Atlassian Jira now a hub for knowledge management, AI tools
Atlassian's raft of product updates at Team '24 spanned long-awaited product integrations and previews of AI automation updates, all of which advanced broader ambitions.
A bevy of feature updates for Atlassian's cloud platform this week marked the culmination of years of integration work between its products and data sets, setting the stage for knowledge management workflows that go far beyond its roots in IT.
A core group of four products -- Jira; Confluence; asynchronous video communications acquisition Loom; and Rovo, a new AI-assisted search and automation tool -- now form the "teamwork foundation" in the Atlassian cloud, according to execs during keynote presentations at this week's Team '24 conference.
While Confluence, Loom and Rovo were positioned as equal pillars within the Atlassian cloud product line, Jira hogged the spotlight on the keynote stage Tuesday. It subsumed the formerly separate Jira Work Management product to become a unified but customizable work management portal that users can tailor for individual teams from marketing and finance to software development and incident management.
"We simply call it Jira, and it brings together the best parts of Jira Work Management and Jira Software into one project management [tool] to help teams pursue shared goals where they plan and track their work," Atlassian president Anu Bharadwaj said during a keynote presentation. "Goals are now directly surfaced in Jira so every team has visibility on what they're working on -- from developers coding up the website to marketers planning their campaigns. … For any given work item, teams can see the goal that work is contributing to."
Jira appeared in nearly every keynote feature demo, most often in the form of the automatically opening and populating Jira tickets from other newly-integrated parts of the product portfolio. In other cases, the Jira product line added long-awaited integrations between existing Atlassian products, such as full Opsgenie integration into Jira Service Management six years after Atlassian's acquisition. Other long-discussed feature additions announced at Team included a Jira Align flow metrics database for value stream management within the Atlassian Analytics service.
"[Atlassian] is putting knowledge at the forefront. It's not 'we'll help you build a bot.' It's 'we're going to help you build better workflows. We're going to help your knowledge workers get to information more quickly,'" Mohr said. "From a strategy perspective, it's a very unique position in that market."
In some areas of ESM, Atlassian will never unseat ServiceNow from its dominant position, but it can carve out its own niche with a lighter-weight, more accessible set of products, Mohr said.
Atlassian co-founder and co-CEO Mike Cannon-Brookes introduces Atlassian Rovo during the Atlassian Team '24 conference keynote.
Unified cloud search a welcome development for users
All of Atlassian's cloud products are now underpinned by a unified data lake and a knowledge graph, which Atlassian calls the teamwork graph, that automatically maps relationships between customer data objects, teams of workers and workflow actions. That data layer and teamwork graph then feed into analytics dashboards and automated workflows that will soon be performed by AI agents.
But first: an improvement in the quality of search results within Atlassian's existing portfolio, based on AI-assisted search with retrieval-augmented generation, an area where Atlassian officials acknowledged past versions of products left a lot to be desired.
"Over the last six months, we've made it 25% more likely that you'll find what you're looking for in the first position of search," said Jamil Valliani, vice president and head of product for AI at Atlassian -- a position he left Microsoft's search and AI team for in October -- in an interview with TechTarget Editorial. "We've actually gone from being behind industry leaders in search in terms of [normalized discounted cumulative gain], which is a standard way of measuring search relevance in the industry, to over eight points ahead of … the industry leader."
These investments in search are a good sign, said Andy Rosequist, vice president of engineering at Sector Alarm Group, based in Oslo.
"Atlassian's search has long been a top complaint," he said. "It's so hard to tell if the information doesn't exist in Jira or Confluence, or if the search is just bad."
Unified search across different cloud products, along with new integrations, means time savings for cloud customers, according to one Atlassian Community Leader.
"The fact that we can now search across products and remove that context switching [between tools] is … something that Atlassian has always struggled with in some of their search [interfaces], especially from a Confluence perspective," said Dan Tombs, Atlassian architect at a satellite communications company. "It's nice to see that this is getting some real attention. And with AI, the fact that you can just ask natural language questions is really, really cool."
Atlassian co-founder and co-CEO Mike Cannon-Brookes introduces Atlassian Rovo during the Atlassian Team '24 conference keynote.
Atlassian Intelligence leads AI automation rollout
Throughout Atlassian's cloud, generative AI put a new sheen on existing products, such as automated service desk generation within Jira Service Management for finance, legal and HR departments -- a significant expansion from past releases' preconfigured request templates. Atlassian officials previewed upcoming AIOps features for automated root cause analysis and incident resolution in Jira Service Management.
