Self-hosted · for dbt Cloud

Your dbt Cloud jobs, in order, on time.

Chain jobs into pipelines, schedule them, and when a model breaks, re-run only what failed.

Request a demo How it works

Replay · Nightly close 02:00:00

stg_paymentspayments stg_ordersorders fct_revenuerevenue dashboardsBI reverse ETLrETL

How it works

One pipeline replaces a crontab full of guessed gaps.

Stages are barriers

Jobs in a stage run in parallel. The next stage starts when all of them succeed.

Pipelines join

A pipeline waits for all or any of its parents, within a time window.

Schedules know what they missed

Cron with a misfire policy: skip, queue one, or catch up once.

Smart Retry

Re-runs only the failed models and their children, then resumes the pipeline.

Features

What you need once pipelines are live.

Alerts: Slack, webhook or email when a pipeline fails or recovers
SLAs: A "done by" time per pipeline, with an alert when it's late
Auto retry: Up to five retries with backoff, kept across restarts
Failure explainer: An AI summary of the error and a suggested fix (opt-in)
Insights: Duration trends, success rates and the models that fail most
Beyond dbt: Fivetran, Airbyte and HTTP steps in the same pipeline
Pipelines as code: YAML export and import, with a dry-run diff
Access: Invite-only roles, SSO and an audit log

Under the hood

One process, one SQLite file. It stores run metadata, never warehouse data.

Stack: Python, FastAPI, SQLModel, plain JavaScript; no external requests
Safe runs: The database blocks duplicate runs; a reaper unsticks stalled ones; cancel stays cancelled
Secrets: The dbt Cloud token is never written to disk and only goes to dbt Cloud hosts
Tests: pytest against a mock dbt Cloud with fault injection

Request a demo

Access is by invitation. Ask for a login to the demo workspace, or for a personal demo running against your own dbt Cloud account. You'll get an email once it's approved.