Yes, works great for anything that is amenable to reinforcement learning. For other domains, not so much.