Better Experiments with LLM Evals — A funnel, not a fork | Spotify Engineering

TL;DR  LLM evals, automated judges that assess relevance, coherence, and quality at scale, are a powerful new tool. Paired with online experiments, they raise the hit rate of what we test and create a feedback loop that makes both evals and experiments smarter over time.

Raccontata daengineering.atspotify.com