Power Your Agents with dbt and Snowflake

The Problem

Most teams run dbt and Snowflake side by side, but connecting governed data to AI agents still means custom plumbing. Practitioners want to build Cortex agents on trusted, well-modelled data, yet the path from transformation layer to agent layer is undocumented and unclear. The result: duplicated logic, ungoverned context, and a lot of guesswork about what "production-ready" actually looks like.

The Solution

This session walks through a single, opinionated reference architecture that closes the gap. dbt handles transformation and governance, the dbt MCP server exposes that governed context to AI tooling, and Snowflake Cortex agents consume it, all as one joined-up flow. No custom middleware, no duplicated definitions: one stack from data model to AI agent.

What You Will Learn

• The reference architecture connecting dbt, the MCP server, and Snowflake Cortex agents end to end
• How to configure the dbt MCP server to expose governed metadata and lineage to AI tooling
• How Cortex agents pick up semantic context from dbt, and what that looks like at query time
• What a production setup requires: configuration, testing, and operational considerations
• Where governance responsibilities sit across dbt and Snowflake in this architecture

Who This Is For

Analytics engineers, data engineers, and platform leads who already use dbt and Snowflake and want a clear path to building AI agents on governed data. Especially relevant if you are evaluating how MCP and Cortex agents fit into your existing stack.