Review generative and agentic AI designs-including retrieval-augmented generation, vector databases, third-party models, APIs, plugins, tools, and autonomous actions-and address controls to mitigate the risks of prompt injection, data leakage, excessive agency, insecure output handling, and AI supply-chain compromise. Define enterprise security architecture strategy, principles, reference architectures, reusable patterns, and multi-year roadmaps across network, endpoint, identity, application, cloud, container, data, and detection and response domains.