Every action governed at runtime.
Governance in aixplain isn't a layer you add after. It's embedded in the execution loop. Every agent action is inspected, scoped, and auditable before it completes — by default, not by configuration.
What governance means in practice
Runtime policy enforcement
Define what agents can and cannot do at the action level, evaluated before execution.
Action-level scoping
Permissions are set per agent, per tool, per data source. No agent can exceed its defined scope.
Audit trails
Every agent action is logged with full context: who triggered it, what it did, what it returned.
Inspector agents
Built-in Inspector micro-agents validate quality, compliance, and feasibility before outputs reach users.
PII redaction
Automatic detection and redaction of personally identifiable information in inputs and outputs.
HITL escalation
Configurable human-in-the-loop triggers. Agents escalate when confidence is low or when policy requires review.
Governance ships with every agent.
No configuration required. No bolt-on layer.