AI & detection engine
LLM / sLLM
Definition
Large language models and self-hostable small models (sLLM), used for intent classification.
For executives (CEO·CISO)
Powerful at catching subtle intent (impersonation, payment pressure), but a hosted LLM effectively sends message bodies outside. Commonly missed: "has AI" matters less than where data goes — if sovereignty matters, confirm an sLLM/on-prem option and cost control (tiering).