Ticket Triage for Jira Service Management with Gemini AI Audit and Guidance

An extendable triage workflow that classifies severity, sets components, and posts actionable guidance for support engineers using n8n + Gemini + Cache Augmented Generation (CAG). Designed for Jira Service Management, but easily adaptable to Zendesk, Freshdesk, or ServiceNow.

Description
Support teams loose valuable time when tickets are misclassified: wrong severity, missing components, unclear scope. Engineers end up re-routing issues and chasing missing info instead of solving real problems.

This workflow automates triage by combining domain rules with AI-driven classification and guidance, so engineers receive better-prepared tickets.

It includes:
✅ Real-time ticket capture via webhook
✅ AI triage for severity and component
✅ CAG-powered guidance: 3 next steps + missing info
✅ Internal audit comment with justifications & confidence
✅ Structured metrics for reporting

⚙️ How It Works
This workflow runs in 4 stages:

📥 Entry & Setup
Webhook triggers on ticket creation
Loads domain rules (priority policy, components, guidance templates)
Sets confidence threshold & triage label

🧠 AI Analysis (Gemini + CAG)
Builds structured payload with ticket + domain context
Gemini proposes severity, component, guidance, missing info
Output normalized for safe automation (valid JSON, conservative confidence)

🤖 Update & Audit
Updates fields (priority, component, labels) if confidence ≥ threshold
Posts internal audit comment with:
3 next steps
Missing info to request
Justifications + confidence

📊 Metrics
Captures applied changes, confidence scores, and API statuses
Enables reliability tracking & continuous improvement

🌟 Key Features
CAG-powered guidance** → lightning-fast, context-rich next steps
Explainable automation** → transparent audit comments for every decision
Domain-driven rules** → adaptable to any product or support domain
Portable* → swap JSM with Zendesk, Freshdesk, ServiceNow* via HTTP nodes

🔐 Required Credentials

| Tool | Use | |------|-----| | Jira Service Management | Ticketing system (API + comments) | | Google Gemini/Gemma | LLM analysis | | HTTP Basic Auth | For Jira API requests (bot user) |

⚠️ Setup tip: create a dedicated bot user in Jira Service Management with an API token.
This ensures clean audit logs, proper permissions, and avoids mixing automation with human accounts.

🧰 Customization Tips

Replace https://your-jsm-url/... with your own Jira Service Management domain.
Update the credentials with the bot user’s API token created above.
Swap Jira Service Management nodes with other ticketing systems like Zendesk, Freshdesk, or ServiceNow.
Extend the domain schema (keywords, guidance_addons) to fit your product or support environment.

🗂️ Domain Schema

This workflow uses a domain-driven schema to guide triage.
It defines:
Components** → valid areas for classification
Priority policies & rules** → how severity is determined
Keywords** → domain-specific signals (e.g., “API error”, “all users affected”)
Guidance addons** → contextual next steps for engineers
No-workaround phrases** → escalate severity if present

✨ The full domain JSON (with complete keyword & guidance mapping) is included as a sticky note inside the workflow.

💡 Use Cases
Automated triage for IT & support tickets
Incident classification with outage/security detection
Contextual guidance for engineers in customer support
Faster escalation and routing of critical issues

🧠 Who It’s For
Support teams running Jira Service Management
Platform teams automating internal ticket ops
AI consultants prototyping practical triage workflows
Builders exploring CAG today, RAG tomorrow

🚀 Try It Out!
⚙️ Import the Workflow in n8n (cloud or self-hosted).
🔑 Add Credentials (JSM API + Gemini key).
⚡ Configure Setup (confidence threshold, triage label, domain rules).
🔗 Connect Webhook in JSM → issue_created → n8n webhook URL.
🧪 Test with a Ticket → see auto-updates + AI audit comment.
🔄 Swap the Ticketing System → adapt HTTP nodes for Zendesk, Freshdesk, or ServiceNow.

💬 Have Feedback or Ideas? I’d Love to Hear
This project is open, modular, and evolving. If you try it, adapt it, or extend it, I’d love to hear your feedback — let’s improve it together in the n8n builder community.

📧 ascuncia.es@gmail.com 🔗Linkedin