policy-evidence-engine

Story Connect AI Copilot is an AI-powered employee intelligence and policy-evidence platform designed for the hospitality industry, particularly luxury hotels. The project helps employees and HR teams understand internal policies, resolve conflicting information, access relevant training guidance, and make better-informed decisions using organizational knowledge. Rather than functioning as a conventional chatbot, the system focuses on evidence-backed answers and auditable reasoning. It combines AI-powered analysis with structured evidence retrieval, source attribution, citation validation, contradiction detection, and a human-review workflow. I integrated Meta's Muse Spark Open model through the Meta Model API using Baseten as the inference provider. The model contributes to analyzing evidence and generating structured candidate relationships, including potential contradictions. These outputs are treated as proposals, not automatically verified conclusions. The system independently validates evidence and citations, applies its own verification rules, and preserves human oversight before decisions can be considered final. My goal is to demonstrate how powerful AI models can support more transparent, reliable, and accountable knowledge systems in real workplace environments.