At Atlassian Team Barcelona, we sit down with CTO Rajeev Rajan. He outlines the pragmatic approach to artificial intelligence that sets the company apart from many other companies. Most important of all: Atlassian doesn’t train AI models on customer data. In an industry that seems to rely more and more on user information for competitive advantage, this commitment isn’t without its challenges. Dive into the world of Atlassian with us in this new episode of Techzine TV.
Subscribe to Techzine Talks on Tour and watch and/or listen to our other episodes via Spotify, Apple , YouTube or another service of your choice.
In a conversation with us, Rajan explains how Atlassian positions itself closest to end-user scenarios without crossing ethical boundaries. “Customer data is the customer’s data and we are custodians of it,” he emphasizes. This principle shapes everything from how Rovo delivers search results to how the platform’s AI agents operate across enterprise environments.
The Teamwork Graph: connecting work without compromising privacy
Central to Atlassian’s AI strategy is the Teamwork Graph, a knowledge graph with extra capabilities. It maps who works with whom on which documents at which point in time. Unlike traditional knowledge graphs, this technology maintains real-time and historical connections across an organization’s entire work ecosystem, according to Rajan..
The graph integrates data from over 80 applications, including Google Docs, Office, Slack, and Zoom, while respecting existing permission structures. When users search through Rovo, they only see results they’re authorized to access. “If there’s a sensitive financial report that only the CFO can see, it doesn’t show this data in the search results to other people,” Rajan notes.
This permission-aware approach operates in real-time, using customer data for the customer’s benefit without training proprietary models on it. The technical implementation allows Rovo to surface relevant suggestions without ever violating data boundaries, according to Rajan.
Rovo’s three pillars: search, chat, and agents
Atlassian’s AI platform Rovo delivers capabilities through three distinct mechanisms. Search spans entire company data repositories with permission awareness. Chat provides contextual conversational AI within Confluence, Jira, or as a standalone interface. Agents, both pre-built and custom, perform actual work autonomously.
During his keynote at Atlassian Team, Rajan demonstrated how Rovo can take a new idea, create a project in Jira Product Discovery, establish tasks in Jira, and even write initial code. This automation builds on Atlassian’s existing workflow capabilities and adds generative AI.
The platform uses multiple large language models strategically. OpenAI’s GPT powers many features, including GPT-5 for cutting-edge capabilities. Claude handles coding tasks. Open source models fill specific niches. This multi-model architecture through an AI gateway allows Atlassian to optimize for different use cases without vendor lock-in.
The 4% problem: why AI adoption remains challenging
Rajan candidly addresses research showing only 4% of companies currently see company-wide AI benefits. This statistic reflects the gap between AI’s promise and practical implementation challenges facing enterprises today.
The adoption barrier isn’t purely technical. Organizations struggle with determining appropriate use cases, measuring ROI, and bridging the gap between executive mandates and practical implementation. “Every time there’s a new technology, it takes some time for the world to adapt and learn and understand,” Rajan observes. Still, if the statistic is correct, 4% is a very low success rate.
At Atlassian itself, the company has set internal goals to use AI across every department. Engineering, finance, HR, legal, and sales teams experiment with Rovo to discover practical applications. This internal experience directly informs product development, creating a feedback loop between Atlassian as user and Atlassian as builder.
Top-down vision meets bottom-up experimentation
When we ask him whether companies should pursue AI from executive levels or grassroots initiatives, Rajan advocates for both. C-level commitment provides permission and resources for experimentation. Bottom-up discoveries identify what actually works in practice.
The most successful approaches combine executive goal-setting with freedom for teams to explore and iterate. “If you don’t have the top level goals, it doesn’t give permission or it doesn’t empower people in the organization to go try things,” he explaines. Conversely, executives can’t simply mandate AI transformation without practitioners finding practical applications.
This balanced approach acknowledges that AI won’t solve every problem automatically. Organizations must still define goals, create workflows, and determine success metrics. AI should enhance these efforts, not replace strategic thinking.
Data sovereignty and enterprise requirements
For European enterprises concerned about data sovereignty, Atlassian offers data residency across 11 regions. The company has achieved FedRAMP Moderate certification in the United States and provides country-specific data hosting where required. Rajan notes that some customers request bring-your-own-model capabilities, particularly in regulated industries. While Atlassian is exploring these options, the vast majority of customers achieve strong results with frontier models and general-purpose LLMs that evolve rapidly.
According to Rajan, this pragmatic stance towards bringing own models versus using general-purpose LLMs reflects Atlassian’s broader philosophy: focus on delivering value at the application layer rather than competing in infrastructure. The company isn’t building chips, creating proprietary LLMs, or training on customer data. Instead, it wants to connect best-in-class AI capabilities to real work scenarios while maintaining ethical boundaries.
From his previous roles at Meta and Microsoft, Rajan brought experience building world-class engineering teams. At Atlassian, he’s focused on broadening the company’s reach from developer tools to solutions for technical and non-technical teams across all company sizes. Based on what we have discussed in this episode of Techzine TV, Atlassian uses AI as an accelerant for this rather than the entire strategy.
Podcast player:
Techzine Talks on Tour becomes Techzine TV Podcast
Starting 1 January 2026, Techzine Talks on Tour will be rebranded as the Techzine TV Podcast. As we add video to every podcast we record, we think it makes more sense to cluster everything around our newly-launched Techzine.tv platform. If you don’t like watching conversations but rather listen to them without video, that will remain possible. The Techzine TV Podcast will be available on Spotify, Apple and many other podcast platforms. Just search for Techzine Talks on Tour in your favorite podcast app.
Our commitment to our listeners and viewers hasn’t changed. We aim to publish interesting conversations with leaders in the IT space. Tell us how we do by leaving a comment or rate us. We are always looking to improve, but obviously also like to receive a compliment every now and again.
Previous episodes of Techzine Talks on Tour:
- Pax8 wants MSP’s to become MIP’s: what does that mean?
- Workday CTO outlines bold AI agent strategy and major acquisitions
- Navigating VMware’s transformation under Broadcom
- Connected from curb to gate at Harry Reid International Airport
- The browser is the next security frontier: how do we secure it?
- Slack is evolving into a work operating system
Get in touch
We hope you like this podcast series. If so, please let us know. If you have suggestions on how we can improve, we would like to hear those too. We’re also open to suggestions around specific topics, or specific people that want to appear in an episode of Techzine Talks on Tour. You can find both Coen van Eenbergen and Sander Almekinders on LinkedIn, or you can send an email to info@techzine.eu.