How would you design a system to prevent unintended medical advice from being generated in Assort Health's AI agent?
