Contradish

Contradish builds the reliability layer for AI reasoning. LLMs can produce correct answers once and then contradict themselves when the same question is phrased slightly differently. Today there is no standard way to test whether an AI system’s reasoning remains stable under these variations. Contr... Read more