[datadrivenangel]

Makes sense. Any thoughts on expanding scope to have multiple 'analyst' roles per LLM model? Could be interesting to see if changing roles/prompts yields better results.