From Workflow Automation to Capability Closure: A Formal Framework for Safe and Revenue-Aware Customer Service AI
This tackles a critical safety problem in customer service automation for businesses deploying AI agent networks.
The paper addresses the safety gap in customer service AI where individually safe agents can combine to achieve forbidden goals through emergent conjunctive dependencies, proposing a formal framework to ensure safety and revenue-awareness in dynamic agent networks.
Customer service automation is undergoing a structural transformation. The dominant paradigm is shifting from scripted chatbots and single-agent responders toward networks of specialised AI agents that compose capabilities dynamically across billing, service provision, payments, and fulfilment. This shift introduces a safety gap that no current platform has closed: two agents individually verified as safe can, when combined, reach a forbidden goal through an emergent conjunctive dependency that neither possesses alone.