AI-Powered Jira Service Management Automation with Make.com

Introduction

In many support teams, Jira Service Management (JSM) tickets can sit idle for minutes or hours before the first response — especially when volume spikes. Customers quickly grow frustrated by slow acknowledgments, even when the issue itself takes time to resolve.

For a recent client project, we solved this by introducing AI-driven automated responses powered by Make.com and OpenAI, seamlessly integrated into Jira Service Management. The goal: simulate a human-like reply from the support agent, 10–15 minutes after ticket creation, keeping engagement high while maintaining authenticity.

This article walks you through how we designed, built, and deployed this automation — step-by-step — so you can replicate it for your own JSM environment.

---

Prerequisites

Before you begin, ensure the following:

• Jira Service Management (Cloud) with admin access
• Make.com account with a scenario editor
• OpenAI API key (via Make.com or direct HTTP module)
• API token for the Jira user who will post replies
• Permissions to create webhooks in Jira
• Basic understanding of JQL (Jira Query Language)

---

Step 1: Triggering the Automation via Jira Webhook

The process starts when a new ticket is created in your JSM project — in this example, the RS project.

1. Go to Jira Settings → System → Webhooks
2. Create a new webhook called New Ticket Listener
3. Set the JQL filter to define which tickets should trigger the automation:

project = YOURPROJ

✅ Pro Tip: You can expand this to multiple projects or issue types later using advanced JQL filters like:

project in ("Your Project Name", "Another Project") AND issuetype = "Bug"

Paste the Make.com Webhook URL (you’ll get this in the next step).
Select “Issue Created” as the event trigger.
Once saved, every time a new ticket is created, Jira sends a JSON payload containing key fields such as summary, description, reporter, and timestamps to Make.com.

---

Step 2: Receiving and Processing Ticket Data in Make.com

Inside Make.com, create a new Scenario and follow these steps:

1. Receive the Webhook Payload
   Add a Webhook module and select Custom Webhook.
   Make.com will generate a unique URL that you’ll paste into Jira’s webhook settings.
   Once connected, trigger a test by creating a sample issue in Jira — this helps Make.com detect the payload structure.
   

2. Add a Randomized Delay
   To make the response look natural, introduce a random wait time between 10–15 minutes:
   Use a “Set Variable” module to generate a random number:

random = 10 + floor(rand() * 6)

Then use a Sleep/Delay module:
Wait random minutes
This gives each ticket a unique and realistic delay before the reply is sent.

3. Generate an AI-Powered Response
   Next, pass the ticket details to OpenAI’s API using a Make.com OpenAI module (or an HTTP request if you want more control).
   Prompt example:

You are a Jira Service Management support agent named Support $Agent Name.
Analyze the following ticket and craft a helpful, empathetic response.
Prefer the description over the summary if both are provided.
Ignore test or placeholder content.

Ticket summary: {{summary}}
Ticket description: {{description}}
Customer name: {{reporter.displayName}}

This ensures context-aware replies tailored to the issue.

4. Post the Response Back to Jira
   Finally, use the Jira Cloud API (POST comment) module to send the AI-generated text back to the issue.
   Endpoint example:

POST /rest/api/3/issue/{{issue.key}}/comment

Make sure to authenticate with the API token of the designated support user (e.g., example@domain.com).
This makes the comment appear as if Support Agent Name personally responded to the customer.

---

Step 3: Customization Options

🔍 Expand Your JQL Filters
You can tailor which projects, issue types, or request types trigger automation:

• Multiple projects:

project in ("Your Project Name", "Another Project")

• Specific issue types:

project = "Your Project Name" AND issuetype = "Incident"

🧠 Modify the AI Behavior
Edit the OpenAI prompt in Make.com to adjust the tone or depth. For instance:

• Add a “technical deep dive” instruction for advanced users
• Include canned signature lines or SLAs dynamically

🔄 Change the Reply Account
To switch the responding user:

1. Create a new API token under the desired Jira account
2. Update the Make.com HTTP module authentication settings

---

Step 4: Maintenance and Monitoring

• Execution History: Every Make.com run logs inputs, outputs, and errors — perfect for debugging.
• Error Handling: Use Make.com’s “Error Handler” routes to retry failed calls or notify admins via Slack.
• Disable/Enable Anytime: You can toggle the webhook in Jira or pause the scenario in Make.com without losing configuration.

---

Step 5: Limitations and Best Practices

Area	Recommendation
Webhook Filtering	Avoid overly complex JQL to prevent performance issues.
Delays	Chaining multiple delay modules can simulate >10 min waits safely.
AI Quality	Keep prompts short and context-focused — avoid vague summaries.
Security	Store API tokens securely and rotate them regularly.
Testing	Use a non-production project first to validate output tone and accuracy.

---

Real-World Impact

For the support team, this automation reduced average first-response time by 83%, while maintaining a human-like tone in replies. Agents could focus on actual troubleshooting, while customers received immediate acknowledgment and reassurance. The workflow proved especially valuable during off-hours and peak load periods — effectively extending team presence without additional headcount.

---

Final Takeaway

AI-powered automation isn’t about replacing your support team — it’s about augmenting human response with smart, context-aware systems. By integrating Make.com, Jira Service Management, and OpenAI, you can deliver faster, smarter, and more consistent support experiences.

👉 Contact me for expert Atlassian consulting
Subscribe for more Atlassian ideas and real-world tips and strategies.