Phases in my LLM use for programming (www.asfaload.com)

A programmer reflected on their evolving use of large language models (LLMs) for software development, highlighting a journey that began just six months ago. Initially, they utilized chat interfaces to address simple programming questions while learning Rust, gradually increasing the complexity of their queries by pasting code snippets and allowing LLMs to generate test cases. This led to significant time savings, as LLMs handled mundane tasks while the developer focused on more critical aspects of their project. As their requirements grew, they embraced coding agents, giving them limited write access in a controlled environment, which enhanced their development efficiency. The programmer ultimately adopted a skills framework to streamline collaboration with LLMs, employing structures that helped organize tasks and improve project management. By outlining plans and reviewing LLM-generated code, they identified common pitfalls, like unused functions and redundancies. With their experience, they transitioned from using free LLM services to more affordable subscription plans, finding that the right tools enhanced productivity without compromising on quality. This account underlines the growing integration of LLMs in programming workflows, illustrating both the benefits and challenges developers face as they adapt to this evolving landscape.