Why Company-Wide AI Efficiency Is Harder Than Individual Speed

A new report from Atlassian reveals a major gap in how companies adopt artificial intelligence. While nearly 9 out of 10 executives say individual employees are working faster with AI, only 6% can point to clear returns on their investment. According to Dr. Molly Sands, head of Atlassian’s Teamwork Lab, most organizations are focusing on the wrong thing: they optimize how one person uses AI instead of how teams work together. Sands explains that AI tools make individuals faster, but without aligning teams, that speed can backfire. “Speeding up people who are pointed in slightly different directions means they quickly crash into each other,” she says. Only about 14% of teams have turned AI use into real value. These high-performing teams share three traits: they capture shared knowledge in a “context graph,” redesign entire workflows instead of just speeding up single tasks, and work under leaders who encourage experimentation. To close the gap, Sands suggests teams create “AI working agreements” at the start of projects. This means deciding together what to use AI for, what to avoid, and which tools to share. Imposing short-term constraints, like writing no code by hand for a week, can also speed up learning. The key lesson: AI doesn’t create new management problems—it exposes old ones, like hidden assumptions and poor communication.