In modern SaaS enterprises, security questionnaires are a major bottleneck. This article introduces a novel AI solution that uses Graph Neural Networks to model the relationships between policy clauses, historical answers, vendor profiles and emerging threats. By turning the questionnaire ecosystem into a knowledge graph, the system can automatically assign risk scores, recommend evidence, and surface high‑impact items first. The approach cuts response time by up to 60 % while improving answer accuracy and audit readiness.