AI Lab: LLM-Assisted Log and Alert Triage

Measure what a language model actually contributes to a SOC queue — against a fixed corpus with known answers, graded on precision, recall, hallucinated entities and cost per 1,000 events.