Loki dissects long texts into individual claims, determines which claims warrant verification, generates search queries to crawl for supporting evidence, and ultimately verifies each claim’s factual accuracy. It leverages OpenAI’s API for claim extraction and evaluation, and uses search APIs like Serper for evidence retrieval. The tool targets journalists, researchers, and anyone needing to validate the reliability of information, directly addressing hallucination detection in language model outputs.