Cortex acts as a safety gate for AI agents that spend money, providing a layer of oversight that allows, escalates to humans, or denies financial transactions. It monitors agent spend proposals, validates them against policy, and records decisions to a ledger, making agent-driven spending safe for enterprise use.