EMOS — Enterprise Memory Operating System

Every meeting produces decisions and promises, but almost none of it survives past the calendar invite. EMOS solves this by turning raw meeting transcripts into structured organizational memory. When a transcript is uploaded, EMOS sends it to an LLM (via Fireworks AI) that extracts a summary, decisions made, action items, risks, dependencies, and open questions. Every commitment identified is automatically saved with an owner, deadline, priority, and status — no manual data entry required. The system then tracks these commitments over time: overdue status updates itself the moment a deadline passes, and blocked tasks are surfaced by tracing dependency chains between commitments. A live dashboard shows pending, overdue, and blocked work at a glance, and a built-in AI chat assistant lets users ask natural-language questions ("What's overdue for me?") answered strictly from their own meeting history — with no hallucinated information. EMOS is built as a full-stack application: a Next.js/React frontend, a FastAPI (Python) backend, and a PostgreSQL database, all containerized with Docker Compose for easy deployment.