A practical guide to the internal champion behind an AI agent project: who should own the pitch, what evidence they need, why deals stall in the middle, and how to package the project so it survives budget, procurement, and operational skepticism.