When a Problem needs more than a list of correlated alerts, AI Analysis explains why it happened and what to do next. Click AI Analysis on any Problem in Atatus AI SRE to investigate the logs, metrics and traces behind it and get a probable root cause with a confidence score.
It runs on demand, separate from the automatic correlation that groups related alerts into a Problem. The AI Root Cause Analysis docs walk through each tab.
What AI Analysis gives you
- A root cause you can check: The Primary tab states the cause with a likelihood label and confidence percentage, the AI reasoning, and the evidence with its source. Evidence tied to an error opens in the error inspector.
- Steps to confirm and fix: Validation tests let you confirm the cause yourself, such as running
EXPLAIN ANALYZEon a slow query, and a recommended action says what to do about it. - The full picture: Contributing lists secondary factors by weight, Alternatives shows the hypotheses the AI ruled out, and Prevention suggests longer-term changes so the same kind of Problem does not come back.
- An action plan your team owns: AI-generated steps carry a Critical, High, Medium or Low priority. Add your own steps, check them off as you work, and keep them when the analysis is re-run.
- Hypothesis, not confirmation: Until automatic correlation confirms a root cause, the Detected Root Cause panel shows the AI conclusion as its current best hypothesis, labeled as coming from AI Analysis.
- Re-analyze as data arrives: Each run is saved as a version labeled with the AI model behind it. Mark it as helpful or not helpful to tell Atatus how useful it was.