide redesigns.

• The technology works with existing infrastructure. Transit signal prioritisation uses the traffic signals a city already has and adds a transponder to each bus. No roads were closed, no new lanes were built, and the bus network continued operating throughout the rollout. The system was added route by route without disrupting existing services. For cities with limited capacity for major transport infrastructure projects, this approach offers a way to improve bus performance using what is already in place.

• The benefits grew as the system scaled. The pilot on two routes produced a 50% reduction in red-light delays. The citywide rollout across all 24 routes produced a 20% increase in overall bus speeds. Each additional route benefits from the same signal adjustments, and the network-wide effect increases.

• The system generates operational intelligence, not just speed. Real-time data from the signal network gives the city continuous visibility over bus performance across all routes. That data helps identify problems before passengers report them and distinguishes among different causes of delay. For transport authorities that currently rely on passenger complaints or periodic reviews to understand how services are performing, this kind of live monitoring changes how quickly and precisely problems can be addressed.

This case study was written with assistance from artificial intelligence.