This project aims to evaluate and compare different Large Language Models (LLMs) for the task of extracting email signature information and structuring it into a JSON format. The evaluated models are OpenAI GPT-3.5 turbo and Anthropic Claude 3. The project includes prompt engineering, testing, and iteration to achieve the best results.