Try out Jira’s new triage agent and share your feedback

[#1]

Jira Triage Agent is now available, and it’s about to save your team a ton of time. Instead of new work items sitting unlabeled, unprioritized, and unassigned in your backlog, the agent gets to work the moment a work item lands.

• Auto-categorize incoming items: The agent reads the summary, description, and context to set the right priority and labels automatically

• Route to the right owner: It identifies missing work item links and suggests who the work item should be assigned to. No manual triage required.

• Clean up your queue at scale: Especially powerful for high-volume scenarios like security vulnerabilities, support escalations, and bug reports

Watch this video for a walk through of the Jira Triage Agent and steps on how to get started:

Have questions or feedback? Drop a comment below and let us hear them. We'd love to hear how the Jira Triage Agent is working for you and any ideas you have for improvement.

[Jason Krewson]

Sounds cool, great job! 

Also the Loom video is set to private, I see the error "Sorry, Due to the privacy settings for this video, it cannot be played here at this time." 

[Braelyn Johnson]

@Jason Krewson - Thank you! Can you tell I shared this before having my morning coffee? :) The Loom video link is now updated! Take a look at it here! 

[__ Jimi Wikman]

A few things to unpack here...

Triage is for me associated with Incidents as part of incident management, but as the term is being used for everything these days as a way to be fancy about sorting and prioritising, I assume that this agent has nothing to do with Incidents, correct?

Is it just a helper for improving the quality of a work item in general?

The reason I ask is that I see several things in this video that are concerning me from an incident management perspective (and the fact that a backlog column is being used when backlogs are a built-in function in both board types...).

In the video, we go through a bug, which is a generic type, and I assume this is a defect rather than an incident. Prioritization is therefore different, and I want to make sure that for an incident, we get recommendations based on proper incident management.

For example, priority is based on Impact + Urgency connected to a system's business criticality (Tier). So recommendations would help us adjust the levels of Impact and Urgency, not suggest priority changes, as that is based on the priority matrix?

Incidents and defects both need steps to reproduce, which is the main focus for an incident triage: Verify that the incident or Defect can be reproduced. This is nothing that the agent can help with, I assume, or can we set some values that are required, such as starting point, browser, environment, and so on?

Can the agent connect this to test cases or test areas? Can it connect it to recent deployments? Can it connect to affected versions and so on?

We see the agent suggesting a change of work item type, but based on what? Bug, in my experience, is not an actual term that is used, as most mature organizations clearly differentiate between defects and incidents, and the term bug is used mostly by people as a general term for quality overall. So will this agent understand what the work items that exist are actually for, and if so, based on what? Description? 

If I am being honest, this agent feels like a crutch for organisations that do not have proper ways of working and where the users have not learned how to use the tool yet? I do not think a tool should also be the teacher, so is this agent meant as a temporary way to help people work properly, or is it to encourage dysfunctional ways of working because an agent will fix it for you?

If you create work items correctly, would this agent have any use?

[Josh]

Hi @Braelyn Johnson .

How close does this Jira functionality mirror what's in place for JSM? As the two products continue to diverge in UI and features, it's getting tough to tell what's common and what's not.

Is this feature designed in such a way that teams sharing a space will still be able to leverage it? A lot of the recent AI features that have been released appear to start from an assumption that there is a 1:1 mapping between teams and spaces. However, that's not always a safe assumption.

[James Rickards _SN_]

@__ Jimi Wikman , the word Triage come from the medical industry from late 18th century. It has been used for "prioritisation of new work" in business for many decades. I recall it being a well-established term when I started working 25 years ago in Australian Superannuation industry during the initial triage of mail arriving in the mail center.  Nothing new or fancy about it.

[__ Jimi Wikman]

@James Rickards _SN_ Yes, it was used in wartime to determine which patients would get treated and who would die. It was used before that, of course, and in a different way, like sorting wool or grading coffee beans, so yes, I understand it has other uses.

I just don't hear it in the areas I have worked professionally. This might be different in different areas of the world, but in the Atlassian platform, which has primarily been used for the build process of IT products and services, I have not heard anyone use the word triage for anything beyond incidents.

It is an old word, but in my experience, no one triages work items for development, test, requirements, design, deployments, service requests, or projects. We rather prioritise such requests based on value creation or time-based requirements. 

At least that is my experience here in Sweden.

@Josh that is a good point. For me, this agent does not fit in Jira, and it seems to be designed more from the perspective of JSM as "security vulnerabilities, support escalations, and bug reports"; to me, all should be in JSM and not in Jira.

I wrote a lengthy article about this, as the agent itself seems out of place and, rather than assisting in getting work done, it seems to be curling users instead of correcting bad work practices. I also think it is added to the wrong product, as this seems to be better placed in JSM...

[Dave Rosenlund _Trundl_]

Interesting, @__ Jimi Wikman. I hear the word "triage" used all of the time, in many geos. Mostly in a service management context, I agree, but also in software development when it comes to addressing bugs. 

That said, I have every reason to believe you when you say it may not be widely used on Sweden. 

[__ Jimi Wikman]

@Dave Rosenlund _Trundl_ I don't think I have ever heard anyone use the word in development settings, unless you have the incident flow there for some reason. In my experience, triage happens in JSM and incidents are then distributed to the teams to fix. There is no additional triage in incident management, so I am not sure what people would use that word for in Jira?

In development, you normally work with defects, which are a part of the development quality process. They are never triaged, they are fixed or de-prioritised as known issues to be fixed after production.

So, I guess the term is added to Jira when there is no JSM, and people try to do incident management in the development Jira instead. That makes sense, but I don't think that this agent is targeted towards that only?

It is interesting to hear how words are being used differently, especially when it comes to words that are pretty defined in incident management processes and things like that, but have no place in development processes (that I know of).

[Avinash Bhagawati]

Sounds cool @Braelyn Johnson !

Is this something can be extended to JSM and CSM as well?

[__ Jimi Wikman]

@Avinash Bhagawati I believe there already is one, or one is on the way. You can see a JSM variant briefly in the video if you look for it :)

[Ashik Muhammed _ AIO Tests]

This looks really useful, especially for teams that spend a lot of time reviewing incoming issues manually. Taking what @__ Jimi Wikman raised a bit further — when it's reading context to set priority and labels, can it also flag when a similar or duplicate issue already exists, or when the same defect touches multiple testing areas? That kind of signal would make a big difference for QA-heavy teams working at scale.

[Rune Rasmussen]

I like the idea of this, I think.
I like the idea of having quick and broad insights into where this work item may relate.

I don't necessarily think this should be scoped only to Jira Software though.
From where I'm sitting Atlassian seems to be the only ones that treat Jira Software, Service Management, and Product Discovery, as totally independent items.
In the real world, or at least the part I'm subjected to, it's all just Jira.
So having the Tirage agent in Jira Software, but not the rest of Jira, is just going to frustrate our users. Which then in turn quickly becomes another argument for switching to something else like Azure DevOps.

I also like that even in this demo Rovo keeps calling it an Issue. Something I would've expected Atlassian to have aggressively scrubbed from it's vocabulary.

And most of all, I like @__ Jimi Wikman's question of "...or is it to encourage dysfunctional ways of working because an agent will fix it for you?".