Just call one business internal and the other one LLM inference. You don’t want AI crafting packets on business internal.