Explainable Decision Layer for Agent Classification & Routing

A hybrid rules-plus-LLM decision layer for the boring, high-volume classification work agents actually do (ticket routing, tagging, prioritization) — where every decision carries a human-readable reason, deterministic rules handle the clear-cut cases, and the LLM only handles genuine ambiguity. Built for the failure mode where 92% accuracy with no explanation makes a team stop trusting the system and check everything twice.