The integration of Large Language Models (LLMs) with Xml query languages like XPath and Xquery has led to notable advancements in element retrieval and data extraction and has greatly improved how we find and get data from documents. multiple algorithms and tools have been introduced among them there is VON Similo, an algorithm that employs a multi-locator approach, utilizing properties such as ID, XPath, label, and tag to accurately identify web elements. This algorithm is the result of a significant development and the used method enhances the precision of web element localization, especially in dynamic web environments. Additionally, the application of Retrieval-Augmented Generation (RAG) techniques helps fix problems where LLMs might generate incorrect or unrelated information and addresses the hallucination issues by retrieving relevant information from a set of documents in order to leverage the generation process. This approach ensures that the outputs of LLMs models is contextualized and more accurate especially for XML retrieval tasks based on XPath or XQuery expressions. The combination of LLMs with XML querying language expression can help improve the efficiency and accuracy of web data extraction and interaction tasks based on the recent demonstrated advances in this area.