Comment by udave

Let's be real, if you and I both ask claude to generate a feature on the same project, what are the chances that it spits out 100% replicated code? But if we are to build the project using a Dockerfile, we will get the same binary and the same image. Products around LLMs are non deterministic unlike compilers.