Arcanna Decision Models is a decision layer platform designed for Security Operations Centers (SOCs) that uses analyst-taught, supervised machine learning models to automate security alert triage and response decisions. The platform operates on a confidence-based decision framework where actions are automatically executed at 90-100% confidence, require human validation at 70-89% confidence, and provide recommendations only below 70% confidence. The system grounds decisions in organizational context by incorporating historical decisions, organizational ontology, and validated human feedback rather than relying on isolated AI agent determinations. All decisions are versioned and auditable to maintain traceability. The platform consists of four main components: structured decision pipelines that move events from ingestion to action with human gating controls, analyst-taught supervised models that apply consistent decision patterns with explainability, execution workflows that apply decisions across integrated security tools within defined boundaries, and a decision graph that links signals, decisions, and outcomes for context and audit purposes. Arcanna integrates with SIEM, SOAR, and endpoint security platforms to govern decisions before action execution occurs. The platform aims to provide automation speed while maintaining human control over critical security decisions through configurable guardrails and oversight mechanisms.