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

The initial vision for AI agents was to automate tasks and liberate human workers. However, businesses in Asia are finding that integrating these agents creates unexpected challenges, requiring more human oversight and management. This reframes the role of AI from a purely autonomous solution to one that needs active human direction.

The reported experience of "Xiao" illustrates this friction. Her agent and its supervisor's agent engaged in a bureaucratic debate over responsibilities and task ownership, mirroring the very human-centric inefficiencies they were meant to replace. This suggests that simply deploying AI agents without addressing underlying process complexities can replicate existing problems.

This situation highlights a critical distinction between AI's potential for automation and its current reality. While agents can perform tasks, their integration into complex workflows introduces new forms of communication and coordination challenges. The "communication overhead" cited suggests that the intelligence of these agents is not yet sufficient to navigate ambiguous roles or self-resolve conflicts within an organisational structure without human intervention.