Guardrails is a Python framework that helps build reliable AI applications by performing two key functions: Guardrails runs Input/Output Guards in your application that detect, quantify and mitigate the presence of specific types of risks. To look at the full suite of risks, check out Guardrails Hub. Guardrails help you generate structured data from LLMs.