Use case · On-call & SRE teams
The worst part of the page isn't the fix, it's the first twenty minutes of hunting: which dashboard, which deploy, which downstream dependency. With Sentinel, that hunt starts the moment the alert fires. By the time you open your laptop, the evidence is gathered, the timeline is building, and a cited diagnosis is waiting for your review.
What changes on call
The correlation work, error spike against deploy timeline against trace, happens in the first minutes, not after a human has woken up and remembered where the dashboards live.
Artifacts are captured at the moment of collection, while the incident is live, not reconstructed later from screenshots and memory. Nothing depends on who thought to save what.
When the incident outlives your shift, the next responder inherits the investigation, the evidence, and the reasoning so far, not a chat scrollback and a guess.
The post-incident review starts from the provenance trail: what triggered it, what was checked, what was found, what was decided, already in order, already stamped.
Human in the loop
Nothing is acted on without review. The report is a proposal, a diagnosis with its evidence attached, and because every claim cites the artifact behind it, you can verify it in minutes or challenge it on the spot. Your challenge joins the record too, which is exactly what makes the next investigation better.
How the runtime worksEvidence gathered while you're still waking up, a diagnosis you can verify claim by claim, and a trail your whole team can build on.