Taking a closer look at aircraft flight data recorded for safety purposes, we quickly realized that it also very precisely captures individual aircraft performance in different configurations. We therefore decided to set up an in-house data science research lab to deploy predictive and prescriptive solutions, using Machine Learning performance models for each tail to optimize specific flight phases through in-flight recommendations to pilots. We started out with climb-out phase, which has the highest fuel burn, and developed OptiClimb, a unique in-flight solution that recommends customized climb speeds to pilots for each flight to save significant amounts of fuel and reduce emissions. We then extended to other flight phases with OptiFlight.