Why it matters
Natural-language-to-data interfaces are only as good as the schema and metric context fed to the model, and this event is squarely about the tooling layer that governs that context before it reaches an LLM.
The tokenmaxxing angle
A governed semantic layer is a context-efficiency play — an NL-to-SQL agent queries a curated metric definition instead of raw schema, cutting the kind of bloated, ambiguous context that drives up both cost and hallucination risk.
From the organizers
The event page names the mechanism directly: Snowflake CoCo automates semantic modeling in a dev environment and syncs it back as governed "Semantic Views" feeding Cortex Analyst, notebooks, and dashboards.