Babel is a copilot for VC negotiations. It finds the Zone of Possible Agreement — the overlap between what a founder will accept and what an investor will offer — and uses it to score every clause in a term sheet against who has leverage in the deal.

Babel is named after the Tower of Babel. In Genesis, humanity spoke one language and began building a tower to the heavens. God confused their speech into many tongues, and construction collapsed; structural failure because the workers could no longer understand each other.

This is not dissimilar to what happens in venture deals: founders, investors, and their lawyers are negotiating the same term sheet, but “market” means one thing to a first-time founder, another to a Series B lead, and something else to outside counsel. Everyone’s talking past each other adding negotiation overhead.

Babel institutionalises the tribal knowledge I built doing deals at Sequoia Capital, encodes it as math, bridges the terms both sides are arguing over, and generates consensus. The LLM doesn’t hallucinate because the math is the source of truth; the model only explains what the engine already decided.

Here are the posts related to this project: