This article explores a novel architecture that couples retrieval‑augmented generation, prompt‑feedback cycles, and graph neural networks to let compliance knowledge graphs evolve automatically. By closing the loop between questionnaire answers, audit outcomes, and AI‑driven prompts, organizations can keep their security and regulatory evidence up‑to‑date, reduce manual effort, and boost audit confidence.

This article introduces a novel architecture that combines self‑supervised learning on edge devices with dynamic knowledge graph evolution, enabling real‑time compliance monitoring, automated policy enforcement, and zero‑latency insights for regulated enterprises.