Bury — stress-test any claim before it costs you

Bury is an adversarial research engine. Give it a claim, a thesis, a pitch or a set of numbers, and four AI agents argue over it: a Proponent builds the strongest defence, a Contrarian attacks it with disconfirming evidence, a Judge scores what survived, and an Epistemic Auditor checks the debate itself for drift and ignored evidence.

What you get back

• An argument integrity score from 0 to 100, with the reasoning behind it.
• Every objection that landed, and every one that was rebutted.
• The load-bearing assumptions your claim rests on — the ones that break it.
• Deterministic checks on any figures you paste: arithmetic reconciliation, growth divergence, Benford's law, round-number clustering and p-value clustering.
• A shareable report.

Who uses it

Founders pressure-testing a pitch or a strategy, researchers checking a thesis against disconfirming evidence, analysts reading a performance report for where the numbers stop adding up, and anyone who would rather find the hole themselves than have someone else find it.

How it differs from a chatbot

A chatbot agrees with you. Bury is built to disagree: it searches for evidence that contradicts the claim, weights base rates and null results above commentary, and reports the objections it could not answer.