South Africa has some of the worst traffic congestion and road safety in the world. In Cape Town, named ninth most congested cities by the 2024 INRIX Global Traffic Scorecard, drivers lost an average of 94 hours to gridlock in 2023.

The City of Cape Town delegates the management of its traffic systems to a company called Syntell, which has dominated municipal traffic services across six provinces, and almost all the metropoles since it struck a cushy deal with the ANC back in 2006, which has been honoured and even expanded by subsequent local and provincial governments.

Through them, Cape Town runs around 1,750 robot-controlled intersections using fixed timing schedules, road sensors, or the SCOOT system (Split, Cycle, and Offset Optimisation Technique), which adjusts timings based on live traffic data and covers about 25% of the city’s robots.

Much of this is anually run through an office with a small central traffic monitoring centres, where employees manually intervene in the traffic flow by controlling the robots. This sort of intervention makes systematic examination of traffic flow difficult to assess – automated controls allow one to systematically assess the effects of various interventions, but such ad hoc interventions are not as high quality data.

But new AI technologies from companies like Google and Alibaba are addressing urban congestion by optimizing traffic light timings, with significant global impact.

Councillor Rob Quintas, from the urban mobility committee, notes that while the city invests heavily in traffic management, it has not yet found an AI system promising substantial benefits, despite awareness of tools like Google’s Green Light.

A February 2025 Nature study showed big-data-driven systems in China’s 100 most congested cities reduced peak-hour trip times by 11% and off-peak by 8%, yielding $32 billion in benefits (time savings, fuel efficiency, CO2 reduction) against $1.5 billion in costs. Green Light, active at 70 intersections globally, reports 30% fewer stops and 10% lower emissions. Cape Town remains open to adopting such technology if a suitable solution emerges but has no concrete AI plans yet.

However, this would mean a tweak in how Syntell does business.

Their services include “smart roadblocks” (automatic number plate recognition), speed camera systems, fixed camera management systems, support vehicles (towing companies), database tracking of fines and infractions, contracts to manage the remote issuing of summons and warrants and chase up fines, and automated online payment services through payCity.

As the Syntell website informs, the company also runs the traffic light systems, CCTV cameras and digital license plate capturing technology for not only most major metropoles, but warehouses, housing complexes, mines and storage depots around the country, which they have central access to under a national database, which they sell to companies overseas in Canada and Belgium for research purposes.

But their business model on that end is strange, since they also run traffic monitoring centres, where employees manually intervene in the traffic flow by controlling the robots. This sort of intervention is ad hoc, and prone to errors of human judgment - there is no industrial or guild standard for this sort of skill.

With AI-driven traffic management, Syntell would have to refit and centrally integrate their entire system from scratch. Perhaps they are already doing so, and the City is providing free publicity for them - after all, they have ambitions to enter the foreign market, and already make a considerable amount of money selling traffic data to foreign states (POPI compliant?).

But regardless of the appropriateness of the inner workings, this modernisation could well be a considerable improvement.

‍