A system that detects hallucinations in Retrieval-Augmented Generation outputs by extracting and labeling claims, then aggregating results to resolve conflicts and ensure factual accuracy.