One read-only token. Pipesight polls every Databricks run around the clock — the second a job fails, a clean, deduplicated Slack alert lands with the job, the error, and a link to the run.
Work email. No credit card · read-only token · cancel anytime
fct_orders_daily throws an AnalysisException and dies. No dashboard blinks. No email gets past your filters. Silence.
The failed run is detected, the real error is pulled from Databricks, and it's deduplicated against open issues. One failure, one signal.
Job, error, timestamp, link to the run. Whoever's on call taps once and is looking at the failed run — not digging through the jobs UI.
Re-run succeeds at 03:22. The morning dashboards load on time. Nobody ever knows — which is exactly the point.
No SDKs, no sidecars, no infra changes. Pipesight sits outside your stack and watches it — if you can create a token, you can be monitored before your coffee cools.
Create a Databricks token with read scope and paste it once. That's the entire install. Nothing runs inside your workspace.
Every job in the workspace appears on its own. Toggle the ones that matter — no wiring alerts by hand, no YAML.
Pipesight polls every run around the clock. A failure becomes one deduplicated Slack alert — job, error, link to the run.
Nothing more, nothing less — everything you need to triage, the second a job goes red.
One failure, one message. Retries and repeat runs fold into the same issue — never forty pings.
The real exception, pulled from the run — not "something went wrong."
Straight to the failed run in Databricks. Triage starts in seconds.
No trial clock, no locked screens. Free gives you the full triage UI forever — the only ceilings are 5 jobs and 5-minute polling.
First 50 teams on the waitlist get 3 months of Pro free.
Stop finding out your pipelines failed three days too late. First 50 teams get 3 months of Pro free.
No credit card · read-only token · cancel anytime