As more data becomes available through open data initiatives, opportunities to analyze and predict behavior in the real world grow exponentially. Open Data projects such as OpenStreetMaps are now more precise than their commercial equivalent, to which we can also add massive amounts of urban open data. Yet, very little intelligence has been derived from them due to the complexity of dealing with datasets at that scale. In this talk, we'll show how to use graph enabled distributed frameworks like GraphX to turn raw urban data into actionnable intelligence.