Founder-GPT: Self-play to evaluate the Founder-Idea fit

Abstract: This research introduces an innovative evaluation method for the "founder-idea" fit in early-stage startups, utilizing advanced LLM techniques to assess founders' profiles against their startup ideas to enhance decision-making. Embeddings, self-play, tree-of-thought, and critique-based refinement techniques show early promising results that each idea's success patterns are unique and they should be evaluated based on the context of the founder's background.