grounding — Tying a model's answers to checkable sources — retrieved documents, live data, tool results — instead of letting it answer from memory alone.
Why it matters
Shows what a VLM classification stack looks like in production at 30M predictions a day, and why a hand-maintained taxonomy is the piece that makes model output usable downstream.
