# What is Tessary's root cause analysis?

RCA is the third and last layer in Tessary's analysis pipeline: a strong model, running in a microVM, investigates a confirmed case and returns grounded hypotheses for what changed. It only starts once the layer before it, triage, has confirmed a case is real.

A firewall separates the two model-driven layers: RCA receives the finding's id and the case's own trace evidence, never triage's ruling on it. That keeps "nothing happened here" available as a genuine, independent RCA conclusion instead of an echo of what triage already decided, since it's the only check on the layer before it.

RCA's output is [a root-cause report](/answers/tessary-rca/what-is-a-root-cause-report), and how much it can name depends on [whether a repository is connected](/answers/tessary-rca/does-tessary-need-repo-access-to-find-the-root-cause). Not every case ends in a named cause; [when no hypothesis clears the bar](/answers/cause-attribution/what-happens-when-no-cause-can-be-found), that absence is the reported outcome.

---

Source: https://tessary.ai/answers/tessary-rca/what-is-tessarys-root-cause-analysis
More on Tessary RCA: https://tessary.ai/answers/tessary-rca
From Tessary, agent reliability for AI agents in production: https://tessary.ai