Evaluating the effectiveness of LLM-based interoperability