💃 I can do it with a Rovo agent: Ask better questions, faster with Rovo

[#1]

Remember typing out the same support ticket questions over and over again? In 2014, that was my daily routine - and honestly, it wasn't fun. Fast forward to 2024, and I'm letting AI do it for me 🙌

Welcome to the first part of my "I can do it with a Rovo agent" series, where I'll show you how to transform your support workflow using Rovo, the Atlassian AI assistant.

🎬  If you're more of a visual learner, watch the video version of this article.

In this series, I'll show you how Rovo can help you:

• Ask better questions, faster

• Build your own AI support assistant

• Scale your support with Rovo

As support teams, we often receive tickets with minimal information. A classic example is the "it is not working" ticket. We need more details to effectively help our customers, but writing these follow-up questions takes time, especially when you're handling dozens of tickets per day.

What if you could save a few minutes per ticket? ⏰

Step 1: Start with Rovo Chat

Before building a Rovo agent, it's best to test your prompts in Rovo Chat. This allows you to refine your approach and ensure you're getting the exact output you need.

Step 2: Write your prompt

Here’s the prompt I started with:

You are an AI assistant helping an IT support team manage helpdesk tickets. You have access to the Confluence space named “IT Support” to help you understand the processes and troubleshooting steps.

Based on the customer’s request (description, forms, fields, summary, comments, etc.), generate a list of 3 targeted follow-up questions that will help the support team gather the information they need to resolve the issue.

Order the questions from most to least relevant and important.

Step 3: Refine the output

Your first try might not be perfect - and that's okay!

The key is to iterate on your prompt until you get exactly what you need. For example, my first attempt gave good questions but in the wrong order.

After refinement, here’s my final prompt:

You are an AI assistant helping an IT support team manage helpdesk tickets. You have access to the Confluence space named “IT Support” to help you understand the processes and troubleshooting steps.

Based on the customer’s request (description, forms, fields, summary, comments, etc.), generate a list of 3 targeted follow-up questions that will help the support team gather the information they need to resolve the issue.

Order the questions from most to least relevant and important.

The questions should:
- Clarify the problem, question, or need the customer is facing (e.g., when it started)
- Gather all relevant technical details (e.g., system information, error messages, or software versions)
- Understand previous troubleshooting steps (e.g., what the customer has already tried)
- Be clear, concise, and easy for the customer to answer
- Aim to be helpful for both the customer and the support team

The questions should not:
- Ask for details the customer already provided

The goal is for the support team to copy and paste the questions you provided. Be concise, do not explain your answer, and list your suggestions in bullet points using a regular font.

Next up: Turning your chat into an automated agent

Once you've perfected your prompt in Rovo Chat, you can use it to build a Rovo agent that automatically suggests follow-up questions on new tickets, saving your team time and ensuring consistency in your ticket qualification process.

In the next article, I'll show you how to transform this Rovo Chat prompt into a fully automated Rovo agent that saves your team time on every ticket.

I'd love to hear your ideas! Have you tried using Rovo for other aspects of support ticket handling? What prompts have worked well for you? Share your experiences in the comments below 👇

[Rodney Estrada]

Thanks @Manon Soubies-Camy . I appreciate the Rovo feature. I would like to see Rovo (or Atlassian Intelligence) focus on making Atlassian's Documentation & Knowledge Base more accurate and search friendly.

Instead of returning links, it should understand the question and return a natural language response that includes the answer, context and potential pitfalls.

Example

Me: Asking Rovo "Where do I find the import Id in Assets?"

Rovo: "The import Id, also known as the Configuration id, can be found in Assets by clicking Schema Configuration, Import tab and Read details for the specific import. NOTE: The import needs to be run once in order to create the Configuration id.

I hope others agree.