News
4 min read
Atlassian, the Australian company behind Jira and Confluence, has rolled out AI agents within its existing tooling under the name Rovo Agents. The agents handle multi-step tasks, such as triaging support tickets, summarising meetings and generating code, without users having to leave their familiar work environment.
In the second quarter of fiscal year 2026, Rovo counted more than 5 million monthly active users. During the same period, Rovo Agents executed 2.4 million workflow automations for customers. More than 90 percent of Atlassian's enterprise Cloud customers now use Rovo; in one recent month, 14 million Rovo-assisted actions were recorded.
The development fits into a broader shift in which major technology companies are embedding AI agents into existing platforms. OpenAI recently reported that ChatGPT Work reached 10 million users in roughly one month. Meta released the open-weight model Muse Glimmer, aimed at agentic tasks that can run locally on a single GPU. Atlassian is taking a different route, building the functionality directly into its SaaS products.
The technical foundation of Rovo is the so-called Teamwork Graph, a context layer that gives AI agents access to more than 150 billion objects and relationships within the Atlassian ecosystem, including work history and personal data. The graph connects people, active work and knowledge documents, including data from linked external SaaS applications.
Atlassian states that customers using agents built on this graph receive up to 44 percent more accurate answers and consume 48 percent fewer tokens compared with agents that lack this context. Lower token consumption has a direct impact on cost per interaction, which can make a measurable difference for larger organisations.
Rovo Chat has evolved into a hierarchical multi-agent system. This means a single task can be distributed internally across multiple specialised sub-agents, enabling more complex scenarios to be handled without the user having to manage the division of work themselves.
Alongside the general Rovo Agents, Atlassian has released Rovo Dev, a specialised agent for software engineers. Rovo Dev supports code planning, code generation, code reviews and the elimination of repetitive work. For the underlying language models, Rovo Dev uses externally hosted LLMs from OpenAI and Anthropic.
Atlassian supports the business case with an internal figure: customers using AI code generation show 5 percent higher monthly active user counts and expand licences 5 percent faster than customers who do not. This is a correlation that Atlassian cites as a signal for product adoption, though it says nothing about causality.
Rovo's AI features are enabled by default for all Atlassian cloud services at the Standard, Premium and Enterprise tiers, including Jira, Confluence and Jira Service Management.
Rovo Agents do not operate exclusively within the Atlassian ecosystem. The agents are connected to external tools such as Slack, GitHub, Figma, Canva, HubSpot, Intercom, Amplitude, Box, Replit and Lovable. This allows workflows spanning multiple platforms to be managed from a single environment.
Atlassian has extended its multi-year partnership with Google Cloud. The collaboration includes integrations between Rovo and Gemini models, including Gemini 3 Flash, as well as connections with Google Workspace and Vertex AI. As part of this partnership, Atlassian received the Google Cloud Partner of the Year Award 2026 in the Application Development, Developer Experience category.
The primary commercial packaging for the AI functionality is the Teamwork Collection. This bundled subscription passed the threshold of 1 million seats and 1,000 customers in Q2 FY26.
Through Atlassian Ventures, the corporate venture fund established in 2020 by founders Mike Cannon-Brookes and Scott Farquhar, the company also invests outside its own walls. The fund started with $50 million and was later increased to $250 million. It now counts more than 50 portfolio companies, including Snyk, Miro, LaunchDarkly, Postman and AI writing tool Writer.
Atlassian recently participated in the Series D round of legal AI startup Legora, a $600 million round that values Legora at $5.6 billion, as reported in April 2026. Atlassian's own funding history is more modest: the company raised a total of $210 million across four rounds, including $150 million in a Series D in April 2014, before going public in 2015.
Atlassian's approach illustrates a pattern emerging at several large SaaS companies: rather than building a standalone AI product, agents are woven into existing workflows that users already engage with daily. The Teamwork Graph is the differentiating layer here, because it provides context that generic models without integration do not have.
For teams using Atlassian tools, this means the barrier to getting started with AI automation is low: the features are on by default. For developers and policymakers tracking the rise of AI agents, the combination of usage figures from Rovo, ChatGPT Work and Muse Glimmer shows that demand for multi-step automation is growing broadly and rapidly, across both consumer services and enterprise software.
This article draws in part on the following sources.
FutureHouse's Robin system identified new therapy candidates for an eye disease in 2.5 months, while Google DeepMind's AI co-scientist validated drug candidates that had previously been overlooked. Meanwhile, San Jose-based company Zinova demonstrated robots independently operating standard construction tools.
Utrecht-based Brainial is bringing a new product to market that aggregates tender announcements from multiple sources into a single dashboard. At the same time, technology company Croonwolter&dros has extended its contract for Brainial's AI software by three years.
Minister Heleen Herbert of Economic Affairs announced the establishment of the National Agency for Disruptive Innovation (NADI) on Budget Day. The agency will receive €260 million from the government over the next five years and will launch three pilot programmes in 2027.
0:11
21:28
21:31
14:28
1:06
0:24
0:56
1:02
