Ship AI agents like tables: dbt materializations for Snowflake CoWork

At dbt Summit 2026, Under Armour engineers described how they brought Snowflake AI agents and Cortex objects into the same governed dbt deployment workflow already used for thousands of data products. Previously, agents, semantic views, and Cortex Search services were created manually in Snowflake’s UI without version control, pull requests, CI checks, environment promotion, or reliable lineage, creating risks such as production agents returning inaccurate answers after upstream schema changes. The team developed custom dbt materializations for agents and Cortex Search services, while using a Snowflake Labs materialization for semantic views, enabling these stateful objects to be managed as dbt models without replacing them and losing conversation history or analytics. Integrating the objects into dbt’s dependency graph allowed impact analysis, selective CI builds, automated role-based grants, and production registration in Snowflake CoWork. Within five months, Under Armour had deployed more than 30 Cortex objects across three domains and environments, improving traceability, review processes, rollback capabilities, and business access to timely sales, consumer insights, and product knowledge, although migration from the UI and user-access request workflows still require manual steps.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.