Request, AI, and processing limits