You don’t need a team of AI experts. You just need one.
Atlassian’s Teamwork Lab found that adding a single AI-skilled team member significantly boosts project quality, with an 18-percentage-point increase in top scores, while overall team size primarily affects delivery, not quality.
Atlassian’s Teamwork Lab analyzed data from ShipIt 62, a company-wide hackathon held April 16–20, 2026, involving 2,574 participants across 1,271 teams. The study linked project judging scores to internal AI usage metrics to assess how team composition impacts outcomes. Teams varied in size and AI fluency, providing a real-world test environment for evaluating performance drivers.
The analysis revealed that team size strongly influences whether a project ships, with each additional member increasing the likelihood of submission by 79%. However, team size had no significant effect on the quality of the final project. Instead, the presence of AI superusers—defined as those in the top quartile of weekly AI interactions within their function—was the key factor in achieving higher scores.
Teams with no AI superusers were 18 percentage points less likely to receive a top score compared to those with at least one. The quality boost was most pronounced when a team transitioned from zero to one superuser; adding more superusers beyond one did not yield a similarly sharp improvement. This suggests that deep AI fluency in even a single team member can act as a force multiplier for project outcomes.
For organizations investing in AI tools, the findings imply that universal adoption may not be necessary. Instead, the priority should be ensuring every team has at least one member with advanced AI skills. The research highlights a practical approach to maximizing AI’s impact without requiring widespread expertise across all team members.