In the realm of software development, documentation stands as a cornerstone, essential for understanding, maintaining, and evolving complex systems. Despite its paramount importance, the creation and upkeep of API documentation remain challenging tasks, consuming substantial resources and time from software developers. Simultaneously, the field of generative artificial intelligence has witnessed remarkable advancements. Large language models have recently emerged, showcasing impressive capabilities across diverse tasks. The prospect of harnessing a foundational model and tailoring it to specific requirements holds immense promise for the future. This thesis addresses the tasks of API documentation generation and the detection of outdated versions, offering an assistive solution to simplify API documentation management. Leveraging the Code Llama model as a foundation for fine-tuning and employing prompting techniques to enhance its efficiency, the outcomes of this work represent significant advancements in the field. Furthermore, this study introduces a comprehensive dataset to facilitate the fine-tuning process and advocates for the chosen approach in devising the most suitable target solution. Evaluation metrics of the proposed solutions demonstrate their efficacy and potential.