Other features of Atlassian Intelligence, first previewed in April 2023, that were showcased at Team '24 included an AI-driven document editor for Confluence, Jira and Jira Service Management that ties in with Bitbucket to create pull-request summaries and release notes. Atlassian Intelligence can break large Jira projects down automatically into smaller components, and automatically generated issue summaries will become available soon. AI-generated API documentation will follow based on Atlassian's acquisition of Optic this week.
Tombs said his team already has Atlassian Intelligence turned on for all its cloud sites and has used it to write natural language queries that the tool translates into Atlassian's Jira Query Language (JQL).
"It alleviates some of my pressure because I don't have to necessarily be on hand to support them if they're not very familiar with [JQL]," he said. "It's already been used for Confluence, not only from summarizing large docs but helping to produce those docs."
AI-generated references to issue context could be useful for incident responses, Rosequist said, but he expressed doubts about AI-generated Jira issues.
"A big weakness of GenAI is it sounds like something that a person would write, and there's already lots of confusing tickets in the world without AI writing them too," he said.
Tombs said he disagreed.
"Most teams are slammed; everything's really busy," he said. "That whole idea of just giving AI tasks is the future? It's inevitable. This is just the next evolution of [automation]."
That isn't to say AI automation is without its risks and drawbacks, Tombs said. For instance, the quality of the data AI draws on must be high to produce useful results, which means organizations often must clean up existing documentation before applying AI.
It's important for teams to truly understand their objectives before creating automation workflows, he said, which can involve low-tech approaches.
"I always recommend you start with 'what am I trying to do and how?' and then 'what am I trying to use?'" he said. "I know I'm not the only Community Leader that does this. I always recommend that when you are starting with automation, just start listing requirements on a piece of paper."
Knowledge management takes shape with Rovo
Atlassian's forthcoming Rovo product will apply its unified cloud search and Atlassian Intelligence automation features to customers' third-party SaaS tools through custom connectors. Rovo will create knowledge cards and organization-specific term definitions by pulling together disparate data sources on the fly, according to Atlassian co-founder and co-CEO Mike Cannon-Brookes during a keynote presentation.
"Rovo uses AI and the teamwork graph to understand any document, connect it, cross reference it deduplicate it and analyze its concepts, without you needing to do anything," Cannon-Brookes said. "And as more documents are created, Rovo is always working. New statement of work written in a Word document stored in SharePoint? Rovo links the work and rewrites the definition. You list the target customers for [a project] in a Google Sheet that comes in from your sales team? Rovo links the work [and] updates the members of the team and the definition of the project."
A Rovo Chat AI assistant will answer questions users ask based on the teamwork graph even if the answer isn't written down already, according to Cannon-Brookes.
Atlassian is also anticipating common enterprise AI data privacy concerns by building Rovo with a private instance of OpenAI's GPT API that doesn't retain user data and refrains from LLM training that would mix data from multiple cloud tenants, according to Valliani.
Context-aware search in Rovo demos looks good, said Andy Thurai, an analyst at Constellation Research. But for now, Rovo remains in private beta, and a work in progress.
"The chatbots allow for conversation [that draws from] any of the documents within Atlassian's realm, though it is currently limited only to structured data, with no specific plan or timeline for unstructured data," Thurai said.
[Atlassian] is putting knowledge at the forefront. It's not 'we'll help you build a bot.' It's 'we're going to help you build better workflows.'
Julie MohrAnalyst, Forrester Research
Structured data builds in an organizational schema; unstructured data does not. It will be important to support both to integrate with a broad array of third-party tools, Thurai said. An Atlassian spokesperson said this is on the roadmap for Rovo.
Finally, Rovo will support AI agents, a form of bot-based workflow automation that some industry experts predict will be the next big trend in generative AI tools. Rovo will ship with more than 50 agents pre-built by Atlassian and its partners as well as allow users to custom-build their own agents using Atlassian Forge or a built-in no-code interface.
Rovo agents previewed at Team '24 keynotes included a Communications Crafter for marketing teams to ensure blogs and other external communications follow company policy as well as a social media scribe that links to Canva for image generation.
All of this might seem far removed from Atlassian's DevOps roots. But there are tiebacks to Jira in pre-built agents such as Backlog buddy, which organizes, consolidates and prioritizes Jira ticket backlogs, and Feature Flag Cleanup, which Atlassian's internal developer teams used to clean up more than 460 feature flags during the last six months, according to Cannon-Brookes.
Knowledge management is also an important aspect of DevOps, Tombs added.
"You're trying to break down that original silo [between dev and ops] and make things faster and easier from project initialization all the way through to actual management and operations," he said. "A huge part of the development of any product or any kind of service is the knowledge transfer from one team to another."
Beth Pariseau, senior news writer for TechTarget Editorial, is an award-winning veteran of IT journalism covering DevOps. Have a tip? Email her or reach out @PariseauTT.
