Atlassian spent nearly two years running its own enterprise AI transformation before it began selling the results. The company treated AI adoption as a company-wide effort, building and deploying agents across sales, HR, and finance, and rigorously tracking outcomes. What emerged is a playbook grounded in operational experience.
That playbook, along with a robust set of product announcements, was unveiled at Atlassian’s recent Team ’26 conference in Anaheim. The new and updated offerings put Atlassian’s learnings within reach of its enterprise customers.
The timing is right. Enterprise leaders are writing large checks for AI and struggling to explain what they are getting. The productivity story is compelling at the individual level, with analysts, engineers, and salespeople working faster, summarizing documents, and generating first drafts.
The picture at the executive level, however, looks very different. Atlassian’s internal research, conducted through its 13-person behavioral science Teamwork Lab, quantifies the disconnect. Individual productivity at Atlassian rose roughly 33% year-over-year from 2025 to 2026, while C-suite visibility into enterprise-level gains remained in the low single digits.
The irony is that technology is working, but the organization is not always capturing the value. This gap is the defining challenge of enterprise AI adoption in 2026.
Customer Zero: How Atlassian Built Its Own Playbook
Atlassian is different from most enterprise technology companies in that its tools come with an intrinsic opinion on how best to use them to solve business problems. Most of the world’s top software development organizations, for example, have workflows built around Atlassian’s Jira. These organizations trust that Atlassian’s point of view is credible and well established.
While every technology has a concept of “eating its own dog food,” where the company uses its own products. Atlassian goes further by defining the products it builds to serve its own needs, before sharing the outcomes with its customers. Atlassian is always customer zero. It’s ingrained in its culture.
I sat down with Avani Prabhakar, Atlassian’s chief people and AI enablement officer, to walk through the internal transformation program she has led since late 2024. Avani oversees IT, internal data science, and customer-facing technical support, a consolidation that reflects the company’s conclusion that AI transformation cannot be delegated to any single function.
“We knew that we’d require some forward-deployed engineering motion internally to make this happen,” she explained. “Traditional IT won’t work. Enterprises need a cultural transformation.”
She went on to explain that CEO Mike Cannon-Brookes wanted to produce a playbook that an enterprise can adopt, based on real operational experience rather than consulting frameworks. “He said we should not look or sound like the consulting firms of the world,” Prabhakar said. “It should be like, hey, this is our lived experience. This is how we did it.”
That mandate shaped three operating principles Atlassian developed for its own transformation, which Prabhakar now carries into customer conversations:
- Lead AI through an innovation narrative rather than an efficiency mandate.
- Acknowledge openly what is unknown rather than overpromising outcomes.
- Run fast experimentation loops rather than treating AI like a digital transformation program.
The workflow selection process was deliberately bottom-up. Atlassian ran internal AI innovation days to surface the highest-ROI use cases, producing roughly 14 workflows across functions, including sales, HR, finance, and legal. The common thread across each is that AI capability was not imposed from the top but rather grew from within.
Team ’26: Context as Infrastructure
Atlassian’s product portfolio is a direct extension of what Prabhakar’s teams learned internally.
The central thread in every announcement is organizational context, which Atlassian stores in its Teamwork Graph. Think of this as a connected map of people, projects, documents, code repositories, goals, and decisions. Atlassian describes the Teamwork Graph as the connective spine that enables agents to reason across Jira, Confluence, Jira Service Management, and connected third-party systems.
Atlassian revealed that its Teamwork Graph now contains more than 150 billion connections, updated by 12 billion daily changes across its customer base.
CEO Mike Cannon-Brookes laid out the premise during his keynote. “In 2026, anyone can buy smarts by the token,” he said. “The real moat is your institutional memory, every plan, document, and decision your teams have ever made.”
One of its most significant announcements at the event, Atlassian is opening the Teamwork Graph to external systems.
The Teamwork Graph command-line interface now provides developers and coding agents, such as Anthropic’s Claude Code and OpenAI Codex, direct, structured access to Atlassian’s work context from local tooling and CI pipelines, without requiring them to work through individual product APIs.
In parallel, Teamwork Graph tools now better support agentic workflows. Rovo, Atlassian’s core AI framework, includes an MCP server that exposes the same context to any MCP-compatible AI client. This includes tools such as Figma and Replit.
