Imagine a world where the complex, time-consuming task of creating tender documents is automated. No more manual drafting, fewer errors, and increased consistency—that's the promise of new research leveraging Large Language Models (LLMs). Tender documents are the backbone of procurement, providing crucial guidelines for acquiring goods and services. Historically, creating these documents has been a laborious manual process, prone to errors and inconsistencies, especially challenging for those without specialized procurement expertise. LLMs offer a powerful solution by automating the generation of these documents. However, off-the-shelf LLMs lack the specific knowledge needed in procurement. This research introduces a novel framework that combines LLMs with retrieval-augmented generation. This means the LLM learns from a database of existing tender documents, policies, and project information. When a new procurement need arises, the system retrieves the most similar historical documents and uses this information to generate a tailored template. The system also verifies the coherence between retrieved documents, policies, and new requirements, modifying the template accordingly. A key innovation is the use of “smart tags” within document templates. These tags act as placeholders for key information, and the LLM acts as an agent, interacting with users to gather the necessary details and fill in the tags. The system also leverages a knowledge base of procurement items to refine the generated document, ensuring it aligns precisely with user requirements. The results of this research are promising. In experiments, the AI-generated documents achieved high scores for semantic accuracy and content relevance. Compared to using a standard LLM without the retrieval and modification modules, the new framework produces significantly better results. While template-driven LLMs often produce more coherent documents, they can still lack the specific style and format of standard procurement files. This research provides a significant step toward fully automating tender document creation. Future research will focus on expanding the tender document database and improving the system's ability to handle complex table formatting and content generation. This framework not only promises increased efficiency and accuracy in procurement but also allows non-experts to easily create professional, compliant tender documents, opening up new opportunities for streamlining the entire procurement process.