TfL says 3 minutes. It arrives in 7. We catch ghost buses before they vanish. BusBrain uses machine learning trained on millions of journeys to tell you when your bus actually arrives.
We spot buses that TfL says are coming but will never arrive. No more standing in the rain for a phantom 205.
Every prediction comes with a confidence percentage. 95% means run. 40% means maybe get a coffee first.
Real-time service alerts, curtailments, and diversions. Know before you walk to the stop.
Rain slows buses. We factor in live weather data so our predictions adapt to conditions on the ground.
A ghost bus is a vehicle that TfL's countdown system says is approaching your stop, but which will never actually arrive. It might have been curtailed mid-route, broken down, or gone out of service.
BusBrain tracks every vehicle's trajectory and flags predictions we believe are unreliable — so you can make a real decision about whether to wait or walk.
Wherever there's a bus, there's a prediction gap. We're closing it.
GPS pings every 30s from every bus. Weather. Traffic. Roadworks. Events. School terms. Time of day. We eat data for breakfast.
Our LSTM network learned from 2B+ data points. It sees how rain at 8:07am on a Thursday cascades through routes across South London.
Every 30 seconds we compare prediction to reality. The model gets smarter with every bus that passes every stop, every day.
We're opening the API to developers first. Build with predictions that actually work.
Request API Access