Other product announcements at Team ’26 filled in the operational picture:
- Rovo Studio reached general availability as a no-code environment for building agents, automations, and applications grounded in the Teamwork Graph context, with built-in roles, approvals, versioning, and audit controls.
- Agents in Jira are generally available, allowing teams to assign work items to AI agents using the same interface as for human assignees, with full audit logging designed for compliance teams.
- DX AI Experience, now generally available, gives engineering leaders visibility into where AI generates code, how agents perform, and the ROI of AI-assisted development, turning agent output into a governed part of the SDLC.
- Rovo Max mode, in early access, enables multi-step task execution across connected tools with an advanced planning engine that decomposes vague requests into structured workflows.
- Remix with Rovo, now in beta for Confluence, lets teams transform text, tables, and lists into charts, infographics, and slides without leaving the page, maintaining a single source of truth while enabling format flexibility.
- The Product Collection entered early access, combining Jira Product Discovery, a new Feedback app, and a planned Pendo integration into a connected product management workflow from customer signal to shipped work.
Competitive Landscape
The enterprise collaboration tool market in which Atlassian competes is substantial and increasingly competitive:
- Microsoft, with Microsoft 365 Copilot embedded across Teams, SharePoint, and the Office suite, holds the broadest footprint in enterprise knowledge work.
- ServiceNow has moved aggressively into AI-orchestrated workflows, particularly in IT service management and operations.
- Salesforce’s Agentforce extends its AI strategy into sales and customer experience workflows.
Each of these vendors has its own version of the organizational context story, backed by years of customer data and deep integration hooks.
Atlassian’s differentiation rests on the depth of its development and product teams’ footprint. Jira and Confluence are the dominant systems of record for software teams at scale, giving the Teamwork Graph a density of engineering context, including code reviews, sprint histories, incident records, and architectural decisions, that general-purpose collaboration platforms cannot replicate.
The decision to open that graph via the MCP and CLI interfaces, rather than requiring all AI interactions to run through Atlassian products, is a meaningful shift. Rather than positioning Rovo as the sole interface to Atlassian context, the company is treating the Teamwork Graph as infrastructure that other AI systems, including competitors’ tools, can plug into.
Atlassian is betting on ubiquity over exclusivity to reduce adoption friction for organizations whose AI stack already centers on Claude, Copilot, or another assistant.
The risk in that bet is data quality. Rovo’s MCP server and Teamwork Graph CLI expose the quality of an organization’s Jira and Confluence practices directly to AI agents. Organizations with stale documentation, inconsistent ticket ownership, and fragmented knowledge will surface those weaknesses as agent failures rather than as human workarounds.
Atlassian’s internal journey illustrates the investment required. Building the playbook required over a year of close coordination among the CISO, CIO, CTO, data science, and people teams before results materialized.
Analyst’s Take
Rovo is now deployed across more than 90% of Atlassian’s enterprise cloud customers and used by more than 75% of Fortune 500 companies. Agentic automations across the customer base have grown seven times in the last six months.
Those adoption figures are outpacing most enterprise software categories, and the opening of the Teamwork Graph to external agents extends Atlassian’s surface area well beyond its installed base. Any organization using Claude Code, Codex, Figma, or Replit is now a potential entry point for Atlassian context, even without a Rovo license.
The competitive pressure is only accelerating. Microsoft’s advantage in enterprise identity and calendar data, Google’s advantage in email and document context, and Salesforce’s advantage in the depth of customer records mean that no single vendor owns the full picture of organizational context.
Embedding MCP into the Teamwork Graph is a hedge against that fragmentation, a way to participate in multi-vendor agent architectures rather than compete against them. Cannon-Brookes captured this succinctly. “We don’t want to be a control tower,” he said. “I want to be a really important station on your subway network.”
What makes Atlassian’s position credible at this moment is its customer zero discipline. Enterprises evaluating AI collaboration platforms will increasingly demand this kind of evidence. Vendors who can demonstrate real internal adoption, measurable outcomes, and a transferable methodology will stand out from those offering AI as an aspirational overlay.
Atlassian has invested two years building that case from the inside out. Team ’26 is where Atlassian shared it with the world.
