Locate the onset
Find when the error rate changed, from your observability backend.
Agentic · ships today
Find the deploy
that broke it
Tie an error onset to the deploy that most likely caused it, from your deploy log and observability backend.
Ships as incident/change-correlation
When errors start, the first question is which change did it. This pipeline answers with deterministic timestamp math over your deploy log and metrics, then lets an agent narrate the evidence for the on-call engineer.
The pipeline
Correlate deploys
Deterministic timestamp math ranks deploys and config changes against the onset.
Narrate
An agentic step explains the ranking with the evidence it used.
Engineer decides
Roll back, hotfix or dismiss. The decision is recorded next to the ranking.
What the record keeps
The onset window, the deploy log, the ranked candidates with their scores, the narrative, and the engineer's decision.
Deterministic first
The ranking is math; the model only explains it.
Minutes, not a war room
The likely culprit is on screen before the call starts.
Composable
Feeds incident/rca when the cause needs a deeper investigation.
Ships as incident/change-correlation on the git operator. Deploy-log and metrics sources depend on the installed operators.
See how failure can trigger a reviewed improvement cycle →Run it on your own infrastructure.
Apache 2.0 licensed, self-hosted. cairn pipeline run incident/change-correlation