## LLMs and Software Development: A One-Year Analysis The integration of large language models (LLMs) into the developer workflow is becoming increasingly common. A recent article explored the benefits and challenges of using these tools over the past year. ## Strengths and Weaknesses Chatbots have proven to be valuable allies in exploring existing code and checking for regressions during the development of new features. Codex excels at debugging, while Claude stands out for the quality of the code produced. Gemini, on the other hand, seems to have difficulty following complex instructions. The highest-performing open-source models are approaching the capabilities of proprietary models, but local execution remains a problem due to high hardware requirements. As a result, many developers rely on third-party services, raising concerns about the security of company code. ## Implications for the Future The ease with which LLMs generate code makes it even more important to focus on thorough testing, code readability, and accurate documentation. The role of developers is evolving: from code generation to checking and supervising the work done by LLMs. It is essential to maintain and develop one's skills to avoid becoming overly dependent on these tools. The use of LLMs allows for a faster approach to a wide range of problems, accelerating the development cycle. The ability to quickly understand code is one of the main advantages offered by LLMs, especially when working with complex and context-rich projects.