Text to JQL

Inspiration

Jira is a necessary and multifunctional tool. Millions of users around the world have appreciated its benefits. But there is one nuance. Everyone who gets to know her for the first time is faced with three mysterious letters – JQL. It is a powerful and convenient way to search and filter the necessary information in Jira. But not all users have a technical background, and some have no idea how to simplify their lives with the help of these three letters.

Therefore, we wanted to create an app that allows you to communicate with Jira in a simple human language.

What it does

Our add-on after installation could be added to your Dashboard as a Text to JQL gadget. This AI/ML-based solution not just converts entered by the user natural-language text into a corresponding JQL query. It shows the search results on Jira's advanced search page in a 1-click.

How we built it

We started to play with Artificial Intelligence and Machine Learning technologies earlier this year. After the war in Ukraine began, we held our activities for about one month, but then resumed work. Now we're working on a big and complex solution for our new product for Marketplace, and "Text to JQL" technology is a part of it. As soon as Atlassian opened its yearly hackathon, we decided to provide this technology as a stand-alone product.

Challenges we ran into

Mixing technologies is always challenging, but adding Artificial Intelligence is the hardest we've ever integrated into our products.

Accomplishments that we're proud of

Add-on brought a qualitatively new experience for Jira users. We hope Atlassian will product fewer videos and documentation tutorials for users about how to use JQL :). And the users will be happier using something simpler to querying the data within Jira.

What we learned

Integration and utilizing AI/ML technologies is much more complicated than we estimated on a project start.

What's next for Text to JQL

This technology has significant potential, so we have many items on our roadmap. The quality of natural language conversion should be improved, including the ability to build complex JQLs based on lists, JQL functions, and dates.