Module 02dbt Transforms

Your dbt project, run by an agent

The Transforms tab runs, schedules, and watches your dbt project — and when something breaks, the agent diagnoses it against your live schema, ships the fix as a diff, and re-runs. Not just an orchestrator: a colleague.

Replacesdbt Cloud
Mako — Live Miniature
models/marts/fct_mrr.sql
- GROUP BY plan
+ GROUP BY plan_id
Agent: fct_mrr failed on a renamed column — upstream now emits plan_id. I updated the ref and re-ran the 3 affected models.
34 models · 12 tests passed · lineage up to date

Live miniature, open on dbt Transforms — the agent just fixed the nightly run. Click the fixed model for the diff, or read the chat on the right.

Scheduling

Runs, schedules, and lineage — the table stakes

Point Mako at your dbt repo and you get scheduled runs, per-model status and timing, logs, and lineage out of the box. Everything you'd expect from a hosted orchestrator, next to the same warehouse connections your Consoles and Flows already use.

Mako — Transforms
helio-prod › Transforms › schedules
nightlydbt build --select tag:core
hourlydbt run --select tag:realtime
on-syncdbt run --select source:stripe+
34 models · 12 tests · lineage up to date
Agent on call

Failures get fixed, not just reported

A renamed column upstream, a type change in a source, a test that started failing at 3am — the agent reads the error, the model SQL, and your actual warehouse schema, then proposes a concrete diff. You review it; it re-runs the affected models and reports back in the chat.

Mako — Chat
Chat — run #482 failed → fixed
fct_mrr: column "plan" does not exist
Upstream stg_stripe__subscriptions renamed plan to plan_idin yesterday's sync. I've updated the ref and the two downstream models that used it.
- GROUP BY plan
+ GROUP BY plan_id
dbt run --select fct_mrr+Done
✓ Re-ran 3 models — all green
Modeling

New models, written on request

Describe the metric you need and the agent drafts the dbt model — staging, intermediate, and mart — following the conventions already in your project. It knows your sources because it synced them, and your schema because it queries it every day.

Mako — Chat
Chat — new model
Add a model for net revenue retention by cohort month.
Creating models/marts/fct_nrr_by_cohort.sqlDone
Adding schema test: cohort completenessDone
Drafted fct_nrr_by_cohort on top of fct_mrr, following your fct_ naming and folder layout. Ready for review.
The agent fixed our nightly dbt run before I'd finished my caf. I felt a great disturbance in the on-call rotation.
Owen L.Moisture Farm Operations

Give your dbt project a colleague.

Connect your repo, schedule a run, and let the agent take the 3am pages.