This was a quick experiment on creating multiple LLM (GPT) agents with their own objective to see if outputs can be improved.