Discover why Atlassian customers win in the AI service era

[#1]

Hi Community! 👋

Welcome to Week 1 of Shatter the Service Quo: Break Stuff, Earn Badges! 🎉

Be honest – how much of your service workflow still looks like log it, route it, wait, report on it?

That playbook made sense when work moved slower and lived in one system. But now your team is fielding requests across Slack, Teams, email, and portals while managing code deploys, SaaS sprawl, and a backlog that never quits. The old queue-and-form model wasn't built for this.

We just published a white paper that digs into what AI-native service actually looks like in practice – not AI bolted onto legacy tools, but service rebuilt from the ground up around context, teamwork, and experience.

👉 Read the full paper, or if you want the TL;DR, we’ve shared what stood out to us below.

The challenge

We’d love to hear your perspective after reading! Comment below and you’ll unlock:

• 🏆 Service Disruptor badge

• 💰100 Kudos points

                                                             

Here are a few prompts to get you started:

• Where are you on the AI spectrum: still stuck in log-it-route-it-wait, or already experimenting with AI-native approaches?

• If you could fix one “broken” thing about your service experience with AI, what would it be?

🔥 The big shifts happening right now

1. From reactive queues → proactive resolution

The goal isn't faster ticket response – it's fewer tickets in the first place. When AI is grounded in a living map of your people, tools, work, and their relationships – not just a static CMDB – issues can be detected and resolved before someone has to file a request. That's what the Teamwork Graph gives Atlassian customers: the context AI needs to answer better, act smarter, and orchestrate across tools.

2. From siloed teams → human + AI teamwork

Great service isn't one team's job. It's Dev + IT + Business + Customer Support on one platform – and now, AI agents like Rovo Service joining as teammates that learn, act, and escalate with humans still in the loop. That's Human+AI Teamwork: not AI replacing you, but AI thinking with you. Think less "ticket hand-off" and more "collaborative resolution."

3. From ticket metrics → experience outcomes

The win isn't "we closed 500 tickets this week." It's fewer dead ends, fewer form fills, fewer transfers – and more journeys that feel intuitive and human for the people you're serving (whether that's employees or customers).

[Tomislav Tobijas]

We're definitelly exploring what's possible with AI, but I see it as 'long road ahead' until we actually implement something. I've heard some blockers that our colleagues have when it comes to going back and forth with our/their customers, and I could maybe see CSM as a potential solution 🤔 The goal is to try to build something next quarter and see if it can be helpful.

From my experience, people tend to prompt AI more than they would actually raise a ticket, so, apart from fewer tickets, it might be that users/customers get 'more help' from solutions like these. Well... at least that's what I think or hope for 😅

[John Funk]

In order for AI to be truly successful at this, you must have a robust source of truth and data - a really great knowledge base if you will. The AI Agent is only as good as what it has access to. 

Having said that, I have found the Atlassian AI agent to be very helpful when reporting a problem. Maybe 40% of the time I am able to solve the issue without having to submit an actual ticket. That's helpful for me because I get a solution quicker. And helpful for Atlassian because there is one less ticket to have to triage and respond to. This will only get better in my opinion. 

[Marcell Bendik]

I believe having a high maturity level ITIL solution with proper JIRA support from before the AI age is still great today. Adding AI would further enhance it. However, reaching high maturity level is a very challenging task we rarely succeed.

So the question for myself, can we get to a higher level faster with AI? Can I skip a level with the help of AI? Like without proper documentation it is hard to train people. However, AI can actually put things together from 1000s of tickets feeling like there is a documentation. That is great shortcut. So where are the shortcuts?

[František Špaček _MoroSystems_]

I still think that having a good backend, functional ITIL porcesses and also good maturity in the mindset itself is first needed step. Otherwise we would be implementing processes where people do not understand what AI is doing and will only demand to have "AI automatically solve everything". I still think that human will have to stay in the loop and that for me means humans will have to understand the process either way. Processes will look a bit different from what we are used to now for sure, but if we leave e.g. lower priority Incidents to be solved by AI, will agents be able to solve higher priority incidents if AI fails to solve it? I think leaving everything up to AI means loosing ways how to teach your team by doing things and I am afraid that at certain point some people or even whole teams might lose some abilities in the long term and will be back at the beginning.

[Sam Okell]

We are still starting out on our AI journey, but one point that we have started looking at is automated ticket triage.

Due to various factors - different teams, different requests, different knowledge / skill sets - it's increasingly difficult to have a single triage process that covers everything without it being a basic set of guidelines.

From our initial review of AI, we are looking at having Rovo review each ticket when it's raised and suggest items that seem to be incorrect, for example if a ticket is raised and the Description says 'I need access to ABC', but it's been raised as an Incident, then it should flag it to the agent and the agent should then take appropriate action.

This should be built on top of the ITIL framework so we are 're-inventing the wheel'!

[gunjanahuja]

Good observations 

[Keiran]

Working in enterprise IT at a university, the focus on outcomes over outputs is especially relevant. We've measured success through ticket volume, response times, and SLA adherence for years. Those metrics are useful for operational visibility but they don't always reflect the actual employee or faculty experience.

The phrase that stuck with me was "service so good you don't even know it's there." That's a tough target but it's the right one.

Whether we're supporting faculty, staff, or students, the future probably isn't about processing tickets faster. It's about creating experiences where support is embedded into daily workflows and delivered before users need to ask for help. We've started moving in that direction with automation and proactive alerting, but there's a lot of room to grow.

[Brita Moorus]

I'm still in the "exploring what is realistically useful" phase with AI in service management, but automated triage and better context are the areas that feel most promising to me. 

If AI could help with:

• suggesting the right request type or category

• spotting missing information before the ticket reaches an agent

• linking relevant knowledge base articles, Assets objects, previous incidents, or related changes

• summarising the issue clearly for the next team

…that would already remove a lot of friction.

But I agree with others here: AI needs a solid source of truth. Without good knowledge, service data, and ownership, it can easily become “confident guessing as a service” 😄

So my ideal AI service experience would be: fewer forms for users, better context for agents, and humans still clearly in control of the process 💙

[Anwesha Pan]

We have been using and exploring AI. Mostly using Rovo agents for automation, triaging and creating Release notes and helping with KB wiki content and creation.