PromiseOS - Commitment Graph & Risk Cascade Engine

Most team commitments never make it into formal project management tools like Jira or Asana. Instead, they live scattered across informal chats, WhatsApp groups, emails, and quick check-ins. The real danger isn't just forgetting a task - it's the invisible dependency chains. Person A promises a critical deliverable to a client, silently assuming Person B will deliver a prerequisite first. When Person B slips by just two days, the entire downstream delivery collapse before anyone even notices. PromiseOS solves this by transforming unstructured conversation threads into an active, living commitment graph. Powered by an autonomous 5-agent pipeline orchestrated with LangGraph, it extracts commitments, resolves cross-person dependencies into an interactive DAG, and deterministically computes cascading delivery risk using verifiable mathematical factors rather than vague LLM estimates. It features a counterfactual "What-If" simulator that lets team leads model delay impacts before they happen, grounded evidence citations to prevent AI hallucinations, and a strict Human-in-the-Loop approval gate for mitigations. Optimized for AMD ROCm hardware, PromiseOS delivers over 5.3x inference speedups for real-time risk intelligence.