Private Knowledge Assistant Prompts for Open Weight Models

# Private Knowledge Assistant Prompts for Open Weight Models

**Knowledge Source Inventory**: Role: Knowledge management consultant. Task: Create an inventory plan for documents that should power a private AI assistant for [team]. Include source owner, sensitivity, update frequency, permission group, and whether the document is ready for indexing.
**RAG Answer Policy**: Role: AI policy designer. Task: Write answer rules for a retrieval-based knowledge assistant. The assistant must cite sources, refuse unsupported answers, avoid guessing, and escalate restricted topics. Output policy text plus 10 example responses.
**Evaluation Question Set**: Role: AI evaluator. Task: Generate 50 evaluation questions for a private knowledge assistant using [document categories]. Include expected answer type, required source, risk level, and pass criteria.
**Access Control Map**: Role: Security architect. Task: Map access controls for a private knowledge assistant used by [departments]. Output roles, allowed document collections, restricted topics, audit events, and admin workflows.
**Employee Launch Guide**: Role: Change management lead. Task: Create an employee launch guide for a private knowledge assistant. Include what it can answer, what it cannot answer, how to verify citations, and how to report bad answers.
**Bad Answer Review Template**: Role: AI quality manager. Task: Create a review template for failed knowledge assistant answers. Include question, retrieved source, generated answer, failure category, correct answer, root cause, and fix owner.
**Document Refresh Workflow**: Role: Operations designer. Task: Design a monthly document refresh workflow for an SME knowledge assistant. Include owners, approval steps, re-indexing trigger, regression tests, and communication to users.
**Model Selection for RAG**: Role: AI architect. Task: Compare open weight models for a private RAG assistant. Evaluate answer quality, citation behavior, latency, context length, cost, and deployment complexity for [company constraints].
**Citation Quality Scoring**: Role: AI evaluator. Task: Create a scoring rubric for citation quality in a knowledge assistant. Include source relevance, answer support, conflicting evidence, missing context, and user trust impact.
**Private Assistant Roadmap**: Role: Product manager. Task: Create a 90-day roadmap for a private knowledge assistant. Include MVP scope, pilot team, metrics, governance, feedback loop, and scale criteria.