The growing complexity of manufacturing systems calls for innovative approaches like biologicalisation (application of biological principles to industrial process design). In this context and building on our previous work on altruistic collaborative healing, where manufacturing elements (“Donors”) share resources with those in need (“Recipients”), this paper addresses key research challenges by integrating Large Language Models (LLMs). By enabling natural language communication between altruistic and recipient resources, LLMs enhance reasoning, coordination, and adaptability in dynamic manufacturing environments. To demonstrate this concept, we propose an LLM-driven framework for decision-making and resource allocation, validated through a peer-to-peer energy-sharing scenario involving Automated Guided Vehicles (AGVs). Using the gpt-4o-mini model in simulation, the results show improved system adaptability and 80% reliability. The paper concludes by discussing the implications, limitations, and future directions of this approach, emphasizing the potential of LLMs in smart manufacturing.