Scaling agentic AI: Inside Atlassian’s culture of experimentation

Scaling Agentic AI is not just about having the latest tools – it requires clear guidance, the right context and a culture that experiments champion to unlock real value. With Venturebeats Transform 2025Anu Bharadwaj, president of Atlassian, shared useful insights into how the company enabled its employees to build thousands of adapted agents who solve real, everyday challenges. To build these agents, Atlassian has promoted a culture that is rooted in curiosity, enthusiasm and continuous experiments.

“You hear a lot about AI top-down mandates,” said Bharadwaj. “Top-down mandates are great to make a big splash, but really, what happens afterwards and with whom agents require constant itteration and adjustment. Top-down mandates can encourage people to use it in their daily work, but people have to use it in their context and repeat it over time to realize maximum value.”

This requires a culture of experiments-a culture where setbacks are not punished in short to medium-term, but are embraced as stepping stones for future growth and high-impact use cases.

Create a safe environment

The agent building platform of Atlassian, ROVO Studio, serves as a playground environment for teams in the company to build agents.

“As leaders it is important for us to create a psychologically safe environment,” said Bharadwaj. “At Atlassian we have always been very open. Open Company, no nonsense is one of our values. So we focus on creating that openness and creating an environment where employees can try out different things, and if it fails, it’s okay. It’s fine because you have learned something about how to use AI in your context.”

You also have to create a balance between experiments with guardrails of safety and auditability. This includes safety measures such as ensuring that employees are registered when they try tools, to ensure that agents respect permissions, understand -based access and offer answers and offer actions based on what a certain user has access.

Supporting cooperation between team agent

“When we think of agents, we think about how people and agents work together,” said Bharadwaj. “What does teamwork look like in a team that consists of a bunch of people and a number of agents -and how does that evolve over time? What can we do to support that? As a result, all our teams use ROVO agents and build their own ROVO agents. Our theory is that once that kind of teamwork is more common, the whole business changes.”

The magic really happens when several people collaborate with several agents, she added. Nowadays many agents are single-player, but interaction patterns evolve. Chat is not the standard interaction pattern, says Bharadwaj. Instead, there will be several interaction patterns that stimulate cooperation between multiplayer.

“Fundamental, what is the teamwork about?” She posed in front of the public. “It is collaboration with multiplayer – several agents and several people who work together.”

Make agent -experiments accessible

ROVO Studio from Atlassian makes Agent Building available and accessible to people of all skills, including no-code options. A customer of the construction sector built up a series of agents to shorten their route map creation time by 75%, while publishing giant HarperCollins built agents who reduced the manual work by 4x in their departments.

By combining ROVO Studio with their developer platform, Forge, technical teams get powerful control to deeply adjust their AI work flows – Define context, specify accessible sources of knowledge, form interaction patterns and create more – and highly specialized agents. At the same time, non-technical teams also have to adjust and repeat, so they have built experiences in ROVO Studio to enable users to use the natural language to make their adjustments.

“That will be the great unlocking, because if we are fundamentally about agent transformation, it cannot be limited to the Codegencenarios that we see today. It has to penetrate the whole team,” said Bharadwaj. “Developers spend 10% of their time coding. The remaining 90% cooperates with the rest of the team, sorting out problems with customers and solving problems in production. We create a platform so that you can build agents for each of those functions, so the whole loop becomes faster.”

Create a bridge from here to the future

In contrast to the previous shifts to mobile or cloud, where a series of technological or go-to-market changes took place, the AI-transformation is fundamentally a change in the way we work. Bharadwaj believes that the most important thing is to be open and share how you use AI to change your daily work. “As an example I share loom videos of new tools that I have tried out, things that I like, things that I didn’t like, things I thought, oh, this could be useful if it had the right context,” she added. “That constant mental iteration, for employees to see and try every day, is very important if we shift the way we work.”