3. Answering repetitive questions
Filling out request forms on behalf of customers
Going back and forth with customers on a ticket
Catching up on long comment threads
Drafting KB articles
Creating a change request
Tracking assets in a spreadsheet
Re-routing tickets
Setting up a service project
*POOF*
If you could wave a
magic wand to make a service
management task disappear, which
task would you choose?
9. GENERATIVE AI
Summarize issue details
Adjust tone
Generate knowledge base articles
Group similar incidents
Surface relevant context
Automated virtual agent support
ASSISTIVE AI
10. GENERATIVE AI
Summarize issue details
Adjust tone
Generate knowledge base articles
Group similar incidents
Surface relevant context
Automated virtual agent support
ASSISTIVE AI
11.
12. Your data is safe.
The data you submit and the responses you receive via Atlassian Intelligence are used
ONLY to serve YOUR experience.
You’re in control.
You have the power to decide when, where, and how you use our AI features.
Atlassian Intelligence is built for
trust
26. Configuring the virtual agent
Use generative AI to search
across your linked knowledge
base spaces and generate
responses.
INTENT FLOWS
AI ANSWERS
27.
28.
29.
30.
31. Configuring the virtual agent
Use generative AI to search
across your linked knowledge
base spaces and generate
responses.
INTENT FLOWS
AI ANSWERS
32. INTENT FLOWS
Craft a custom conversation
progression based on help
seeker responses to the virtual
agent.
Configuring the virtual agent
Use generative AI to search
across your linked knowledge
base spaces and generate
responses.
AI ANSWERS
44. INTENT FLOWS
Craft a custom conversation
progression based on help
seeker responses to the virtual
agent.
Configuring the virtual agent
Use generative AI to search
across your linked knowledge
base spaces and generate
responses.
AI ANSWERS
46. BEST FOR REQUESTS THAT…
Are covered by your knowledge base
Can be resolved with instructions
Don’t require a human agent
INTENT FLOWS
AI ANSWERS
BEST FOR REQUESTS THAT…
Require guidance or troubleshooting
Require collecting info, action, or triaging
May require a human agent
47. Customer
request
Intent
Matching
Enter matched
intent convo flow
Send generated answers
derived from customer’s
KB
How do they come together?
IF: CONFIDENCE IS HIGH
Matches to a well defined intent
IF: CONFIDENCE IS LOW
Check if it is answerable by KB
56. How to get started?
New teams often don’t know
which request types to use
What other use cases?
Existing teams may not realise
which other request types they
might consider
Need ideas for request types?
57.
58.
59.
60. AI answers
AI issue summaries
Virtual agent
AI project configuration
Generative AI in the
editor
We’ll get into demoing some of these innovations shortly, but first we’d like to tell you a bit more about our approach to AI and demystify some of the magic behind it.
Going back to my question, I imagine the tasks that first came to mind for a lot of you are probably those that are tedious, or relatively manual, or time consuming…and generally tasks that don’t add a ton of business value or feel like a great use of your skills
<click>
More specifically, things like answering the same questions over and over again, or writing and re-writing a customer response to strike the right tone. But what if there actually was a way that you could eliminate some of those tasks from your plate?
Atlassian has introduced Atlassian Intelligence, which is a collection of AI-powered capabilities built into the Atlassian Platform.
Obviously, AI is a hot topic these days, but what’s unique about here is that Atlassian AI models are layered on top of what the Atlassian Teamwork Graph, which is made up of over 20 years worth of connected insights into how teams operate and how work flows across an organization and across an ecosystem of tools. Because of this, Atlassian Intelligence is able to provide a differentiated, native AI experience that’s contextual to you, your teams, and your workflows.
