Raw LLM Responses

(disclaimer. I use gpt for pretty much everything now. The text below are my original ideas refined by GPT. If it sounds AI written that's why) Essentially, I go to GPT with the problem. I will give it information I copy from my companies website, to give it context about what we do, and then I explain the problem or the goal I'd like to accomplish. I will come to it with some rough solutions so it has an idea of what I'm looking for. Depending on the problem, I will ask it to generate several ideas. Then, I will ask it to expand on the ones I like the most. Once I feel that it has a good idea for the solution, I will ask it to create a proposal document with several sections. Purpose, Outcome, Objective, Next Steps, and then anything else relevant to the specific project (for example, an agenda of the programming if it's an event, something like that). Now, I have a document that contains the context for a project in a way that both my co-workers and GPT can understand. It's useful for presenting to teams, and also giving GPT a bird's eye view of every aspect of my project. If I feel that any aspect of the project needs to be refined, I will copy the whole document, paste it in GPT, and ask it for context feedback. Usually, it will come up with some ideas on how to improve it, and I ask it to expound on the ones I like the most. Then, I will ask it to incorporate those new ideas into the proposal, and I update it in the google doc. One thing I've noticed is that if the document is too large, GPT-4 will get "dumber". So, if there's a particular part that I'd like to focus on, I will give it the whole document for context, and then ask it to provide feedback on a specific part. I keep all following dialogue focused on that part, and it will use the context of the entire project I gave it before but be smarter in its responses to that one particular part I'm focusing on. While I could provide specific prompts, I believe that the true power of GPT lie