🚀 Your Knowledge Base Is the Engine Behind Your AI (JSM Feature Roundup)

[#1]

Hi, Atlassian Community! 👋

If you've ever wondered why your AI features aren't quite hitting the mark or why draft replies miss the mark, the answer is almost always the same: it's not the AI. It's the knowledge behind it.

Your knowledge base is the engine that powers everything. Self-service suggestions, virtual agent conversations, Rovo draft replies, and more. A connected, well-maintained knowledge base is what turns Rovo in Jira Service Management from a nice-to-have into a genuine productivity multiplier for your team.

Over the last year, we've been on a mission to make that easier. Whether you're starting from zero, managing knowledge scattered across multiple tools, or trying to keep your knowledge base from going stale, we've built something for every stage of that journey. Here's what's new.

Not sure where to start? We've got you.

Setting up a knowledge base sounds straightforward until you're staring at a blank page with no idea what to write first. This is one of the most common reasons teams put it off entirely.

Create Knowledge Base with Rovo solves this with an AI-assisted creation flow. Bring your existing references, such as a PDF, a Loom recording, a URL, or even pasted notes, and Rovo will suggest starter article topics grounded in that context. Review the topics, confirm what you want, and Rovo generates full draft articles ready for your review. From there, publish, and your knowledge base is live. All in one session, without writing a single article from scratch.

It's the fastest path from zero to a working knowledge base we've ever shipped.

Learn more about building your knowledge with Rovo →

Your knowledge lives everywhere. Now your knowledge base can too.

Most teams don't have a knowledge problem; they have a knowledge scattered everywhere problem. Guides in Google Drive, runbooks in SharePoint, documentation split across multiple Confluence sites. Getting agents and helpseekers to the right answer means pulling all of that together.

Multi-source knowledge linking lets you connect multiple knowledge sources to a single service space simultaneously. Confluence spaces (from the same site or different sites), Google Drive, and SharePoint, all at once. No imports, no duplication, no manual cross-referencing. Your team gets a single place to find answers, and Rovo draws on all of them when generating replies and suggestions.

Learn more about connecting third-party knowledge sources →

Too much noise? Link only to what matters.

Connecting an entire Confluence space to your knowledge base is sometimes too much. Not everything in a space is relevant to your support team, and surfacing the wrong content can have consequences.

Granular linking gives you surgical control over what goes into your knowledge. Instead of linking a whole space, you can now link a specific folder, a page tree, or even a single standalone page — or any combination of those. Your knowledge base stays focused, relevant, and something your team can trust.

Learn more about granular linking →

Some knowledge is just for your team. Now it can stay that way.

Not everything belongs on the customer portal. Escalation paths, security procedures, compliance protocols, internal triage guides; this content is essential for your agents, but it should never be visible to helpseekers.

Internal knowledge lets you create a dedicated layer of knowledge that only your agents and Rovo can see. Rovo draws on these internal articles when drafting replies, so responses reflect how your team actually operates - not just what's on the public-facing portal. Customers continue to see the right articles for self-service while your agents get the deeper context they need to resolve requests confidently.

And with internal articles in the agent glance panel, the right runbook or procedure now surfaces directly in the work item view - no tab switching, no searching. The guidance agents need appears at exactly the moment they need it.

Learn more about internal and external knowledge →

Keep your knowledge fresh — automatically.

The hardest part of running a knowledge base isn't setting it up. It's keeping it up to date. Content goes stale, new issues emerge, and nobody has time to manually audit hundreds of articles.

Suggested topics take this a step further by analyzing thousands of past work items to surface what your customers keep asking about that you haven't documented yet. Think of it as your always-on content strategy telling you exactly where to focus next.
Learn more about Suggested Topics →

And the Rovo knowledge skill brings all of this directly into the authoring experience. Identify gaps, generate drafts from resolved work items, and flag outdated content without ever leaving the editor.
Learn more about Rovo knowledge skill ->

What it all adds up to

When the right sources are connected, the right content is surfaced to the right people, and AI continuously helps you build and maintain it, your knowledge base stops being a static library and becomes a living system.

Helpseekers find answers before they raise a work item. Agents resolve requests faster with the right runbook in front of them. Rovo drafts replies that actually reflect how your team works. And every resolved work item is an opportunity to make the next one faster.

Have you tried any of these features yet? Are you connecting knowledge from multiple sources, using Rovo drafts, or setting up internal articles for your agents? Drop your experience in the comments. We'd love to hear what's working and answer any questions you have. 💬

[Dave Rosenlund _Trundl_]

Love this, @Arzoo Jain. Leveraging Rovo and/or other AI systems to help humans create and curate knowledge is the right move.

I’d just add that so much of what teams rely on is tacit. AI still can’t see what we never codify, and that’s the very human, and harder problem to solve.

[Avinash Bhagawati]

Great article @Arzoo Jain ! 

[zoltanersek _outpostlabs_dev_]

great read, I like using resolved work items to identify knowledge gaps

[Brita Moorus]

I really like the direction this is heading. One thing I've noticed with customers is that AI usually isn't the limiting factor - the quality of the knowledge base is.

The more structured, up-to-date, and trustworthy the knowledge is, the better the suggestions, virtual agent responses and draft replies become. Features like granular linking and internal knowledge make it much easier to give AI the right context instead of all the context. Looking forward to seeing more teams adopt these.

[Wenjie Liang]

Great article!