The AI promise was automation. The reality? Managing digital workers

The initial vision for AI agents was to deliver fully autonomous digital workers that would eliminate human drudgery. However, the operational reality in Asian businesses suggests a different outcome, with AI implementation leading to increased management overhead and new forms of workplace friction.

One reported instance describes two AI agents, one for an employee and one for a supervisor, locking into a lengthy debate over task ownership and verification. This scenario, as described by Xiao, became “very formal and cumbersome,” creating communication overhead that mirrored the human bureaucracy they were intended to replace.

This experience indicates that the integration of AI workforces is not a simple delegation of tasks but requires active supervision and management, similar to human teams. The complexity arises from agents needing clear directives and boundaries, which, if not properly defined, can lead to inefficiencies.

The challenge for businesses now lies in effectively supervising these AI workforces to prevent them from generating new forms of bureaucracy. This involves understanding how to structure tasks and responsibilities for AI agents to avoid unintended friction and ensure they genuinely contribute to productivity.