ANAHEIM, Calif. -- Lines are blurring between business and technical processes at many organizations. Add agentic AI, and the field of enterprise knowledge management -- specifically, enterprise search -- suddenly has a much broader audience and set of tools from which to choose.
Agentic AI integrated with enterprise search tools uses large language models (LLMs) to summarize multimodal information in context and return responses to open-ended natural language queries such as, "What are all my high-priority Jira tickets?" or "What is the new vacation policy?" Agentic AI, through direct UI integration or a chat interface, can then answer further questions, suggest next steps or take action instructions from the user within the same interface, such as, "Email this to my manager."
These increasingly centralized workflows shared by business and tech teams -- what Atlassian has dubbed a "system of work" -- have easy, comprehensive, flexible and centralized access to data at their heart. To take advantage of them, most enterprises will need to catch up in knowledge management (KM) practices, skills and tools, said Julie Mohr, an analyst at Forrester Research.
"During research for a recent KM Wave report, what I heard from clients was, 'We found out very quickly that the quality of our knowledge was not where we thought it was' [when adopting generative AI]," Mohr said.
Increased emphasis on knowledge and data management also influences vendor product development. It increases competition between vendors, such as Atlassian and Google, which each updated agent-assisted enterprise search products this week.
Atlassian Rovo updates led the news out of the Atlassian Team '25 conference.
Google Agentspace vs. Atlassian Rovo
Google Agentspace launched in early access in December, and the vendor made some features available to a limited number of customers this week during its Google Cloud Next conference. It also added preview-stage integration with Chrome Enterprise and built-in agents for idea generation and deep research, capable of carrying out long-term, multistage research projects.
A Google-approved list of customers has access to built-in connectors to third-party tools such as Atlassian's Jira Cloud and Jira Data Center -- Google's blog launching Agentspace in December prominently features a Jira workflow. According to Google documentation, users must set up their own permissions for such integrations.
Atlassian made connections and automation for third-party tools a centerpiece of its product line in 2021 with Open DevOps, which contained the first iteration of its Teamwork Graph. Its Rovo Search, made generally available in October and updated this week, now has more than 50 built-in connectors with third-party applications, including built-in permissions management.
Over the last two years, Atlassian built a centralized data lake for its cloud platform, added an Atlassian Intelligence analytics tool and expanded Teamwork Graph well beyond DevOps tools. It is much newer to search than Google, but has been working to catch up quickly, according to Tiffany To, senior vice president and general manager of enterprise and platform at Atlassian.
"What we announced yesterday was really Teamwork Graph 2.0 because we recruited a lot of engineers from Meta who had worked on their knowledge graph, and they used a very similar architectural approach," To said. "One of our challenges was, we brought the data in, but it wasn't fast enough. We wanted it to be really fast. Otherwise, those use cases aren't very helpful if you've got to take a long time to query."
This revamped Rovo Search will soon become available to all cloud platform subscribers. During its Team '25 conference this week, Atlassian also previewed its own Deep Research agent.
One enterprise customer of both vendors who had early access to Google Agentspace and Atlassian Rovo Search said their strengths and weaknesses have been predictable so far.
We could search inside Jira tickets or Confluence before -- the biggest difference is that Rovo Search is a one-stop shop, almost like a Google search engine for the enterprise.
Kasia WakarecyVice president of enterprise data and apps, Pythian
"Google is really good at searches, so they had that advantage," said Kasia Wakarecy, vice president of enterprise data and apps at Pythian, a data and analytics services company that is an Atlassian customer and partners with Google. "When we did an exact comparison side by side with the same prompts and the same data, we found Google Agentspace was slightly better, not necessarily at summarizing, but thinking ahead to the next step."
However, Pythian chose to put Rovo Search, Chat and Agents into production. At the time, Rovo was generally available, while Agentspace was still in early access, which made Rovo the more proven product, Wakarecy said. But Rovo's strong integrations with third-party tools, along with the workflow tools that developers already used, were the major deciding factors.
"We could search inside Jira tickets or Confluence before -- the biggest difference is that Rovo Search is a one-stop shop, almost like a Google search engine for the enterprise," she said. "You go to this one window, type in what you want, and you don't care if it's a Slack message, Confluence document, Google Doc or email."
Enterprises weigh competing sources of truth
According to IT pros, it's also unlikely that any agentic AI tool will exclusively capture most enterprise IT environments. Pythian also uses Google Gemini and Code Assist, and employees will continue to use whichever generative AI or agentic AI feature is native to their preferred tools, Wakarecy said.
So far, early adopters of Atlassian Rovo tend to be among its existing following for corporate daily workflow and software development tools. For example, Microsoft and Atlassian have multiple integrations available that connect Atlassian Jira with Microsoft Copilot, but one user of both tools said he prefers Atlassian Rovo.
"With Rovo, we didn't have to do contextual switching -- jumping from one application to the next, dumping something into Copilot, trying to templatize it, bringing it back in," said Fred Frenzel, director of the project management office at HarperCollins Publishers in New York City. "It also has information about people in the Teamwork Graph, so it could look through all the users and start recognizing personas. I was writing notes for someone in sales in a sandbox environment, with no other data to look through except what I was feeding it, and after a while, it started saying, 'Oh, so and so, they'll be interested in this.'"
Google has less of an advantage in developer tools and IT service management (ITSM) than Atlassian. However, Google's expertise in databases, analytics and data management shouldn't be underestimated in agentic AI and enterprise search, said Chirag Dekate, an analyst at Gartner.
"Agents bring AI to life in enterprises, and this week at Next with Data Science Agent, Google basically provided all the core layers for any data engineering team to simplify the curation of AI and these data foundations," Dekate said.
Enterprises have many further options in this market beyond these two vendors, all of which position themselves as a single source of truth for all organizational systems. Microsoft Enterprise Search and Power Apps are longstanding players in business intelligence and organizational data management, and Microsoft has many large enterprise customers for its Office 365 productivity tool suite. For ITSM teams, ServiceNow, with its recent Moveworks acquisition, is also a major contender.
But Teamwork Graph makes Atlassian an early contender in knowledge graph-based retrieval-augmented generation, or GraphRAG, for enterprise AI agents, said Charles Betz, an analyst at Forrester Research.
"The graph makes all the difference to me," Betz said. "LLMs hallucinate, and you can't trust them. LLMs plus RAG are somewhat better, but they still can hallucinate. When you ground it in the graph, the metaphor I use is that LLMs are like flesh, just kind of floppy. But then the graph is the skeleton."
Beth Pariseau, a senior news writer for Informa TechTarget, is an award-winning veteran of IT journalism covering DevOps. Have a tip? Email her or reach out @PariseauTT.
