The author was initially skeptical about vibe coding and only used LLMs for a small portion of the code, but found them helpful for bug fixes, new domains, and quick answers.
When building a Phoenician transliteration tool in ReactJS without prior ReactJS knowledge, the author found success using ChatGPT for crafting prompts and generating code, but faced challenges during deployment.
Issues arose during website deployment as ChatGPT and Cursor couldn't provide a complete list of libraries, leading to manual library installations and debugging.
Despite successfully deploying the site, the author realized the importance of investing time in learning React properly to avoid debugging and improve code structure, highlighting the need to treat LLMs like a Lego toolkit for code generation.