We have LLMs now. Why are the docs still wrong? (amendary.com)

In a recent discourse on documentation challenges in software development, the author highlights the persistent issues with keeping technical documentation accurate and up-to-date despite the availability of large language models (LLMs). While LLMs can efficiently generate initial drafts for documentation based on code changes, ensuring that the correct sections are updated remains a complex problem. The risk of overwriting critical context or introducing inaccuracies underscores the importance of understanding the relationship between documentation and the specific code changes as well as the needs of different audiences. To address this challenge, the author introduces Amendary, a tool designed to integrate with repositories like GitHub and documentation platforms such as Notion. Amendary monitors code changes, identifies discrepancies in documented information, and drafts corrections for review without automatically implementing them. This approach aims to streamline the documentation process by ensuring that changes are executed with proper context and are subject to review before integration. Through this workflow, the author hopes to reduce the time spent on maintaining documentation while enhancing its relevance and accuracy, ultimately improving the overall user experience for those relying on these guides.