Every Responsible AI principle gets translated into operational controls at specific stages. Fairness, for example, becomes intake assessment for bias-relevant data, deployment gates requiring bias-test pass, and production monitoring for fairness-metric drift. Enzai codifies this translation per principle.

RAI controls live in shared workflows that engineering, security, product, and operating teams use natively, pushed into Jira, ServiceNow, and Slack. The RAI function defines the controls; operating teams execute them at deployment without context-switching into a separate RAI tool.

Enzai maintains the Compliance Framework library centrally. When NIST publishes a new AI RMF profile, when ISO updates AI standards, or when a sectoral regulator issues new RAI guidance, the library absorbs the change once. Affected systems re-trigger assessment automatically.

Agentic and generative AI governance is built into the same RAI platform. Agentic-specific controls (autonomy classification, action whitelisting) and GenAI-specific controls (training-data provenance, watermarking, prompt-injection mitigation) operate alongside the broader RAI controls.

Maturity is dimensioned across fairness controls coverage, transparency disclosure cadence, accountability documentation, safety monitoring, and robustness validation. Each dimension scored across the AI estate with trend lines over time, drawn from the same operational evidence base operating teams use.

AI governance is the broad programmatic discipline. AI compliance is meeting specific regulatory obligations. Responsible AI is the principles-led discipline - fairness, transparency, accountability, safety, robustness - operationalised across the lifecycle. The three overlap, and Enzai serves all of them.