Leadership as a Service? Delegating Leadership Functions to Agentic AI—Boundaries, Risks and Accountabilities

Leadership is increasingly delivered, in part, by software. Agentic artificial intelligence (AI) systems now monitor team activity, coordinate work across people and tools, schedule and assign tasks, deliver feedback and performance nudges, and broker information between leaders and followers — functions that organisations once obtained only from human managers. This has prompted the suggestion that leadership itself is becoming a service: a set of functions that can be specified, subscribed to and supplied by systems rather than embodied in persons. This conceptual paper examines that suggestion critically. Drawing on functional leadership theory, the automation–augmentation paradox, and the literature on algorithmic management and trust in AI, I develop a function-by-delegability framework that classifies five leadership functions — sensing and monitoring, coordination, deciding and allocating, developing and giving feedback, and representing and meaning-making — according to how far each can responsibly be delegated to agentic systems. I argue that delegability is governed less by technical capability than by accountability: I introduce the accountability floor, the principle that functions whose delegation would break the chain of accountability may be automated in execution but never in ownership. I identify four risks of delegation — legitimacy and follower trust, bias at scale, accountability gaps, and erosion of the leadership pipeline — propose governance and design principles, and derive three propositions for empirical research.