It looks very interesting. What do you feel are the main take-aways with regard to reinforcement learning?