To meet the demands of integrated communication and computation design as well as network native AI, there are some challenges in current networks. These include complex network functions, poor flexibility, and low Intelligent level. In order to handle the above challenges, this talk mainly involves the agentic artificial intelligence (AI)-empowered integrated communication and computation design. First, this talk deals with the the agentic AIradio access network (RAN) architecture. The basic features and classifications are roughly provided. Typically, an agentic AI RAN assisted unmanned aerial vehicle semantic communication system is considered. In the considered system, the sense-transmit-decide-control scheme is proposed with jointly optimizing the AI inference indicator, semantic compression ratio, and UAV transmit power. Then, this talk presents the deployment of the integrated communication, sense, compute, and control system. In the integrated system, both communication-computation parallel split learning design and federated low-rank adaptation for wireless large AI model are provided. Finally, for the applications of agentic AI RAN，the talk shows the edge and cloud cooperation of quradratic robot in complex environments, as well as the agantic navigation based on intelligent sensors.