WRITING / AI STRATEGY

Your Judge Needs an Eval Too: Validating LLM-as-Judge in Production

An LLM-as-judge isn't a measurement instrument until you've measured the instrument—pairwise labels, kappa against humans, and a version pinned in telemetry.

Published January 24, 2026. Reading time: 9 min read. Author: Morgan Wilson.

This public writing shell gives crawlers the canonical title, topic, excerpt, date, author, and next-step links before the React app renders the full essay.

Canonical next steps