Companies are increasingly adopting AI Agents.
However, there’s a frustrating problem encountered during work:
AI is smart, but Context keeps getting lost.
A product manager might spend two hours discussing with AI,
consolidating:
customer problems,
feature priorities,
constraints,
launch plans.
Then they pass it on to Marketing.
The first thing that Marketing does is:
explain everything all over again.
The same happens when it moves to Customer Support,
requiring another re-explanation.
Even though AI saves you time, teams still waste time on:
transferring Context.
Atlassian’s latest update to Rovo Chat addresses exactly this issue with four new features.
What is Rovo Chat?
Rovo is Atlassian’s AI work platform.
It understands Atlassian tools like Jira and Confluence, and uses a Teamwork Graph to connect the relationships between:
- Work Items
- People
- Projects
- Knowledge
- Decisions
So Rovo Chat is not simply about:
“Asking AI a question.”
It operates on the existing corporate work context.
The update released on September 17 takes this further:
AI Context no longer belongs solely to one person’s chat.
First New Feature: Visible Memory
Many AI tools have memory,
but the real challenge is:
What exactly does it remember?
The new Rovo Chat adds a visible Memory View.
For example, it might remember:
You prefer bullet-point summaries.
A certain project launched in Q1.
Customer data cannot be included in drafts.
Previously this info was only influencing AI behind the scenes.
Now you can:
See it directly.
Plus, you can:
- View
- Edit
- Adjust preferences
- Disable memory
Memory can be controlled separately at the User and Site level by users and administrators.
This distinction matters because:
More memory stored by AI isn’t always better.
The key is:
You know what it remembers and can modify it.
Second New Feature: Direct @Agent Handovers
Imagine discussing a new product launch in Rovo Chat.
Market problems, customer needs, product constraints, launch dates, and competitor analysis have been covered.
Suddenly, you realize:
You need an agent specialized in Competitive Intelligence.
Previously, you would:
Create a new Agent
→ Copy prior conversation details
→ Explain everything again
→ Hope no important context is missed.
With the new Rovo Chat, you can directly:
@mention the Agent.
Examples:
@Competitive Intel Agent
@Launch Planning Agent
@Data Analyst Agent
The new Agent joining receives all the thread’s:
Context, decisions, and artifacts.
No need to start from scratch.
This capability is actually more important than just having multiple Agents.
Because what organizations really need is:
Different AI agents to seamlessly take over the same work.
Third New Feature: Share Entire Chat with Colleagues
This might be the most practical update.
Suppose you spend an hour with Rovo analyzing:
Why the product launch got delayed.
AI summarizes:
- What happened
- Data reviewed
- Reasons ruled out
- Remaining assumptions
- Final decisions
Now the Marketing Manager needs to take over.
Usually, they write a summary.
But summaries often contain only:
The final conclusion.
Why some options were discarded,
Which data was considered,
Which issues remain unanswered,
tends to get lost.
The new Rovo Chat lets you:
Share the full Chat.
The next colleague doesn’t just see:
“We decided to do A.”
They get the entire context to continue working.
Sharing a Chat Doesn’t Grant New Permissions
This is very important.
If the chat references:
Engineering Jira tickets,
Restricted Confluence pages,
Specific customer data,
Sharing the chat with another colleague does not grant them access to information they don’t already have permission for.
Atlassian states explicitly:
Shared chats remain permissions-aware.
This means:
Colleagues only see what they are authorized to see.
Just because an authorized person chatted with AI,
sharing the chat won’t circumvent access controls.
This is a necessary boundary for enterprise AI collaboration.
Fourth New Feature: Turn Repeated Prompts into Custom Skills
This feature closely aligns with real workflows.
Say a Product Manager creates a Sprint Summary every Friday.
Each time, they tell AI:
Read this week’s Jira,
Identify completed items,
Classify blockers,
Organize next week’s priorities,
Then output in company format.
Repeatedly entering this prompt means you’re not really “using AI.”
You’re:
Having to retrain AI on the same SOP over and over.
The new Rovo lets you create:
Custom Skills.
You don’t even need to design complex workflows first.
You can say in Chat:
“Create a Skill that generates Sprint Summaries weekly from the Board.”
Rovo will ask:
Which items to include?
How to define blockers?
Which exceptions to exclude?
How to format output?
It packages these rules into reusable Skills.
Custom Skills Are More Than Saved Prompts
This is different from simply saving a prompt as text.
What you really want to preserve is:
How teams do their work.
For example, customer support can create a:
Ticket Triage Skill.
Product teams may create:
Weekly Sprint Summary Skill.
Marketing can create:
Launch Brief Skill.
Leadership might set up a:
Leadership Weekly Update Skill.
When new members join,
they don’t have to ask:
“How exactly do you create this report?”
Methods once locked in a senior employee’s mind start to be:
Documented → Validated → Saved → Reused.
This is the true value of Custom Skills.
How Does a Project Flow Seamlessly?
Atlassian’s example is easy to grasp.
A project might start with:
Jira Engineering Sprint,
then move to:
Confluence Marketing Launch Plan,
and finally:
Jira Service Management Customer Support Knowledge.
These aren’t three separate tasks.
They’re:
The same project passing through three teams.
Critical context should flow with the project.
Such as:
Why was the feature delayed?
Which issues remain unfixed?
Which customers will be affected?
What can Marketing promise?
How should Support respond?
If each stage re-explains everything,
no matter how fast AI is,
much time is wasted on handoffs.
Rovo Chat’s real goal is:
Keep Context tied to the work, not to a single person’s chat window.
Who Is Rovo Chat Best For?
Rovo Chat is not designed as a personal lifestyle chatbot.
It’s better suited for teams already using:
- Jira
- Confluence
- Jira Service Management
- Atlassian Cloud
Examples include:
Software companies,
IT teams,
Product teams,
Customer support teams,
Enterprise project teams.
Especially if you encounter:
The same information being reformatted repeatedly across departments.
Then this update is worth close attention.
Is It a Separate Purchase?
Rovo is currently provided to eligible Atlassian Cloud customers.
Officially supported plans include:
Jira, Confluence, Service Collection, or Teamwork Collection:
- Standard
- Premium
- Enterprise
cloud plans.
However, AI features must be enabled within the organization for Rovo Chat and Agents to be available.
Some Rovo AI operations also consume Rovo Credits depending on interaction complexity.
So, it’s not simply:
“If you have Jira, you get unlimited free Agent use.”
Available usage depends on your organization’s plan and credits.
Custom Skills Are Being Rolled Out in Phases
Another timing note:
When announcing the update, Atlassian said:
Custom Skills will be gradually available across instances within a week.
If your Rovo Chat already has:
Memory,
Share Chat,
@Agent,
but you can’t find Custom Skills yet,
it doesn’t mean your account is faulty.
It may still be in rollout.
The Real Significance of This Update
Looking at the four features together:
Visible Memory
Solves:
What does AI really remember?
@Agent
Solves:
How does another specialized Agent take over?
Share Chat
Solves:
How does a colleague take over full Context?
Custom Skills
Solves:
How to avoid retraining AI on the same workflows every time?
They address the same challenge:
Can AI work be handed off?
This could mark a key turning point for enterprise AI agents.
Because real corporate work is not:
One person asks AI,
AI answers,
End of story.
Instead, it’s:
A completes part,
B takes over,
Experts join in,
Rules are preserved,
The process repeats.
When AI really fits into this chain,
it starts functioning like a work system,
not just a smarter chat interface.
Today, make progress with AI.
Learn one AI skill every day.
Save a little time daily.
Improve your abilities bit by bit.
SasaDaily, growing with you.
