ActivTrak finds heavy AI users work longer days and have shorter focus sessions

AI may speed up a task without shortening the working day. ActivTrak’s latest workplace study found that heavy AI users worked longer and had shorter stretches of uninterrupted focus. The findings, detailed by CIO on November 3, challenge a simple time-saving narrative—but do not establish that AI caused those differences.

The productivity-tracking software company analyzed behavioral work data from 9,140 knowledge workers at 467 enterprises, collected between April 1 and September 25, 2026. Its measurements follow activity across applications. That offers a view of how employees spend their digital workdays, rather than a direct answer to how much valuable work they finish.

Longer days, more fragmented attention

ActivTrak recorded daily work durations of 8.42 to 9.08 hours for heavy AI users, versus 7.58 hours for nonusers. Gizmodo’s coverage adds a useful definition: heavy users spent more than 45 minutes using AI across at least five sessions daily. They also logged 18 more minutes of after-hours work than nonusers.

The focus comparison used a different baseline: heavy users’ uninterrupted sessions averaged 37.8 minutes, against 48.2 minutes for light users—a 22% gap. The distinction matters when reading the results: the longer-day finding compares heavy users with nonusers, while the focus finding compares two groups already using AI.

Tool placement also tracked with different behavior. Users of embedded AI switched applications 0.6 times an hour, compared with 1.7 to 4.2 times for users of web-based AI. ActivTrak calls the extra movement between tools a “swivel-chair tax”: effort spent moving across applications rather than staying within the work environment.

The earlier gains measured something different

Earlier this year, ActivTrak published a much larger study covering 443 million work hours from 163,638 employees at 1,111 organizations. That dataset spanned January 2023 through December 2025. It found average workdays decreased by about nine minutes, while daily time classified as productive increased by 19 minutes to six hours and 36 minutes.

Even that more positive picture contained signs of fragmentation. Focus efficiency—the share of work time spent in uninterrupted activity—fell to a three-year low of 60%. Average focused sessions in 2025 were 9% shorter than in 2023. More recorded productive time and less sustained attention appeared together, rather than representing mutually exclusive outcomes.

These are different samples over different periods, not a before-and-after experiment following the same employees. The latest results therefore do not simply overturn the earlier gains. They show why the length of a workday, time labeled productive and the ability to concentrate need to be read as separate measures.

The [enterprise] board should be able to say whose day improved and whose work made that improvement possible.

Sanchit Vir Gogia, chief analyst at Greyhound Research, speaking to CIO

What the activity tracker cannot settle

Gogia cautioned that employees can use AI and still appear in the nonuser group. Work on unmanaged personal devices may be invisible, while background agents or software connections can finish tasks without a desktop session. Application switching is also easier to observe than shifts of attention inside a single application.

He also noted a commercial interest: ActivTrak sells the tracking capability behind these measurements. Passive screen time, calendar-recorded meetings and paid hours are not interchangeable. His concern is what happens after an initial task gets faster, including whether an apparent improvement per paid hour conceals effort that never reaches a timesheet.

Workload decisions offer another explanation for where saved time might go. Sarah Altemus, manager of the ActivTrak Productivity Lab, warned that managers may treat faster task completion as room for more responsibilities. Without clear workload boundaries, she said, time savings can become additional work—not additional time employees get back.