Distraction at work has become one of the central productivity problems of the last decade. Personal devices have become harder to put down by design. The tools we use to do our jobs have multiplied. The pace of communication has accelerated. And the gap between what people want to do in a day and what they actually get done has turned into a real source of frustration.
Organizations have responded with effort and investment: focus apps, attention training, advice on personal habits, no-meeting Fridays, communication charters, protected focus time on team calendars. Most organizations have tried several. The intentions are good and the budgets are real.
But after all that effort, the problem isn’t solved. If it persists despite the investment, it’s reasonable to wonder whether we’ve been addressing the whole issue. Looking back at the list of interventions, most are aimed at the individual: their habits, their attention, their personal practice. Building better attention habits is genuinely hard, and worth supporting, but even people committed to the practice find themselves slipping back.
Why focus habits keep slipping
Gloria Mark’s research at UC Irvine helps explain why. She has studied attention at work for two decades. One of her central findings is that an environment of constant interruption changes how the brain operates. When people are pinged repeatedly throughout the day, the brain gradually adapts to that rhythm and starts to expect the next interruption. Over time, when the external pings stop, the brain produces them itself. Someone trying to build new attention habits is not only doing personal work. They are doing it inside an environment that is actively shaping the habit they are trying to change.
So what about that environment? How meetings get scheduled. How decisions get captured. The way information moves between people. How communication flows across every platform that’s accumulated over the last three years. This is the day-to-day reality of work, and it sets the limits on what attention can do, regardless of how disciplined any one person is.
275 interruptions a day
Microsoft’s 2025 Work Trend Index, based on a survey of 31,000 knowledge workers and analysis of trillions of Microsoft 365 productivity signals, gives us the most current picture of what that environment looks like. Across meetings, emails, and chats, employees are interrupted every two minutes during core work hours, adding up to 275 interruptions a day.
The human cost is exactly what you would expect. Eighty percent of the global workforce, employees and leaders alike, say they lack the time or energy to do their work. Forty-eight percent of employees and 52% of leaders describe their work as chaotic and fragmented. Meanwhile, 53% of leaders say productivity must still increase.
Numbers at that scale aren’t a story about anyone’s habits. They describe the conditions everyone is trying to focus in. If those conditions are producing 275 daily interruptions, how much of the individual distraction problem is actually downstream of the environmental one? How big a problem would personal distraction be if the environment had been better designed?
Why AI agents won’t fix this
Microsoft’s own response to this data is to argue for redesigning work around AI agents. The case is that agents can take on the repeatable tasks, freeing human capacity for the work that requires judgement, decision-making, and strategic thinking. There is real merit in that. Used well, AI can absorb routine work and give people back time for the thinking only they can do.
But technology applied to broken conditions doesn’t automatically produce the freedom it promises. We have seen this before. Email was supposed to reduce communication overhead but it has created the always-on culture we are now trying to recover from. Teams was supposed to streamline collaboration. Instead, the average knowledge worker now processes 117 emails and 153 Teams messages a day. The newer tool didn’t replace the older one. It was added on top.
AI agents will follow the same pattern unless something else changes alongside them. If the conditions producing 275 daily interruptions stay the same, adding AI to the mix will not reduce the fragmentation. It will compound it.
Microsoft itself names this risk. In their follow up paper, Breaking Down the Infinite Work Day, they warn that “we risk using AI to accelerate a broken system.” That is exactly the concern this blog is raising. The difference is what we do with the warning. Microsoft’s answer is to reimagine work through AI adoption. The argument here is that the reimagining has to happen first, or alongside, not after.
The conditions need attention before the tools do.