## How to Diagnose and Fix AI Inference Latency on Kubernetes It's Thursday, around quarter past two. Your team shipped an internal assistant two weeks ago. The demo went well enough that someone in finance asked whether it could read contracts. Word spread. Today, for the first time, everyone is using it at once. The support channel has the same complaint arriving in six different tones. Is it down? Mine is just spinning. It worked this morning. Somebody has posted a screenshot of a loading indicator with no caption, which you suspect they're enjoying. So you open the dashboard. Nodes ready.…