Establish AI guardrails so generated content stays on-brand for [Company] using [GenAI Tools]. Define style constraints, prohibited phrases, and safe defaults ([Temperature Range]). Provide a canonical “brand system prompt” and task-specific prompt snippets (email, landing page, release notes). Address [PII Rules] and [Approval Workflow].
Output format: 1) Guardrail policy (one page). 2) System Prompt (copy/paste block). 3) Prompt snippets (6). 4) Evaluation rubric (1–5) with [Metrics] (brand fit, clarity, compliance). 5) Human-in-the-loop steps.