What an enterprise AI gateway must enforce, per-user throttling, hierarchical budgets, and automatic provider failover, to control LLM cost and uptime at scale. LLM adoption inside enterprises is outpacing the governance layer around it. Three problems keep showing up in platform team backlogs: one agent or power user drains the shared provider quota and starves every other workload, the monthly L