Use AI Ticket Triage
What happens when AI Triage classifies a new ticket, runs diagnostics, and posts a triage note — and how to give feedback on its accuracy.
Overview
When AI Ticket Triage is enabled (see Configure Autonomy Policies), Plenix automatically analyses every new support ticket the moment it's created — before an agent even opens it — and posts an internal note with a category, a confidence score, and (where possible) live diagnostic results from the affected device.
What happens automatically
- A ticket is created — via the Service Desk, the customer portal, or an inbound email.
- AI classifies the ticket into one of: Connectivity, Performance, Printer, Software, Storage, Updates, Errors, or Other — with a confidence percentage and a one-line reasoning summary.
- Plenix resolves the affected device — using the device linked to the ticket, a hostname mentioned in the AI's classification, or a hostname pattern found in the subject/description.
- If a device is found and the policy mode is Auto, Plenix runs the matching diagnostic playbook automatically (e.g. a connectivity ticket triggers Network Connectivity, a printer ticket triggers Printer Queue State) — see Run a Diagnostic Playbook for what each one checks.
- An internal note is posted on the ticket, tagged [AI Triage], containing the category, confidence, linked device, diagnostic output (if it ran), and suggested next steps for the agent.
This typically all happens within a few seconds of the ticket being raised.
Where to see it
Open any ticket in the Service Desk. If AI Triage ran, you'll see an internal note (not visible to the client) near the top of the comment thread, marked [AI Triage], with the classification, any diagnostic results, and a Suggested next steps section.
Giving feedback
Below every AI Triage note is a "Was this helpful?" prompt with thumbs up / thumbs down buttons:
- 👍 if the classification and suggested next steps were accurate and saved you time
- 👎 if it was off-base or wasn't useful
Ratings aren't currently used to retrain anything automatically, but they're stored against the comment and give your team a quick signal for how well triage is performing for different ticket types.
If no device was found
If AI Triage couldn't confidently match a device (no device linked to the ticket, and no recognisable hostname in the text), the note will say so and skip the diagnostic step. Link a device to the ticket and re-run the relevant playbook manually from the Linked Device panel — see Run a Diagnostic Playbook.
Where to go next
Was this article helpful?