Fire Your Jira Admin – Automate Everything with Rovo AI and Forge

If you clicked on this article hoping to discover how to completely replace your Jira admins with Rovo AI and Forge, I've got news for you—you've been successfully jebaited!

While your Atlassian admins aren't getting replaced by AI anytime soon, what I can offer is something potentially more valuable: a practical guide to automating those mind-numbing permission tasks that consume valuable IT resources. By the end of this article, you'll learn how to free your Jira admins from mundane tasks so they can focus on more creative and complex challenges.

But first, let me share a story that fueled my passion for this particular automation challenge.

My Administrative Nightmare (That Started It All)

It was late 2019—a simpler time before GPT revolutionized how we think about AI. I was working in DevOps for a large organization, primarily managing Atlassian products and automating processes across our tech stack.

Our small support team of 10 members handled all queries from internal development teams. Customers would either raise tickets through ServiceNow or contact team members directly via MS Teams or email if they knew who could solve their problem.

Then came an unexpected opportunity: my manager asked if I'd lead the support team while the current lead took six months of maternity leave. With my experience and system knowledge, I was the natural choice—especially since the entire team consisted of newcomers with less than two years of industry experience.

Initially reluctant, I eventually agreed, viewing it as a chance to develop my leadership skills. I threw myself into the role, determined to excel. I spent countless hours outside work understanding every aspect of the support process—the types of requests, resolution steps, KPIs, OKRs, and SLAs.

The Revelation in the Data

After exporting ticket data from ServiceDesk and conducting a quick exploratory data analysis, I discovered something striking:

• 70% of all tickets were Jira-related

• Of those Jira tickets, a whopping 40% dealt with user permission problems

These permission issues ranged from requests for access to new Jira projects to troubleshooting existing roles when users couldn't perform certain actions. Resolution times varied dramatically—from 20 minutes to 8+ hours—depending on when the ticket arrived, whether the request details were clear, and if support agents were available across different time zones.

The permission structure itself wasn't complicated. We had configured Jira projects with a simple permission scheme using roles to map different scopes. Users were added directly to projects with appropriate roles. Almost all projects used the same permission scheme, but they were restricted with Browse permission scope, preventing anyone without the appropriate role from viewing or modifying issues.

I began wondering: could we automate this entire process and eliminate 40% of our tickets?

The Solution That Never Was

After a month of intensive research and consultations with experts, I designed a comprehensive solution centered around a conversational AI chatbot. Here's what the architecture looked like:

The solution included:

• User Authentication via Azure AD SSO: Seamless login through existing credentials

• Multiple Communication Channels: Integration with MS Teams and Email

• Bot Connector: A unified interface connecting all channels to the bot engine

• Bot Engine: Managing conversational flow and user engagement

• Azure Key Vault: Secure storage for API keys and credentials

• Azure Storage: Maintaining conversation state and session data

• Cognitive Services: Intent recognition using LUIS and QnA Maker

• Application Insights: Real-time monitoring and performance tracking

• ServiceNow Integration: Automatic task creation for complex queries

• Process Automation: API Gateway connected to UI Path for RPA tasks and other product services.

Excited about the potential impact, I prepared a detailed presentation for my manager—complete with problem statement, architecture design, process flows, and projected savings.

The Corporate Reality Check

My manager's response left me stunned. After acknowledging my effort, he revealed why we couldn't implement it: the support team's annual budget increases were justified by the rising ticket numbers. By reducing tickets by 40%, we would undermine our case for additional funding—and the revenue from implementing the automation wouldn't come close to matching our current growth projections.

I was devastated. Not just for myself, but for all the junior team members condemned to these mind-numbing tasks. Eventually, I accepted reality and returned to my previous role when the team lead returned from leave.

Enter Rovo: A Second Chance at Innovation

Fast forward to 2024—Atlassian announced Rovo, and I immediately recognized my opportunity to scratch that years-old itch. With Forge accelerating development, I could finally build what I had envisioned years ago.

Rovo agents built with Forge provide two key modules:

• rovo:agent: AI teammates integrated into Jira and Confluence workflows

• action: Specific tasks the agent can perform, like calling APIs or running code

My vision was clear: create an agent that would solve access and permission problems for end users through two conversation starters:

1. Add to project access role: Direct request pathway for project access

2. Troubleshoot project permission problems: Conversational problem-solving leading to role adjustment

The Technical Implementation

Here's how our Rovo agent definition looks in the app's manifest:

rovo:agent:
    - key: access-o-bot
      name: Access-o-Bot
      description: An agent for solving all your access and permission related problems in Jira.
      prompt: resource:agent-resource;prompts/agent-prompt.txt
      conversationStarters:
        - Add to project access role
        - Troubleshoot project permission problems
      actions:
        - search-projects
        - get-project-role-permissions
        - add-project-access
        - get-user-access

The agent includes four critical actions:

• search-projects: Finds Jira projects based on user input

• get-project-role-permissions: Provides available role information

• add-project-access: Adds users to projects with specific roles

• get-user-access: Fetches current user access roles

The Conversation Flows

Let’s see our Agent in action

The Impact: From Vision to Reality

What began as a frustrated DevOps engineer's dream during pre-AI times has become a practical solution that:

• Reduces support ticket volume by automating common access requests

• Empowers users to solve their own permission problems

• Frees IT support teams to focus on more complex challenges

• Provides immediate resolution instead of multi-hour wait times

• Creates a consistent and user-friendly access management experience

While we can't replace Jira admins (nor should we want to), we can certainly make their lives easier—and our organizations more efficient—by intelligently automating the routine tasks that consume so much of their valuable time.

Have you implemented similar automation in your organization? I'd love to hear about your experiences in the comments below.