Artificial intelligence (AI) adoption in small and medium-sized enterprises (SMEs) is characterised by high initiative failure rates, misaligned resource allocation, and systematically underdeveloped governance. A primary driver of these failures is the absence of a structured method for classifying, comparing, and prioritising AI initiatives before investment is committed. This paper proposes the Four Quadrants of AI Value — a two-axis classification framework that positions AI initiatives along the dimensions of value beneficiary (Internal vs. External) and AI autonomy (Assists vs. Acts). Combined with a six-level AI Capability Maturity Model and a five-criterion Prioritisation Scorecard, the framework provides operations managers and industrial engineers with a practical, evidence-based instrument for AI initiative selection. The paper describes the theoretical foundations of the framework, details the classification logic and scoring methodology, and illustrates application through a multi-initiative case study drawn from SME operations contexts. Results demonstrate that framework-guided selection significantly reduces the incidence of premature high-autonomy deployment and improves governance calibration relative to unstructured approaches. Implications for operations management, decision support systems design, and responsible AI governance are discussed.