This paper presents a comprehensive analysis of large language models (LLMs) capabilities in the machine tool domain, specifically focusing on face milling operations. The research evaluates how different prompt techniques (zero-shot, few-shot, and tree-of-thought) affect LLMs' ability to perform tasks traditionally requiring human domain expertise, such as interpreting G-code, recommending appropriate cutting tools, and calculating machining parameters. Performance is evaluated by comparing LLM outputs to industry-standard CAM software and digital twin simulations to verify practical applicability. The findings indicate that current LLM technology shows promise for transforming and optimizing complex engineering tasks in manufacturing but still requires additional operator input and customized approaches to achieve complete operational accuracy. This work contributes to understanding how generative Artificial Intelligence (AI) can be leveraged to optimize, generalize, and standardize machining procedures in industrial applications.