📄 Abstract
Span-of-control theory has, since Graicunas's 1933/1937 formalisation of how fast superior-subordinate interactions multiply, treated the number of people a manager can effectively supervise as bounded chiefly by the manager's own cognitive and communicative capacity. The deployment of autonomous and semi-autonomous AI agents into team workflows unsettles that boundary in a way earlier information technologies did not: agents do not merely give a manager better information about subordinates, they absorb a share of the routine coordination, monitoring and exception-triage work that previously consumed the manager's attention, functioning in effect as coordination-compressing capital. Industry evidence assembled since 2024 suggests this is already reshaping organisational structure: Gartner has predicted that by 2026 a fifth of organisations will use AI to flatten structure and eliminate over half of current middle-management positions, a 2025 MIT Sloan Management Review/BCG study found 45% of respondents at organisations with extensive agentic-AI adoption expect further reductions in management layers, and Amazon's 2024 mandate to raise its individual-contributor-to-manager ratio by at least 15% is widely read, alongside this evidence, as an early instance of the same structural shift. This paper reviews classical span-of-control theory, the organisational economics of hierarchy, and this emerging AI-agent evidence to develop a conceptual framework in which AI-agent deployment density raises optimal managerial span of control primarily by reducing the coordination cost per subordinate, an effect strengthened for routine, codifiable task portfolios and attenuated where work remains exception-heavy and judgment-intensive. Five formal propositions link deployment density, task routineness, and managerial role adaptation to realised — as distinct from merely announced — span-of-control change, drawn from a structured review of thirty-six cited works. Particular attention is given to India's IT-BPM sector, whose historically wide, standardised-process spans of control interact distinctively with agentic-AI deployment. The paper closes with a research agenda centred on a measurement gap current scholarship has not yet filled: no validated instrument yet exists for "AI-agent deployment density" itself.
🏷️ Keywords
📚 How to Cite:
Dr. N Subbukrishna Sastry, Dr. Manjula Mallya M , AI-AGENT DEPLOYMENT DENSITY AND THE RESHAPING OF MIDDLE-MANAGEMENT SPAN OF CONTROL , Volume 14 , Issue 9, September 2026, International Journal of Asian Economic Light (JAEL) , Pages: 11 - 28 ,