Agentic Nesting proposes a way to integrate existing enterprise applications by wrapping them as autonomous AI agents inside a layered structure.
The paper argues that enterprise systems are often fragmented across silos, while conventional integration methods such as enterprise service buses, API gateways, and robotic process automation can become tightly coupled, costly to maintain, and limited in intelligence. The proposed framework extracts a digital agent proxy from each legacy application and organizes those agents in hierarchies that mirror enterprise workflows.
The idea is not to replace every system with a new AI platform. It is to give old systems agent interfaces that can coordinate work while preserving their specialized roles. That could be useful where business processes span finance, customer service, operations, and internal knowledge tools.
The paper is methodological, so real deployment evidence will matter. Its contribution is a concrete architecture for thinking about enterprise agents as stewards of existing applications rather than as a flat swarm of disconnected bots.