About
What Mad World is
Mad World is a multi-model AI debate lab. You bring a hard question. Models from different labs discuss it in one room — with optional live web search — while a neutral judge tracks agreement, dissent, and lean. Council mode freezes independent answers and ranks them before anyone can talk anyone else into changing their mind.
The product exists because one chatbot can sound certain while only agreeing with itself. Mad World makes disagreement and sources visible on purpose.
What it is not
- Not a coding-agent framework (not Autogen / Crew / LangGraph).
- Not an official fork or endorsement of Karpathy’s llm-council — Council mode is inspired by that idea.
- Not a claim that multi-agent debate always beats a strong single model.
Who it’s for
Researchers, founders, investors, and anyone who wants a harder look at a claim than a polite single-model chat. If you care about dissent, citations, and decisions you can audit, it’s for you.
Who built it
Mad World is a small personal project in early beta — not a funded company. The goal right now is honest usage and feedback, which is why sign-in includes free credits and no card.
Privacy in one line
Debates are private to your account unless you create a public share link. Details: Privacy.