> ## Content Index > Fetch the complete content index at: https://www.zilck.com/llms.txt > Use this file to discover other available public pages before exploring further. # AI Users Just Lost 23 Minutes of Focus a Day - URL: https://www.zilck.com/ai-users-just-lost-23-minutes-of-focus-a-day/ - Published: 2026-09-09T23:16:35.000Z - Updated: 2026-09-09T23:16:35.000Z - Description: AI really did give your team hours back — 443 million hours of work data show exactly where those hours went instead. - Author: Zilck Team - Tags: Productivity The average stretch of focused work now lasts 13 minutes and 7 seconds. That's a three-year low, down 9% since 2023, based on [ActivTrak's Productivity Lab](https://www.prnewswire.com/news-releases/activtrak-productivity-lab-analyzes-443m-hours-of-work-activity-revealing-ai-is-accelerating-work--not-replacing-it-302710985.html?ref=zilck.com) analysis of 443 million hours of real workplace activity across 1,111 organizations and 163,638 employees. Here's the part that should stop you: employees who use AI heavily lost 23 minutes of daily focus time. Non-AI users barely moved. The tools meant to protect deep work are, on the evidence, consuming it — and the reason has almost nothing to do with the tools. ## The Time Really Was Freed Start with the good news, because it's real and rigorously measured. In a randomized controlled trial run across [66 firms and more than 7,000 workers](https://aiinstitute.hbs.edu/ai-is-giving-workers-more-focus-time-now-what/?ref=zilck.com) by researchers from Microsoft Research and Harvard Business School, employees who used Microsoft 365 Copilot heavily cut their email time by 31% — about 3.6 hours back every week. Collaborative documents got finished roughly 25% faster. But one finding in that trial explains everything that follows. Meeting attendance and meeting duration didn't budge. As the researchers put it, they saw "larger changes in behaviors that workers can adjust independently and less movement in behaviors that require coordination with colleagues." AI compresses the work you do alone. It has no authority over the work you do with other people. And that asymmetry is where your afternoon goes. ## Where the Hours Went The ActivTrak data traces the destination precisely. Collaboration time surged 34%, adding 13 minutes to reach 52 minutes a day. Chat and messaging volume rose 145%. Email volume — not time spent, volume — climbed 104%. Multitasking rose 12%, to an hour and 33 minutes daily. Then look at the weekend. Saturday productive hours jumped 46%, to an average of 4 hours 37 minutes, with people starting an hour and 24 minutes earlier than they used to. Sunday hours rose 58%. Meanwhile, the workday itself shrank by all of 2%. That combination is the whole story. Everyone got faster at producing things, so everyone produced more, so everyone had more to review, respond to, and coordinate around. Your AI-assisted colleague's output arrives in your inbox as input. Multiply across a company and individual speed becomes collective congestion. "AI isn't reducing work, it's increasing the speed and density of how work happens," says Gabriela Mauch, Chief Customer Officer and head of the ActivTrak Productivity Lab. ## It's Not Burnout. It's Drift. Now the finding that complicates the obvious narrative — and the reason this deserves more than a scolding about notifications. Burnout risk actually fell 22%, down to 5% of employees. The share of people maintaining healthy work patterns hit a three-year high of 75%. Overutilization dropped 42%. By the standard wellbeing measures, the AI era is treating workers better, not worse. But disengagement risk rose, from 19% to 23% of employees. Read those together and a different problem emerges. People aren't being crushed; they're being scattered. Thirteen-minute fragments are long enough to answer things and too short to build anything. That produces a workforce that is busy, healthy, and quietly checked out — which is a harder condition to detect than burnout and, in a knowledge business, more expensive. ## The Measurement Gap ActivTrak's own framing names the underlying failure: a growing "AI measurement gap." Companies now run seven AI tools on average, up from two in 2023, and 80% of employees use them. Almost none of those companies can say what the tools did to focus, capacity, or output quality. There's a striking number buried in the data on that point. Employees who spend 7–10% of their work hours in AI tools show the strongest productivity profile — and only 3% of users land in that band. The rest are above it or below it, mostly without knowing a band exists. ## What This Means The lever here is organizational, not personal. Three moves follow from the evidence. Cut the coordination layer, not the tool. The RCT is unambiguous: AI can't shorten your meetings because meetings are a group decision. If your team gained hours on email and lost them to collaboration, the fix is fewer standing meetings, fewer required reviewers, and clearer decision rights — changes only a manager can make. Set a norm for AI-generated volume. When drafting costs nothing, sending costs nothing, and reading still costs everything, the polite default becomes overproduction. Agree on what gets summarized, what gets skipped, and what nobody has to read. Measure focus, not adoption. Tool licenses are the easiest metric and the least useful. Track uninterrupted session length before and after a rollout; if sessions are fragmenting, the productivity story is being told backward inside your own company. Software didn't steal the 23 minutes. They were released into an environment with nothing to catch them, and the calendar absorbed them the way it absorbs everything else. Companies that spend the next year designing where those hours land will get the compounding gains. Everyone else will get a faster version of the same day.