Our models embrace two forms of AI. The first is assistive AI which enhances your human abilities and helps improve task efficiency and decision-making <click>
Within JSM, that includes things like grouping similar incidents or surfacing of relevant past issues or confluence articles. It also includes parts of our new virtual agent feature, which we’ll demo shortly, like the ability to capture issue context and seamlessly hand it off to a human agent. <click>
And then there’s generative AI which you’ve probably heard a lot of buzz about recently, Generative AI acts like an additional member of your team. <click>
There’s a number of new capabilities in JSM powered by Atlassian Intelligence that can create and transform content to reduce cognitive load on your team, doing things like summarizing information and generating customer responses. <click>
<click>
While it’s exciting to think about the potential of these AI powered capabilities, with these possibilities comes a lot of uncertainty. There is a lot of pressure these days to be infusing everything we do with AI, but most people don’t even really understand how it works or how their data is being used.
We understand some teams have reservations about adopting AI, and that’s why trust is at the forefront of everything we’re doing with Atlassian Intelligence. That means:
We encourage you to check out the AI trust center which we’ve linked here. It provides more insight into Atlassian’s approach to AI and how they responsibly handle your data.
So with that, let’s see some of the newest AI features available in JSM.
For today’s demo we’re going to walk through capabilities across the platform that help you to…
[CLICK] Unlock answers to deliver instant, always on support
[CLICK] Accelerate time consuming tasks and free up your time to focus on high value work.
[CLICK] And streamline the process of getting started with a new service project, particularly for non-technical teams.
Before I show those capabilities, one quick thing to note: in order to use some of the features we’re about to show, org admins need to first activate Atlassian Intelligence for Jira Service Management.
This is easy to do. Org admins simply need to navigate to admin.atlassian.com, go to Settings at the top, then find the Atlassian Intelligence section on the left, then <CLICK> “Select products to activate”
… <click> select Jira Service Management,
… along with any other Atlassian products you wish to enable Atlassian Intelligence for.
<CLICK>
… then hit next
Review the terms,
check out trust standards that we referred to earlier.
<CLICK>
And hit save…
And that’s it, <CLICK> you’re done.
You’ll now be able to take advantage of powerful Atlassian Intelligence features throughout the products you selected.
With that out of the way…
Let’s talk about our first set of features: the virtual agent and AI answers
Jira Service Management already has conversational ticketing which allows you to capture, manage, and resolve requests directly from your favorite chat tools.
This provides a great experience for help seekers…
It is the tool they use everyday and it provides immediacy…
but it is difficult to scale if your team is resource constrained.
This is why we have the virtual agent…
The virtual agent automates support interactions in your chat tools to free-up agent time to focus on higher value work and help teams deliver exceptional service at scale.
I’m excited to share that Virtual Agent is generally available in Slack.
There are two main ways that you can configure the virtual agent, depending on the type and complexity of requests you’re looking to automate.
You can use either or both of these to help deflect tickets and deliver fast support to your customers.
… We’ll start with Atlassian Intelligence Answers, or AI answers, which uses generative AI powered by Atlassian Intelligence to search across your linked knowledge base spaces to generate responses to customer questions.
As mentioned earlier this feature requires you to activate Atlassian Intelligence in order to use it.
Let’s take a look at how it works.
This team has built a robust knowledge base to handle common IT problems that users and admins at the organization encounter, like troubleshooting laptop malfunctions and getting access to the Wi-fi network. Despite having these articles, the IT team wastes a ton of time re-directing people to troubleshooting resources or answering questions that are already answered by the KB articles.
Let’s see how an agent can use AI to solve this challenge. He heads over to the virtual agent and to start out<CLICK> he needs to follow the prompts to connect the virtual agent to his team’s Slack instance.
Once that’s done and since Atlassian Intelligence is already activated for their site…
From the virtual agent settings menu he can toggle on the Atlassian Intelligence answers feature. <CLICK>
So now, when a help-seeker asks a question that is answered in a KB article, this is what will happen… <CLICK>
A user is having trouble connecting to the school’s VPN
<CLICK>
He can ask for help in the IT support channel
The virtual agent responds immediately
Without agent intervention, the virtual agent searches across the IT team’s knowledge base for information about VPN issues.
