An observability platform for large language models that helps teams detect hallucinations and improve model performance.