Globalbit is an Israeli custom software development company. This page is a case study about Confidential — Israeli Traffic-Tech Company (Smart City & Urban Mobility). Globalbit built an AI system for an Israeli traffic-tech company that reads traffic from existing cameras and selects, in real time, the best approved signal plan for each intersection. The system was deployed for a major Israeli city. Key facts: 25% less waiting time at intersections, up to 30% fewer bus delays, 99% vehicle detection accuracy, detection up to 700 meters from the intersection, full rollout in ~3 months.
We Taught a City's Traffic Lights to See
Less waiting time at intersections
Fewer bus delays across the city
Vehicle detection accuracy
AI that manages traffic signals in real time, without digging up a single road
Globalbit built an AI system for an Israeli traffic-tech company, deployed for a major Israeli city, that watches intersections through the city's existing cameras and updates traffic signal plans in real time. The system detects and classifies vehicles up to 700 meters from the intersection, with 99% accuracy, then selects the best approved plan from 10 existing signal programs. Before it was allowed to control live traffic lights, it ran for about a month in shadow mode and compared its decisions against the fixed timing. The result: 25% less waiting time at intersections, up to 30% fewer bus delays, and more than 20% better traffic flow.
Bringing AI into live city infrastructure without stopping the city
A traffic light looks like a small system from the outside. In practice, every decision affects safety, public transit, congestion, residents, regulation, and public trust.
- -Fixed-time traffic lights cannot see the road — The same plans run during rush hour and at midnight. An empty street can get a green light while a full queue of cars waits on the other side.
- -Traffic is political — Every wasted minute is felt by residents, drivers, and public transit passengers. In a large city, a small improvement at a major intersection quickly becomes visible on the ground.
- -There was no appetite for infrastructure work — Loop sensors mean opening roads, closing lanes, and months of disruption. The city wanted a solution that worked with the cameras and traffic controllers it already had.
- -There was no room for error — A wrong traffic light decision is a safety event, not a screen bug. Before live control, the system had to prove it made better decisions than the fixed plans.
The city was not looking for a smart city slide deck. It wanted unnecessary red lights to disappear.
First we proved it in the data. Only then did we give the AI control
An AI system that understands intersections, regulation, and public transit
The hard part was not only detecting vehicles. The important part was making decisions that can run in a real city, on existing infrastructure, with traffic engineers who need to understand every change.
Planning an AI project with real-world accountability?
We built AI that controls live city infrastructure, a border control system for COVID restrictions in 90 days, and trading apps where every second matters. If your project carries real risk, we know how to build a system around it that holds up.
~3 months from one-intersection pilot to city rollout
A production AI stack for existing city infrastructure
The system was built around one main constraint: work in the field, on the equipment already in place, with full visibility for the traffic engineering team.
Why an Israeli traffic-tech company trusted Globalbit with the AI at the core of its product
A project like this needs more than a good algorithm. It takes transportation systems experience, comfort working within regulation, and software that can run when dangerous trial and error is not an option.
Less waiting, fewer bus delays, better flow
The system proved measurable improvement in the field using existing infrastructure, with no roadworks and no new sensors.
