Your models stay. Agents keep them healthy.
dbt remains part of your stack. Dagen reads the project, heals drift through the DAG, and builds Agent Views on modeled data.
Models as first-class assets.
Parse the whole project
During discovery, agents parse your dbt project — models, sources, tests, docs — and the logic inside becomes reusable input for everything Dagen builds.
Logic travels with you
Moving to Iceberg or restructuring your warehouse doesn't mean abandoning your dbt investment — the logic travels.
Drift healing for dbt flows.
Impact through the DAG
When a source schema changes, agents trace the impact through your model DAG, propose the fix, and validate it against your tests.
Agent Views on modeled data.
Business meaning, MCP-served
Your dbt models already encode business meaning. Dagen builds Agent Views on top of them and serves the result over MCP.
Your repository, your review process.
Read your dbt project
Dagen reads your dbt project from your repository and connects to the warehouse it runs against — under the access you scope.
Map models and dependencies
Agents parse models, sources, tests, and docs so your existing logic becomes reusable input for healing and new generation.
Propose fixes through the DAG
When sources drift, agents trace impact through the DAG, draft fixes, and validate against your tests — arriving as proposals in the workflow you already use.
Agent Views on dbt models
Agent Views sit on your modeled data and are served over MCP, so AI agents consume business-meaningful tables instead of raw schemas.
Dagen and dbt, answered.
Does Dagen replace dbt?+
No. dbt is part of the stack Dagen works with. Your models are read, reused, and kept healthy, and new transformations can be generated in the patterns your project already follows.
Can Dagen fix a broken dbt model automatically?+
At the autonomy level you choose. Start with propose-and-approve: agents diagnose the break, draft the fix, and wait for review. Expand autonomy as the fix history earns it.
Do our dbt tests still matter?+
More than ever. Agents run your tests as validation for every fix and every generated model.
Where do fixes and new models land?+
As proposals your engineers review, in the repository and review workflow you already use — not as silent overwrites.
See it on your dbt estate.
Thirty minutes. Your stack, your scenario. We build a pipeline in front of you.