Snowflake's Cortex Code CLI (GA February 2026) enables full automation of semantic model and semantic view creation for natural language analytics. Instead of manually authoring YAML files over days or weeks, engineers can point the CLI at a schema and have it profile tables, infer relationships, generate synonyms, define metrics, and deploy a semantic view in minutes. Custom skills encoded as markdown files make the workflow repeatable, team-shareable, and CI/CD-triggerable, so semantic layers stay in sync as schemas evolve. At Pfizer, this approach supports over 225 active semantic models. The result collapses the time from data product publication to natural language access from weeks to minutes.
Table of contents
If you’re still writing YAML files by hand, you’re already behind.Three Eras of Semantic Layer CreationWhat Automation Actually Looks LikeMaking It Repeatable: Custom SkillsGet Matt Massey’s stories in your inboxKeeping Semantic Models Current: The CI/CD AngleWhy This Matters for Domain-Driven OrganizationsThe Bottom LineSort: