You have a reinforcement learning system? Sure, it works great, but does it give you any guarantees? I thought not. Maybe synthesis can help. Come listen to my talk.