89% of executives see faster individuals, but only 6% can name AI ROI

Most AI rollouts are making one part of the org go faster and another part stay stuck. Dr. Molly Sands, head of the Teamwork Lab at Atlassian, says that in practice companies are optimizing how individuals use AI, instead of how teams work together. She made that case during a fireside chat with VentureBeat senior technology contributor Sam Witteveen at VB Transform 2026.

The data backing her point is blunt. In Atlassian’s annual State of Teams Report, this year surveying 12,000 global knowledge workers and interviewing roughly 200 Fortune 1000 executives, Sands highlighted a big disconnect between activity and value. “89% of those executives told us that individuals are speeding up in their companies, and only 6% of them said they could point to specific examples of clear ROI,” she said. At the same time, roughly 14% of teams had translated AI usage into real value, meaning an organization can contain a handful of teams capturing returns while others see little or nothing.

If you are a founder, operator, or investor, that last part should raise your eyebrows. It suggests AI adoption is not a uniform uplift you can assume once a tool gets deployed. It is more like a talent and operating-model problem, where some teams redesign work and others just add AI to the same old mess. Sands frames it as an “optimization” mistake: leaders are treating AI like an individual productivity upgrade, when the value actually shows up when teams change how decisions, work, and knowledge connect.

Atlassian points to three characteristics shared by the leading teams: context, workflows, and culture. On context, the top performers build what Atlassian calls a context graph, capturing goals, decisions, and organizational knowledge in shared digital records instead of leaving it in individual memory. Using products such as Jira and Confluence, the graph connects work items, goals, and the people doing them, giving AI access to organizational context it needs. In plain English: the AI is more useful when it is plugged into how the organization already thinks and operates, not when it is guessing based on fragments.

On workflows, the fastest teams do not just accelerate isolated tasks. They redesign end-to-end processes. Sands’ warning is that if you only speed up individuals who are aimed in different directions, they do not magically align. They “very quickly start to crash into each other,” she said. This is where teams, not employees, become the unit of performance. In many orgs, cross-functional handoffs, approvals, and feedback loops are the real bottleneck. Speeding up the wrong step can make those seams more obvious, not less.

Culture completes the loop. The teams pulling ahead work under leaders who explicitly encourage learning and experimentation, and they make it clear that some experiments will fail. That matters because AI can introduce uncertainty: outputs can be inconsistent, assumptions can be wrong, and the team has to learn what “good” looks like. If leadership treats AI as a one-shot transformation rather than an iterative practice, teams will avoid experiments, hide failures, and keep using AI in narrow, low-risk ways that rarely unlock ROI.

So what actually moves the needle? Sands emphasizes experimentation and constraints as the fastest route to learning. The leading teams deliberately impose constraints on how they work, from breaking tasks down into the smallest practical unit of work, like a single story point, to committing to write no code by hand for a week. The point is not to make constraints permanent. Sands says most of it is not sustainable to do forever, but it is a “really, really fast way to learn.”

She also argues that one obstacle is not the technology at all. Employees are figuring out AI on their own. That produces different prompts, agents, and assumptions, which creates another layer of unspoken knowledge inside teams. The catch is that this local improvisation rarely translates into organizational performance. To counter that, Atlassian experimented with AI working agreements at the start of projects. Teams decided not only what they would use AI for, but what they would deliberately avoid using it for, which agents they would share, and what common skills would keep everyone working from the same context. Sands says teams that adopted the practice used AI more, moved faster, made better decisions, and produced higher-quality work.

Zoom out and the lesson is almost uncomfortable: AI is not creating entirely new management problems so much as exposing old ones. Teams have always struggled with hidden assumptions and different mental models of how work actually gets done. AI makes those gaps more consequential. For boards and executives, that has a second-order implication: the governance questions do not end at procurement. They expand to how work is documented, how knowledge is shared, how workflows are measured end-to-end, and how teams learn together without turning AI into a solo sport.