Last released Jun 15, 2026
Guardrail and evaluation framework for business-process AI agents — governs and tests action safety, not just output quality.
Supported by
