Vibe Coding is not a universal key Vibe Coding has been very popular recently, and I am also studying it myself. As a CS PhD student and former code farmer/CTO, let me share my views Suitable scenarios for Vibe Coding - Demo -Simple data acquisition and kanban -Products with low requirements for safety/maintainability/performance/accuracy Scenarios where Vibe Coding is not suitable -Commercialized products -Products with high requirements for safety, maintainability, performance, and accuracy The issue with Vibe Coding -Low security There is a big problem with the illusion of big language models now, and scaling laws are also ineffective. The upper limit of Vibe Coding is the upper limit of the ability of big language models, and the code written may cause problems -Poor maintainability There is a problem that requires a manual team to modify, and the code generated by AI is difficult to maintain -Poor code quality At present, the focus of Vibe coding is on writing code, rather than system software engineering; To give an analogy, it's like a paper house that's built quickly but can collapse at any moment, while an excellent software engineer builds a reinforced concrete building At present, Vibing Coding seems to be a great personal assistant and an upgraded version of chatgpt, but it cannot be used for commercial software and is difficult to support the needs of one person one company, especially for software service companies; I will personally continue to learn and share Vibe Coding, helping with personal investment research and content generation for web3, but I am not proficient in developing core modules for the company What problems do you use Vibe Coding to solve? Feel free to leave a comment in the comment section