# What's the difference between a failure mode and a root cause?

A failure mode is the recurring category of error, like wrong tool selection, a lost instruction, or a loop that never ends; a root cause is the specific change, a commit, a prompt edit, a model swap, that made that category spike right now. Running a check for a failure mode across many traces turns it into a rate you can watch over time. The mode tells you what shape the problem takes; the cause tells you why it's happening now instead of last week. A call site can carry "wrong tool selection" as a standing failure mode at some baseline rate for months, then spike the week a tool's parameter gets renamed. The mode was always possible. The root cause is what turned the possibility into an actual spike, and it's the cause, not the mode's name, that says where the fix belongs.

---

Source: https://tessary.ai/answers/failure-modes/difference-between-a-failure-mode-and-a-root-cause
More on Failure modes: https://tessary.ai/answers/failure-modes
From Tessary, agent reliability for AI agents in production: https://tessary.ai