Atlassian Research Identifies Single AI Superuser as Critical Driver for High-Performance Team Outcomes

The Atlassian Teamwork Lab has released the findings of a comprehensive internal study conducted during the company’s biannual ShipIt hackathon, revealing a significant correlation between the presence of "AI superusers" and the quality of team outputs. The data suggests that universal AI fluency may be less critical to organizational success than the strategic distribution of high-frequency AI users. Specifically, the study found that adding a single AI superuser to a team that previously had none resulted in an 18-percentage-point increase in the likelihood of the team achieving a top-tier performance score.

As corporations worldwide grapple with the complexities of AI integration, the fundamental question for many leadership teams has been the scale of training required to see a return on investment. The Atlassian study addresses whether a company requires a workforce of universal experts or if a smaller, concentrated group of power users can drive meaningful results. By utilizing its internal "ShipIt" event as a controlled testing ground, the Teamwork Lab analyzed the performance of over 1,200 teams to determine how AI proficiency impacts innovation, impact, and project completion.

The ShipIt hackathon provides a unique environment for such research because it imposes a strict 24-hour time limit on all participants. During this period, self-organized teams must move from a concept—ranging from a product prototype to a specific technical fix—to a completed demo. Because every team operates under identical constraints and is evaluated by the same criteria of innovation, customer impact, and completeness, the event functions as a standardized laboratory for measuring team dynamics and technical execution.

The most recent iteration of the study, conducted between April 16 and April 20, 2026, involved 2,574 unique participants across 1,271 teams. To isolate the impact of AI skills, the Teamwork Lab cross-referenced project scores with internal AI usage data. The research focused specifically on 870 teams that successfully submitted demos for judging. The primary metric for success was defined as a "top score," which refers to a rating above the median across all judged categories.

Atlassian defines an "AI superuser" not by a fixed number of interactions, but through a relative performance metric. A superuser is categorized as an employee in the top quartile (the top 25%) of weekly AI interactions within their specific executive group or function. This methodology accounts for the natural variations in AI utility across different departments; for instance, a software engineer’s baseline for AI usage differs significantly from that of a human resources professional or a marketing specialist. By measuring usage relative to peers in similar roles, the study ensured that "superuser" status reflected a meaningful advancement in AI adoption within a specific professional context.

The data revealed that while 49% of ShipIt participants met the criteria for being an AI superuser, these individuals were not distributed evenly across the competing teams. Out of the 870 teams that received scores, nearly 25% operated without a single AI superuser. The performance gap between these teams and those with at least one superuser was stark. Teams with zero superusers were 18 percentage points less likely to receive a top score compared to teams that possessed at least one.

A critical nuance in the data is the "0 to 1" effect. The research found that the most significant performance gain occurred when the first AI superuser was added to a team. Interestingly, the addition of a second or third superuser did not produce a corresponding linear increase in the team’s score. Teams with multiple superusers performed roughly at the same level as teams with just one. This suggests that the presence of a single individual capable of deeply leveraging AI tools acts as a force multiplier for the entire group, providing the necessary expertise to elevate the project’s quality without requiring every member to possess the same level of technical fluency.

The study also identified a clear divergence between the factors that drive the quantity of work (shipping) and those that drive the quality of work (judged scores). The Teamwork Lab discovered that team size and AI fluency serve two distinct functions in a high-pressure project environment.

According to the analysis, team size was the primary predictor of whether a team would successfully "ship" a project within the 24-hour window. Each additional team member increased the likelihood of completing and submitting a demo by 79%. However, team size had no statistically significant impact on the quality of the demo or the final score received from judges. In contrast, the presence of an AI superuser did not significantly increase the likelihood of a team shipping their project on time, but it was the strongest predictor of whether that project would be ranked as high-quality.

This distinction provides a new framework for team design and project management. The findings suggest that if the objective is simply to ensure a project is completed, increasing the headcount is the most effective strategy. However, if the objective is to ensure the project is innovative, impactful, and complete, the presence of a high-frequency AI user is more important than the total number of people on the team.

The practical implications for organizational leaders are twofold. First, the data suggests that the current corporate emphasis on achieving universal AI fluency may be an inefficient use of resources. Instead of attempting to move every employee into the top quartile of AI usage, organizations may find greater success by ensuring that AI expertise is distributed strategically across all teams. If every project team has at least one AI superuser, the organization can maximize its performance outcomes without needing to wait for the entire workforce to reach expert status.

Second, the study highlights the role of the AI superuser as a "quality catalyst." Because the first superuser provides the majority of the performance boost, they likely serve as a consultant or a lead for AI-driven tasks within the team, allowing other members to focus on their core competencies while the project benefits from AI-enhanced workflows.

The methodology of the study was designed to isolate the effect of superuser composition by controlling for team size in all statistical models. The dataset encompassed a wide range of roles and project types, ensuring that the 18-percentage-point jump was a robust finding across various types of work. The teams were judged on three specific dimensions:

1. Innovation: The originality and creativity of the solution.
2. Customer Impact: The potential value the project provides to the end-user.
3. Completeness: The degree to which the prototype or tool was functional and ready for demonstration.

The boost provided by the first AI superuser was consistent across all three dimensions, reinforcing the idea that deep AI integration affects the fundamental nature of the work produced rather than just the speed of execution.

As AI tools become more integrated into the standard corporate tech stack, the Atlassian Teamwork Lab’s research suggests a shift in how talent is deployed. The findings indicate that the most effective way to leverage AI investment is not through broad, shallow adoption, but through the cultivation of a critical mass of power users who can then be embedded into diverse teams. This "force multiplier" effect allows organizations to achieve high-quality, innovative results by relying on a single point of expertise to elevate the collective output of the group.

In conclusion, the ShipIt 62 data provides a clear roadmap for AI-driven team composition. To move the needle on quality and innovation, the priority for leaders should be the elimination of "zero-superuser" teams. By ensuring that every team has at least one individual who uses AI deeply and frequently, companies can bridge the gap between simple project completion and high-impact innovation.