Behind all the collaboration that Atlassian offers other companies lies another kind of complexity: the financial operations that support Atlassian.
At CamundaCon 2025 in New York, Claude Khoury—head of finance enablement and AI at Atlassian—and Varad Seshadri—head of sales and finance technology—discussed their in-house AI-powered platform for real-time visibility into Atlassian’s financial operations.
Runner-up for the Gartner 2025 finance use case award, STAR (system transparency and reliability) provides centralized visibility into Atlassian’s financial data pipelines. It automates repetitive tasks and streamlines the month- and quarter-end close process. At the backend of this AI-powered solution is Camunda, Atlassian’s orchestration engine.
The challenge: Complexity at scale
Atlassian’s mission is to “unleash the potential of every team,” according to Khoury. The company’s goal is to help teams collaborate better and help them build software and drive products in a more effective way. With flagship products like Jira and Confluence, Atlassian helps over 300,000 customers across 15 countries maximize everyday collaboration.
This of course means that Atlassian deals with a lot of complexity when it comes to billing. Atlassian’s financial technology system alone is a web of interconnected systems, some homegrown, some SaaS. “Camunda has been really the central orchestrator for that whole code-to-cache engine,” says Khoury.
According to Seshadri, Atlassian moved away from NetSuite about two years ago due to growth issues and switched to Oracle Fusion. The current goal is to make sure Atlassian’s financial tech has a good process automation layer. Camunda plays a key role in making sure billing transactions are orchestrated across all Atlassian’s billing systems and flow seamlessly into Oracle Fusion on a “near real-time basis.”
Building STAR: System transparency and reliability
Before STAR, Seshadri said, daily reconciliation was slow and error-prone. Manual activities across systems created inefficiencies and left the business operations team with limited visibility into where failures occurred. This “black box” effect meant that when a transaction failed, manual hours had to be expended to dig through multiple logs and systems to identify the issue.
To truly meet the needs of a company the size of Atlassian, its data source systems needed an almost real-time sync with the target system (Oracle Fusion).
STAR has three main tasks:
- Monitor all jobs that are running and flowing into Atlassian’s finance systems
- Facilitate Atlassian’s month- and quadrant-end close process
- Agentically respond to user interaction, with questions like “how ready am I to do the close process? What steps are pending?”
This agentic capability transforms STAR from a static dashboard into a conversational assistant for Atlassian’s finance professionals. Instead of waiting for reports, users can interact with STAR in plain language, a key leap in process orchestration powered by AI.
Orchestration at the core: How Camunda connects it all
Camunda comes into the picture by stitching together all of Atlassian’s homegrown systems and SaaS systems, driving the process orchestration, and then making sure the correct data feeds into STAR and Oracle Fusion. Seshadri points out that they also use Camunda as a business rules engine, transforming complex data from their legacy systems into Oracle. In this way, Oracle Fusion stays aligned and ready for core financials reporting.
With near real-time synchronization across its financial ecosystem, Atlassian could now monitor and resolve issues before they grew too impactful.
Measurable impact: From 16 hours to 30 minutes
Khoury and Seshadri closed their talk with a short video demonstrating STAR in a typical daily use situation. ”John” receives a Slack alert about a failed transaction from STAR’s AI agent. He clicks through Slack to STAR’s UI and checks out the details about the failure, including the error code, the workflow context, and which system had the error. He can request a plain language explanation from STAR and receive recommended next steps (he could also keep the conversation in Slack, if he preferred to never leave his collaboration environment).
The demo highlighted how this AI-driven guidance accelerates resolution and ensures issues are addressed before they impact month-end close.
The demo also ran through an example of job monitoring, which was previously manual and fragmented. Now, STAR aggregates all recent job runs in one place, enabling users to see operational statuses, failure counts, and error details. Role-based access controls ensure only authorized users can view or act on sensitive data, mirroring permissions in Oracle Fusion. Sensitive data stays protected while authorized users gain the insight they need.
The results of this visualization into what happens on the backend systems have been significant and immediate. “We used to spend 16 hours every month just doing manual tasks and a lot of different recons,” Khoury explained. “And now it just takes us like 30 minutes.”
That efficiency of course translates directly into business impact: faster close cycles, fewer errors, and better financial decision-making.
“We used to spend 16 hours every month just doing manual tasks and a lot of different recons. And now it just takes us like 30 minutes.” -Claude Khoury
The vision: accelerating the close process and enabling predictive insights
STAR went live in July, but Khoury explained that this is just version 1.0. “We still have a lot of ideas ahead.” Rather than being able to close the books for a public company of its size in three days, Atlassian would like to be able to reduce that to one day.
A feature they’re currently working on is something called Flux Analysis. Ideally, AI would be able to compare quarters of financial data so the team can simply review results, focus on strategy, and respond accordingly.
They hope to have Flux Analysis launched by December. This would be in line with Atlassian’s financial reporting goal: to have one point of view when it comes to financial reporting, a close possibility with Camunda as the central orchestrator. Due to the many systems in use throughout the global company, such a POV would shorten process times considerably, especially if users can chat with an AI-enabled system in natural language to get the results they’re looking for.
That’s more than automation. Camunda elevates the system to agentic orchestration, bringing intelligence into the daily workflow of financial operations.
Why build, not buy?
The STAR platform was a response to the challenge of monitoring data flows and orchestrating those business processes in a complex environment. It made sense for Atlassian to build STAR from scratch, Seshadri said, rather than look for an external product because nothing in existence would really fit the Atlassian ecosystem. Plus, they wanted to see what their own products could do.
With Camunda’s ability to integrate with all kinds of endpoints, Atlassian had the flexibility to connect all their various systems and continue using them with Camunda as a central orchestrator.
“At the end of the day,” Khoury said, “we’re a company that really wants to help other companies collaborate better and drive efficiency. So we wanted to drink our own champagne, use our own products, and see what we can do with them.”
This concept of experimenting, iterating, and improving is evident as STAR continues to unfold. Rather than treating agentic automation as a one-time project, Atlassian seems to be approaching it as a living system to be continuously refined with new efficiencies.
For organizations grappling with similar challenges of fragmented systems, growing data complexity, and slow process cycles, STAR is an intriguing roadmap. With Camunda’s ability to integrate with all the systems you already use, agentic orchestration is finally within reach.
Start the discussion at forum.camunda.io