n8n AI-Powered Support Automation: Outlook + OpenAI + JIRA Ticketing

Email and Webhook dual-channel input → AI classification with structured parsing → auto-reply or JIRA ticket + notification

Who

Support teams with an n8n instance, Outlook, and JIRA, handling requests from both email and web forms

You get

AI classifies requests and decides auto-resolution; simple issues get instant replies, complex ones create JIRA tickets and notify assignees

You need

n8n · gmail

Steps

1. Configure dual-channel triggers

   Add two trigger nodes in n8n: Outlook Trigger (Microsoft Outlook, operation getAll, receivedAfter last 30 minutes) and Webhook Trigger (Webhook, path support, method POST). Connect both to the Normalize Data node.

2. Normalize and deduplicate

   Add a Function node (Normalize Data) and paste the code below to unify heterogeneous email and webhook fields into a standard structure. Then add a Remove Duplicates node using email+subject as the deduplication key.

   Execution prompt · Source text。Normalize Data1 node Function Code field

   Copy prompt

   const data = $input.all();
   const result = [];
   
   for (const item of data) {
     let normalized = {
       source: item.json.from ? 'email' : 'webhook',
       email: item.json.from?.emailAddress?.address || item.json.email,
       name: item.json.from?.emailAddress?.name || item.json.name,
       subject: item.json.subject || 'Web Request',
       message: item.json.body?.content || item.json.message,
       timestamp: new Date().toISOString()
     };
     result.push({ json: normalized });
   }
   
   return result;

3. Configure AI classification and structured output

   Add a Basic LLM Chain (AI Classifier2), connect an OpenAI Chat Model (gpt-4o-mini) and a Structured Output Parser. In the LLM Chain, set the Text field to the user message template and Messages to the system classification instruction; use the JSON Schema below in the Output Parser.

   Execution prompt · Source text。AI Classifier2 node Text field (user input template)

   Copy prompt

   Subject: {{ $json.subject }}
   Message: {{ $json.message }}
   From: {{ $json.name }} ({{ $json.email }})

   Execution prompt · Source text。AI Classifier2 node Messages → messageValues[0].message (system instruction)

   Copy prompt

   Classify this support request. Return JSON with:
   {
     "category": "technical|billing|account|feature",
     "urgency": "critical|high|medium|low",
     "expertise": "frontend|backend|support|business",
     "canAutoResolve": true|false,
     "isVIP": true|false
   }
   
   VIP indicators: enterprise email domains, mentions of "enterprise", "premium", urgent language.
   Auto-resolve: simple how-to questions, password resets, basic billing questions.

   Execution prompt · Source text。JSON Parser1 (Structured Output Parser) Input Schema field

   Copy prompt

   {
     "type": "object",
     "properties": {
       "category": {"type": "string"},
       "urgency": {"type": "string"},
       "expertise": {"type": "string"},
       "canAutoResolve": {"type": "boolean"},
       "isVIP": {"type": "boolean"}
     }
   }

4. Set up routing decision

   Add an If node (Route Decision1). The template leaves leftValue/rightValue empty; manually set the condition so that canAutoResolve=true routes to Auto Response, and false routes to Ticket Creation.

5. Configure auto-response branch

   On the true branch, add a Microsoft Outlook node (Auto Response1, operation sendMessage) to send the auto-resolution email. Then add a Respond to Webhook node (Webhook Response1) returning JSON: {"status":"success","message":"Request processed","ticketId":"AUTO_RESOLVED"}.

   Execution prompt · Source text。Webhook Response1 Response Body field

   Copy prompt

   {
     "status": "success",
     "message": "Request processed",
     "ticketId": "{{ $json.ticketKey || 'AUTO_RESOLVED' }}"
   }

6. Configure manual ticket branch

   On the false branch, add in sequence: a Function node (Assign Agent1) to assign an agent and compute SLA; a JIRA node (Create JIRA Ticket1) to create the ticket; and a Gmail node (Send Confirmation1) to notify the customer.

   Execution prompt · Source text。Assign Agent1 node Function Code field (replace [email protected] with real emails)

   Copy prompt

   const request = $json;
   const expertise = request.output.expertise;
   const isVIP = request.output.isVIP;
   const urgency = request.output.urgency;
   
   // Agent assignment logic
   const agents = {
     'frontend': '[email protected]',
     'backend': '[email protected]',
     'business': '[email protected]',
     'support': '[email protected]'
   };
   
   const assignedAgent = agents[expertise] || agents['support'];
   
   // SLA calculation
   const slaHours = {
     'critical': isVIP ? 1 : 2,
     'high': isVIP ? 4 : 8,
     'medium': isVIP ? 8 : 24,
     'low': 24
   };
   
   return [{
     json: {
       ...request,
       assignedAgent,
       slaHours: slaHours[urgency],
       priority: urgency === 'critical' ? 1 : urgency === 'high' ? 2 : 3
     }
   }];

   Execution prompt · Source text。Create JIRA Ticket1 Description field (Additional Fields)

   Copy prompt

   Customer: {{ $json.name }} ({{ $json.email }})
   Source: {{ $json.source }}
   Category: {{ $json.output.category }}
   Urgency: {{ $json.output.urgency }}
   {{ $json.output.isVIP ? 'VIP CUSTOMER' : '' }}
   
   Message:
   {{ $json.message }}

   Execution prompt · Source text。Send Confirmation1 node Message field

   Copy prompt

   Dear {{ $json.name }},
   
   Your support request has been received and assigned ticket #{{ $json.ticketKey }}.
   
   Priority: {{ $json.output.urgency }}{{ $json.output.isVIP ? ' (VIP)' : '' }}
   Assigned to: {{ $json.assignedAgent }}
   Expected response: {{ $json.slaHours }} hours
   
   We'll update you as we work on your request.
   
   Best regards,
   Support Team

7. Test and go live

   Save and activate the workflow. First send a test POST request via Webhook (fields in prompt), then verify Outlook email triggering. Check JIRA project key, Issue Type, labels, and priority mappings.

   Setup guidance · Editorial suggestion。Webhook test request example; paste into Postman/curl or a frontend form for end-to-end validation

   Copy prompt

   {
     "email": "[email protected]",
     "name": "Customer Name", 
     "message": "Support request text"
   }
   
   Optional fields:
   {
     "subject": "Custom subject",
     "priority": "high",
     "source": "website"
   }

You are reproducing n8n template 7407: AI-powered support automation with Outlook, OpenAI & JIRA ticketing. Proceed in this order: 1) Confirm you have credentials for Outlook, OpenAI, JIRA, and Gmail; 2) Import workflow.json or build manually per the steps; 3) Replace placeholder emails in Assign Agent with real team addresses; 4) Complete the Route Decision condition (suggest using canAutoResolve); 5) Adjust the JIRA project key and SLA hours; 6) Test with Webhook first, then enable Outlook triggering. All node parameters are governed by workflow.json; this site has not tested the import, so troubleshoot against the source template if issues arise.