Interview Guides
You are designing the first-run onboarding dialogue for a creative writing companion chatbot. The bot should help users brainstorm ideas, develop characters, and continue drafts without sounding generic or overbearing. You also need fallback error messages for cases where the model is uncertain, the request is unsafe, or the user tries to override the assistant's instructions.
How would you draft the onboarding dialogue and fallback error messages so the chatbot feels helpful and creative, while still handling hallucination risk, prompt injection attempts, and evaluation of prompt quality before launch?
