This chapter explores the transformative role of Large Language Models (LLMs) and intelligent agents in optimizing resource allocation for customers in beyond-5G (B5G) network environments. We examine how generative AI technologies enable dynamic orchestration of network resources through predictive modeling, contextual understanding, and autonomous decision-making capabilities. The chapter delves into self-healing network architectures that leverage LLM-driven agents to detect anomalies, anticipate failures, and automatically reconfigure network topology to maintain service continuity. Special attention is given to real-time resource redistribution techniques, multi-agent coordination for complex problem-solving, and the integration of these systems with existing network management frameworks like SDN and NFV. We conclude by discussing the significant challenges, including model reliability, security, and real-time performance, while outlining future research directions toward fully autonomous, intent-driven network ecosystems.