It generates a response to his query along with links to relevant articles.
In this case, it links to a troubleshooting article that looks perfect.
Think about what just happened here. Headmaster Farquhar simply flipped a couple of switches and now help seekers get immediate answers in the tool they use every day — no agent required.
AI answers is a great option for getting started quickly with the virtual agent.
You can deflect help seeker questions just by leveraging your knowledge base
You can also configure the virtual agent with intent flows. These are custom, pre-defined conversation progressions that are dependent on help seeker responses to the virtual agent.
You do NOT need to activate Atlassian Intelligence in order to use virtual agent intent flows.
Let’s take a look at how you build these…
In this example, the IT department is looking to automate more interactions with their service desk, beyond what can be handled by AI answers.
Our agent goes into the Intents section of Virtual Agent and <CLICK> selects create intent…
To start creating, you can browse from a list of templates that come pre-populated
There are standard templates based on common IT requests…
and if you have AI enabled, you’ll see some suggested templates that are based on recent ticket data in your project
<CLICK>
In this case, the agent selects a template from his existing data — Password Reset, he knows that he receives this a lot because the system has arcane password requirements so students keep forgetting.
<CLICK>
This pre-populates the basic intent settings with a name and description.
The next step is configuring training phrases…
The goal here is to help train the virtual agent by entering a variety of phrases people might use when making this type of help request.
Because we used a template we already have some pre-filled, but you can add additional ones, that are particular to your organization. In this case he might add some that use the lingo common to the organization.
Don’t worry you can refine this list after the fact too. Right now, he’s happy so he can hit create…<CLICK>
So the next step is building a flow to guide the virtual agent on how to handle this type of intent. The flow builder requires no code, and allows you to build a back and forth conversation, where you you specify the different actions the virtual agent can take, including providing choices for the customer to choose from, asking for information that can be used to populate a field later in the flow, sending a web request to automate an action, and more.
In this case, the agent wants to guide someone to reset their password.
I’ve fast forwarded a bit to take a tour of the flow he’s created.
You can see where the agent has added choices to ask the help seeker whether they’ve already tried to recover it,<CLICK>
Based on that question, he can unlock different paths to the conversation. <CLICK> If the help seeker says “YES” do this, “NO”, do that
In the case, he sends a message back to the help seeker…
That great for setting a conversational flow, but what about if you actually want to take action?
Well you can use Virtual Agent to hook into remote services to truly automate and deliver service end to end
<CLICK>
Sending a web request, where you can pass any info you want…<CLICK>
Then based on the response, you can take different actions.<CLICK>
In this case, it was successful so you can let the help seeker know and resolve the request…
In this case, it escalates the request to an agent…
Now, let’s see how it works for a help seeker <CLICK>
Poor Nick is having a lot of IT issues today. <CLICK>He asks his question…
Notice that his question doesn’t directly mention “password reset”…
But the intent matches anyway because of the AI model
Nick’s choices guide through the flow, until his issue is successfully resolved.
Again step back and consider what happened here. Like magic!
Over time as your customers use the virtual agent…
You’ll be able to get insights for what help seekers are asking about, giving you the opportunity to refine the flows and training phrases.
So we just talked about two ways to configure the virtual agent… let’s talk about when to use each
Which one should you use?
[CLICK]
Both!
It is when you combine them that you unlock the true power of Virtual Agent
What kinds of requests are each best at?
• AI answers is the perfect way to get started quickly with the virtual agent – it only takes a few minutes to set up and immediately start seeing value.
• Covered by your knowledge base and can be resolved by instructions. The help seeker can solve it without an agent. <CLICK>
• Intent flows require a bit more setup than AI answers, but are great for more complex issues that tend to require guided work or troubleshooting; information collection and triaging, or an automated action. Think things like software access requests. incident reporting, procurement requests, onboarding workflows, etc. Intent flows are also great if you’re team isn’t quite ready to activate Atlassian Intelligence.
