The autonomous AI agent & triage

The autonomous agent works alongside your team. It is suggest-first by design: it always shows its thinking and drafts a reply, but a human is always the one who presses send.

What it does on a new ticket

1. Posts a triage read-out to the timeline — proposed priority, category, sentiment and tags, with a confidence percentage.
2. Auto-applies a priority raise and merges its tags — but only when auto-apply is on and its confidence clears your threshold.
3. Posts a suggested reply you can accept, edit or ignore.

It never emails the customer on its own. When a ticket is assigned, urgent, and the sentiment is negative, it also raises an escalation alert.

Run it on demand

Open any ticket and click Run agent in the AI Assist bar (tooltip Triage this ticket and draft a reply). This runs the full triage immediately, even on older tickets.

Set the guardrails

In Admin → Settings, under the autonomous agent controls:

• Enabled — master switch.
• Auto-run on create — triage every new ticket automatically.
• Auto-apply triage — let it actually change priority and tags (vs. only suggesting).
• Minimum confidence — the bar it must clear before applying anything (default 0.6).
• Suggest reply — draft a reply as part of triage.

Tips

• Start with auto-apply off and watch its suggestions for a few days to build trust, then turn it on.
• Raise the confidence threshold if you see it applying changes you disagree with.
• Even with a real provider, the agent is wrapped so a provider outage can never block ticket creation.