Jekyll2026-08-14T21:49:01+00:00https://percentes.ai/feed.xmlReliability measurement for LLM inference — coordinated-omission-correct, censoring-aware, and honest about the requests that never finish.