Agentic Nesting: A New Methodology for Enterprise Application Integration and Services
Xi Wang, Kun Li, Xianyao Ling, Gang Yin, Liang Zhang, Jiang Wu, Wenbo Lei, Jun Xu, Annie Wang, Fu Zhang, Weizhe Wang
arXiv:2608.05159 [cs.AI] (Submitted on 25 May 2026)
Abstract
Enterprise operations are heavily dependent on a diverse array of heterogeneous business systems and information applications, leading to significant data silos and process fragmentation. Although substantial financial and material investments have been made in building these applications, effectively orchestrating and leveraging them remains a formidable challenge. Conventional approaches to enterprise application integration—including middleware architectures like the Enterprise Service Bus (ESB), API gateway infrastructures, and Robotic Process Automation (RPA)—face inherent limitations such as high architectural coupling, escalating operational and maintenance costs, and constrained intelligence capabilities.
This paper introduces Agentic Nesting, a multi-agent collaboration framework that encapsulates existing enterprise applications as autonomous AI agents within a hierarchically nested structure. Unlike flat interconnection models, this approach organizes agents into layered stewardship topologies that mirror the compositional complexity of real-world enterprise ecosystems. The framework extracts a digital agent proxy from each legacy application, enabling natural-language interaction and autonomous manipulation. A central orchestrator coordinates these agents for task decomposition and dynamic dispatching, while a unified conversational interface facilitates cross-application querying and process orchestration.
Key contributions of this research include: (1) the introduction of the "Application-as-Agent" integration paradigm, (2) the articulation of the "Conversation-as-Integration" interaction philosophy, and (3) an exploration of the methodology's generalization potential in scenarios involving heterogeneous system coordination and large-scale data applications. As enterprises increasingly adopt generative AI and agentic architectures in 2026, this framework offers a pragmatic bridge between legacy infrastructure and next-generation intelligent automation.
via ArXiv AI