JSM Collaboration Assistant

Inspiration

From personal experience, when working in a company with multiple teams, or when joining a team for the first time, it can be hard to figure out if an issue that's raised is new of if there were similar issues that were raised before. JSM Collaboration Assistant aims to help solve that problem.

What it does

When a user creates a Jira issue(JSM request), it triggers a Jira automation which then triggers the Rovo agent to create a JQL query based on the issue's summary. The list issues and assignees of issues returned from the JQL request is then stored in the issue properties. The user is then able to link those issues as well as add those assignees as request participants.

How we built it

I have experience with forge and the stack used was pretty simple. Forge with custom UI frontend. Written in typescript.

Challenges we ran into

Tuning the AI. It is my first time working with AI, hence had some trouble getting it to work correctly. Honestly there's more improvements I can make to the prompt to get the AI to work better.

Accomplishments that we're proud of

My first solo hackathon! Also first experience with AI prompt engineering.

What we learned

Learn primarily about how to write prompts.

What's next for JSM Collaboration Assistant

Tuning the prompt to make the search more reliable and cover a wider range of scenarios.
- Better synonym support
- Support searches in issue description
Feature to send a message or tag in a comment.
UI improvements.
Fixing bugs.