Taking you to the card → The caveat: these proofs fit the reinforcement learning paradigm