Atlassian 与 Google Cloud 扩大 AI 合作…将“Rovo”集成至 Gemini 及 Workspace

Atlassian has significantly expanded its long-term partnership with Google Cloud. The key is to deeply integrate its AI intelligent agent platform “Rovo” into Google Workspace and Gemini Enterprise, and to migrate major AI training tasks to run on Google’s infrastructure.

The announcement was made at the “Google Cloud Next 2026” conference in Las Vegas, USA. Atlassian said it is building a complete stack for training and inference based on Google Kubernetes Engine and Google Cloud’s “AI supercomputer.” The orchestration layer is designed by Atlassian itself, with the goal of being able to flexibly scale workloads across high-performance GPUs and Google’s custom TPUs.

According to the company, some of Rovo’s training tasks have already been running on this infrastructure. Preliminary results show that it can retrieve answers faster and with higher relevance from enterprise internal knowledge bases. For enterprise-grade AI, speed and accuracy are directly tied to real productivity; therefore, this collaboration has been interpreted as not just a simple technical alliance, but an important move aimed at strengthening the competitiveness of overall work tools.

Applying Gemini 3 Flash… Enhancing document summarization and multimodal capabilities

Google’s Gemini 3 Flash model will be integrated into some of Rovo’s functions. Atlassian says it plans to use the model in scenarios such as complex reasoning, multimodal tasks, and large-scale document summarization. They also say they will remain flexible and can switch to other models depending on specific use cases.

Practical use cases have also already emerged. The newly launched Confluence “Remix” feature can convert text-based documents into charts and diagrams, a process that leverages Gemini 3 Flash’s multimodal capabilities. Turning the vast amount of documentation accumulated by enterprises into visual formats is focused on improving collaborative productivity.

Integrating Rovo into Gemini Enterprise and Workspace

The two sides have also strengthened cross-product integrations. Now users can access Rovo directly within Gemini Enterprise. This means there’s no need for separate customized integration work to bring Atlassian’s work context into Gemini-based intelligent agents.

Conversely, in Google Workspace, users can now also make Atlassian-related queries. Through the Atlassian Rovo Model Context Protocol (MCP) server, users can call Jira data in Google Docs or Gmail without switching tabs. This has been interpreted as an effort to reduce efficiency losses caused by switching between work tools.

Atlassian is also participating as an early access partner for Google Workspace MCP servers. With this move, future Google features will be more directly integrated into the Atlassian environment. The two sides believe that joint customers can automate workflows across the two platforms, thereby building an AI-supported pipeline from idea discovery to real operations.

Enterprise AI competition: “work connectivity” is more important than the “model” itself

Jamil Balyani, head of Atlassian AI products, said: “We are designing with Google Cloud the infrastructure and AI agents that our teams rely on to execute core tasks. We will combine Atlassian’s AI-based work system, Rovo, and Google Cloud’s AI stack to give customers more choices and more powerful agent-driven workflows.”

Satisth Thomas, vice president of Google Cloud’s AI applications and platform ecosystem, also said: “By expanding our partnership this time, we are directly embedding Gemini Enterprise and Google Workspace capabilities into the team collaboration stack.”

This release shows that the competitive focus in the enterprise AI market is shifting away from the performance of large language models alone, toward the “connectivity between work tools.” Atlassian and Google Cloud are working to deeply integrate AI into real collaboration workflows rather than treating it as a standalone feature. For enterprise customers, whether existing documents, emails, and project management tools can be seamlessly connected is increasingly becoming a more important decision criterion than the new model itself.

TP AI Tip: This article is summarized based on the language model from TokenPost.ai. The main content of the body text may be omitted or not match the facts.