AI inference is the production-time process in which a trained model receives new inputs and computes predictions, generated tokens, actions, or representations. This guide explains the mechanism, trade-offs, evaluation, and controls that matter in practice.