SHIELDBOT is an AI governance layer that reroutes users to human support agents when an AI support chatbot is likely to fail or hallucinate. Through the attached documents, we show how to incorporate low-friction ethical checks throughout product development and production.