In recent years, large language models (LLMs) have gained significant popularity, driving innovation across various domains. LLMs are widely used across fields, including general-purpose robotics. However, their use in industrial applications still has a lot of unexpressed potential. This paper presents a prototypical implementation of a collaborative robot system which uses a set of carefully instructed LLMs to plan a collaborative assembly/disassembly process. Using natural language descriptions, the system generates action plans, assigning tasks to robots or humans based on capabilities and tools. Central to this implementation are state-of-the-art object detection and a reasoning process to deduce action plans. In addition, we implemented a monitoring system employing sequential LLMs to check for and correct errors during task execution. We detail the integration of system components, including peripherals and prompt engineering methods to optimize task outputs. To validate the approach, we demonstrate its application on the assembly and disassembly of a Raspberry Pi-based product. This work explores how LLMs enhance human-robot collaboration (HRC) in industrial assembly and disassembly, improving task planning, real-time error detection, and adaptive correction.