6. ALM, Security, Governance, and Responsible AI (40–45%)

This phase completes the Deploy domain by covering the operational guardrails that keep AI solutions safe, compliant, and manageable over their lifecycle. ALM (Application Lifecycle Management) ensures you can version, test, and promote AI solutions across environments without breaking them. Security ensures adversaries can't exploit your agents, models, or data. Responsible AI and governance ensure your AI solutions operate within ethical and regulatory boundaries.

The exam tests these topics with scenario-based questions that require you to identify what's missing from a proposed architecture. A well-designed agent running in production without proper ALM will eventually break during an update. Without prompt injection defenses, it will be exploited. Without data residency compliance, it will create legal liability. The architect's job is to design all these safeguards before they're needed.