Security and governance
Control what AI can know, decide, and do before the model participates.
Cortex applies identity, tenant scope, permission-aware retrieval, approved tool execution, validation, handoff, and audit controls before AI sees context or performs a business capability.
Governance model
Cortex narrows what the model can see and which tools it can use before generation begins. Governance is part of the runtime, not an instruction hidden inside a prompt.
- Permission-aware retrieval only fetches knowledge the current user, tenant, and channel are allowed to use.
- Tenant and domain scope keep each assistant bounded to the right workspace, customer, product area, and data domain.
- Approved tool execution exposes business actions through controlled tools instead of uncontrolled model behavior.
- Business-rule validation checks policies, confirmations, and workflow rules before Cortex performs an action.
- Human handoff patterns escalate sensitive, uncertain, or exception-heavy moments to the right team.
- Audit-friendly activity records retain evidence for retrieved context, tool calls, outcomes, and decisions.
Contact Cortex IR at sales@cabin4j.com or +61 466 809 425.