From Features to Actions: Explainability in Traditional and Agentic AI Systems Paper • 2602.06841 • Published Feb 6 • 1