Feature Request: Rovo AI should track and report its own health metrics

[Phil Eichmiller]

Feature Request: Rovo AI should track and report its own health metrics

When Rovo fails to complete an action — like creating a Jira ticket — the only feedback loop is the user noticing, retrying, and eventually giving up or contacting support. There's no mechanism for the AI itself to recognize patterns in its own failures and escalate them.

The problem: If Rovo's ticket creation (or any action) starts failing at a high rate, there's no self-awareness or automatic alerting. Users experience repeated failures individually, with no way to know if it's a systemic issue or something on their end. The burden falls entirely on users to report problems through support channels — which most won't do.

The suggestion: Give Rovo basic operational self-monitoring:

• Track success/failure rates per action type (e.g., Jira ticket creation, page updates, searches)

• Detect anomalous failure rates and surface them to Atlassian engineering automatically

• Optionally, let the AI inform users in-conversation: "I'm seeing elevated failure rates for this action right now — I've flagged it internally"

This is standard observability practice for any production system. The AI layer should have the same operational awareness we'd expect from any service — especially one that users are relying on to get real work done.

Why it matters: As organizations adopt Rovo more deeply, silent failures erode trust quickly. A user who hits three failed actions in a row with no explanation will stop using the tool. Self-monitoring would catch systemic issues faster, reduce support burden, and make Rovo a more reliable teammate.

[Trudy P Claspill]

Hello @Phil Eichmiller 

Welcome to the Atlassian community.

As this is a user community it is not the best path for getting a change request directly to Atlassian.

Better options are:

1. Ask you Administrators to submit the change request through a support case.

2. Under the Help menu in most of the products (i.e. Jira) you can find an option to provide feedback directly to Atlassian.

[Phil Eichmiller]

thanks for the alternatives. those all require human diligence. That's kind of silly don't you think, when we have intelligent agents that could monitor it all with zero human diligence needed. Anyway, I appreciate the response and didn't expect this to go anywhere. I hope the community here sees the message and expects more from the tools we buy from Atlassian. Because why am I the one to suggest software automatically monitor it's own system health.