What do you to prevent your GenAI from hallucinating?

We are building LangWatch, so far great results of companies who improve their AI solution with our insights. Launching soon officially. But wondering how you currently handle this