I Ran Technical Due Diligence on My Own AI Startup. Three Findings Ended It.
The prototype worked. The economics, the market and the ownership review did not.
Read the articleThe prototype worked. The economics, the market and the ownership review did not.
Read the articleA small AI experiment exposed failures in answer selection and grading. Why complete traces matter before trusting or scaling an evaluation.
Read the articleLLM judges made automated AI evaluation practical, and introduced new ways to fool yourself. Here’s how I design judge panels, and what I’d ask before trusting one.
Read the article