A 20-minute hands-on lab. You run dbt Wizard against a retail dbt project on Snowflake to investigate data, extend an intermediate model with a new Fivetran-synced source, and recover from a broken pipeline caused by an upstream column rename.
A 90-second walkthrough of the onboarding scenario at 1x. This is the first scenario every attendee runs.
Onboarding scenario walkthrough. Scenarios 1-4 are linked below.
dbt Wizard is an AI agent built from the ground up for the way analytics engineers actually work. Not just code generation, but the entire data lifecycle: asking questions, investigating, understanding, changing, validating, and shipping. It's grounded in real dbt project context: lineage, tests, contracts, and defined metrics. And it's available wherever analytics engineering work actually happens.
A conversational workspace in the dbt platform, plus an embedded assistant inside dbt Studio. Tailored capabilities ship in dbt Canvas and dbt Insights for visual modeling and analytics work.
A terminal-native agent for local development. Runs against your dbt project on your laptop, with your warehouse credentials, in the editor you already use. This is what the HOL uses.
The agent reads the dbt project natively (lineage, tests, contracts, semantic definitions), edits across files coherently, and validates its own output against expected results before returning.
Ten capabilities that distinguish a purpose-built AE agent from a generic code assistant.
Investigate, change, validate, and ship from one surface.
Lineage, contracts, semantic definitions, tests, governance - all in context.
Reads your project the way dbt itself reads it. No translation layer.
Edits across models, schema files, and tests in a single coherent change.
A decade of analytics-engineering knowledge maintained by dbt Labs and the community.
Built-in comparison loop runs your changes against expected outputs before handoff.
You stay in control. The agent proposes; you approve.
Reasoning and tool calls are visible. Nothing happens in a black box.
Purpose-built for AE workflows, not a generic chatbot in a costume.
Snowflake, BigQuery, Databricks, Redshift, and others.
Each scenario is a self-contained workflow that takes 10-15 minutes. Start with onboarding, then pick scenarios that match your role. Scenarios 2 and 4 include three interchangeable paths.
Tour the dbt project at The Builder Depot. Inventory the models, sample real data, then create and preview a new mart model (orders_by_week).
Identify stores whose actual inventory differs from the expected shipment-plan quantity. Build and materialize inventory_shipment_variance.
Add columns from a Fivetran-synced source to an existing intermediate model without breaking downstream consumers. Three interchangeable paths.
Build a 180-day customer activity layer plus a segment model tagging VIPs, big spenders, and category-loyal customers.
Diagnose a failing dbt run caused by an upstream column rename. Map the blast radius, apply an alias-preserving fix across SQL and YAML, re-run. Three interchangeable paths.
git clone https://github.com/fivetran-jacklowery/dbt_wizard_hol.git"What really sets the dbt Developer Agent apart is its precision in identifying and resolving bottlenecks. It doesn't just suggest code; it understands our existing tests and lineage well enough to troubleshoot issues almost instantly. This has significantly reduced our build times and allowed us to scale our dbt project with total confidence in our data's trustworthiness."
"We went from about 60 conformance errors to 7, using the Fusion migration agent. That's the difference between too hard and actually doable."
Install dbt Wizard, authenticate to GCP, create a dbt account, validate the environment.
One-page printable reference with every prompt in the lab.
Snowflake account, source schema, dbt project layout, and per-attendee credentials.
dbt project, seed data, scripts, and lab assets.