Telemetry data unlocks new use cases of OpenStreetMap by learning from human behaviour and improving the map for everyone. With hundreds of millions of sensors transmitting location updates in real time, we process massive amounts of data to derive up-to-date, continuous information on the collective activities of humans, which we then contribute back to OpenStreetMap and combine with the map to build entirely new services. This talk will explore how telemetry data can help prioritize map improvements, and bring dynamic behaviour to the OSRM navigation engine to provide realistic solutions to routing problems.