How I Build Things

Before AI, I typically started out with an idea of what I’m trying to accomplish based on some need that I have.

I write out a list of steps that I need to accomplish said task.

After my list is written out I would fill in the details beneath each step (like how I would accomplish subtasks to accomplish the overaching step).

If I am writing a script I would then decide of each step should be a function or if it’s okay to just write the code in my main execution area.

Now that we have LLMs I can typically ping chatgpt a few times to see what it thinks about how I can accomplish something. ChatGPT isn’t always correct but it can provide a nice framework for whatever it is I am trying to accomplish (and reduces the mental load of trying to write my own framework). I find that sometimes I have to go back to my old method of developing a new process or (config / script / plan) (ChatGPT isn’t always reliable).