A multi-agent AI system that evaluates tough organizational decisions from **7 specialized perspectives**: legal, ethical, economic, reputational, environmental, public health, and AI risk — then synthesizes these into a clear, actionable recommendation.

 **Key Features:** 
 
 **Multi-Agent Orchestration**: 7 specialized AI agents run in parallel, each with domain expertise and custom prompt templates for comprehensive decision analysis. 
 
 **Semantic Caching**: Vector-based caching system provides 85%+ faster responses for similar queries, reducing response time from 15-20 seconds to 2-3 seconds. 
 
 **RAG Implementation**: AI Risk Agent uses Pinecone vector database for retrieval-augmented generation, grounding recommendations in verified evidence to prevent hallucination. 
 
 **Content Guardrails & Security**: Multi-layered protection including OpenAI content moderation, prompt injection detection with 5 attack patterns, PII redaction, and comprehensive input validation. LLM injection protection prevents role confusion and system prompt bypass attempts. 
 
 **Enterprise Features**: Rate limiting (5 requests/day), PDF report generation with detailed agent findings, and real-time security logging for production readiness.

Moral Compass AI