Atlassian Teamwork Lab head Dr. Molly Sands said many companies are approaching AI "backward," improving how individual employees use the technology while leaving team-level collaboration unchanged. In her view, that helps explain why faster workers have not consistently produced stronger organizations or clearer returns on investment.
According to a VentureBeat report sponsored by Atlassian, Sands made the remarks during a public session at VB Transform 2026. The fireside chat was hosted by VentureBeat senior tech contributor Sam Witteveen. Sands said the core issue is not that people are failing to move faster. It is that team processes have not been rebuilt around AI, so individuals can end up moving quickly in slightly different directions and "crashing into each other."
Faster individual output has not translated into company-wide gains
Generative AI tools have clearly accelerated everyday work for many employees, but that improvement has not smoothly turned into broader corporate advantage. Sands, who leads a team of behavioral scientists and psychologists, studies how AI changes collaboration and also works with organizations on workflow redesign.
Her argument was direct: most companies are teaching employees to use AI more efficiently before they rethink how teams actually coordinate. That imbalance separates personal productivity gains from organizational performance, making measurable ROI hard to find.
Survey data shows a wide gap between speed and value
Atlassian's 2026 State of Teams Report surveyed 12,000 knowledge workers globally as well as about 200 Fortune 1000 senior executives. The findings showed that 89% of executives acknowledged employees were moving faster after adopting AI on an individual level.
Yet only 6% said they could identify a clear ROI example. Just about 14% of teams successfully converted AI use into real value. The figures suggest that even inside the same company, team performance can vary sharply.
Three traits shared by teams that get results
Sands said organizations that turn AI into a team advantage usually share three characteristics.
- Context: Successful teams build a "context graph" that records goals, decisions, and organizational knowledge in shared digital systems such as Jira or Confluence. That gives AI access to organizational context instead of relying only on isolated personal prompts.
- Workflows: They do not stop at making one task faster. They redesign the full end-to-end workflow.
- Culture: Leaders explicitly encourage learning and experimentation and can tolerate failure during that process.
AI working agreements can expose hidden organizational fractures
To help companies close the gap, Sands suggested that teams learn quickly by imposing deliberate constraints. Her examples included breaking tasks down to the smallest units or trying a challenge such as going one week without writing code by hand.
She also said it is important to set clear "AI Working Agreements" at the start of a project. Those agreements cover which tasks should be handled by AI, which ones should be avoided, and which agents and skills should be shared across the team.
AI did not create new management problems, she said
The report ended on a broader point: AI has not created entirely new management challenges. Instead, it has exposed fractures that already existed inside organizations, including hidden assumptions and different mental models. Once AI is introduced, the consequences of those gaps become larger, making shared context and clear ways of working more urgent than before.