It is important to understand how they come together.
When a customer sends a message, the virtual agent will always first try to match it to an intent. If no intents match – either because you’ve chosen not to create any, or none of your existing ones are appropriate – it’ll then try to answer the question using Atlassian Intelligence
That’s why we recommend first starting with AI answers… leveraging your knowledge base.
Then add intent flows to cover specific use cases.
So that’s unlocking answers, with virtual agent and AI answers
* Virtual Agent is generally available in Jira Service Management.
* AI answers, which draws on your knowledge in JSM or Confluence, is in beta
Atlassian is building this offering for Microsoft Teams. You can watch for updates on this in the public cloud roadmap.
Next up is handful of features in the web issue view to accelerate work and improve your ability to quickly respond to and resolve requests or incidents.
Here Sarah has broken her laptop. She thinks she needs a new one, but IT wants to try to fix her existing one first.
Professor Wong was handling it, but it’s the end of the day for him, so he's assigned it to his teammate Professor Abraham who’s just starting her day in a different part of the world.
Instead of spending time reading the entire conversation thread trying to get up to speed
[CLICK]
Professor Abraham can now get a clear, concise summary of the ticket and all the discussion… including what troubleshooting steps have already been sent…and this summary is entirely generated by Atlassian Intelligence.
In this case, Professor Wong suggested a cleaning spell, which apparently backfired, and Sarah has an exam coming up.
Now, Professor Abraham has decided to send Sarah a new laptop and has started writing a reply
<CLICK>
With Atlassian Intelligence, given the slight tension in the situation, Professor Abraham can quickly adjust the tone of her response to handle the situation with the right amount of empathy and care needed.
While Atlassian Intelligence is always there to help, you are always in control. We know it’s not going to be right 100% of the time.
So, here Professor Abraham is able to insert the text…then make a quick edit before responding…to make it perfect. With Atlassian Intelligence she can be thoughtful without sacrificing speed or accuracy.
Bonus, generative AI can also help you create knowledge base articles, saving you more time.
1. Create a new article
2. Bring up Atlassian Intelligence via the toolbar or via the slash menu.
3. You can then write your own prompt… in this case, what school supplies would a first year magic student need?
4. It gets to work and comes back with a draft which you can use as a starting point.
5. Then you can refine as you’d like.
That’s AI issue Summaries and Generative AI in the editor, two capabilities that will help accelerate agents
AI issue summaries is only available in JSM…
Generative AI in the editor, isn’t just limited to KB. AI editing is across the majority of Atlassian Cloud products from the Jira family, Confluence, Atlas and Bitbucket.
The last feature we’re going to show today is brand new and currently being rolled out in limited beta: AI-powered request type suggestions
JSM offers a number of templates and features to make setting up a new service project easy, but new and non-technical users can often get stuck brainstorming which request types to use.
In the Request Types section of Project Settings…
There’s a new option…
<CLICK>
“Suggest”, let’s check out how it works.
It reveals a text area that allows Professor Wong to describe the help his team provides
<CLICK>
He enters a basic description and hits “go”
Comes back with a list of request type suggestions. Orientation information looks perfect.
Right now, we’re only suggesting the request type names
In the future, we’re looking to extend this to field suggestions as well.
This is a new feature that’s being rolled out in limited beta right now. You can keep an eye out for availability updates in the Atlassian community.
So that’s AI project configuration, we’re excited to see how that can help teams accelerate getting started.
Beyond this list here, Atlassian has more exciting AI features in the works, so please keep an eye out for updates on new features.
We encourage you to check out the resources linked here: the Virtual agent product guide which has tips and tricks for getting started, and also a link to an overview of Atlassian Intelligence that includes FAQs and links to additional resources.
If you have any questions, we’ve got a few minutes and we’d love to try to answer them for you so feel free to ask.