Heavy AI Users Work 90 Minutes Longer Per Day, ActivTrak Data Shows

Employees who use AI most heavily work up to 90 minutes longer each day, according to new research from ActivTrak, a productivity tracking software vendor. The finding directly contradicts the company's own earlier study, which concluded AI had shortened the average workday by roughly nine minutes.

The latest report draws on behavioral data from 9,140 knowledge workers across 467 enterprises, collected over five months from April 1 to September 25, 2026. The earlier study, published this year, covered a far larger base: 443 million hours of work by 163,638 employees across 1,111 organizations, spanning January 1, 2023 through December 31, 2025. The discrepancy may stem partly from the different timespans and sample sizes, but it signals that enterprises should treat vendor claims about AI productivity gains with caution. Measuring those gains remains extremely difficult.

What do the numbers show?

Heavy AI users work between 8.42 and 9.08 hours per day, against 7.58 hours for employees not using AI, the latest report found. The prior study had reported a 2% decline in average workday length attributable to AI.

The newest data also points to more fragmented work patterns among heavy users:

• Focus sessions (uninterrupted blocks of concentrated work) run 22% shorter for heavy users: 37.8 minutes versus 48.2 minutes for light users.
• Heavy users switch applications 15% more frequently.
• Light users spend more active time in core job-specific applications: 81.4% of work time versus 78%.

"Over time, those interruptions add friction to high-value work that requires concentration, critical thinking, and sustained attention," said Sarah Altemus, manager of the ActivTrak Productivity Lab, in a blog post. "More frequent AI engagement doesn't necessarily translate into more focused work," she said.

The earlier report told a partly positive story. It found focus efficiency (the percentage of work time spent in focused, uninterrupted activity) fell to a three-year low of 60% with AI use, and average focused session time in 2025 declined 9% from 2023. Yet the same study found daily productive time rose 19 minutes, or 5%, to 6 hours and 36 minutes, while productive sessions grew 13%. Multitasking rose 12%, or more than 10 minutes, with AI in that dataset.

What is the 'swivel-chair tax'?

The latest research compared users of tools with embedded AI against users of web-based AI tools. Employees using embedded AI switched applications 0.6 times per hour. Users of web-based AI switched between 1.7 and 4.2 times per hour — up to four times more switching than employees who use no AI at all. ActivTrak labels the resulting overhead the "swivel-chair tax": the extra effort employees expend moving between applications.

Altemus flagged a management problem beyond the metrics. When employees finish tasks faster, managers may assume they have capacity for more responsibilities, she said. Without clear workload boundaries, time savings can be redirected into additional taxing work.

Can the data be trusted?

Sanchit Vir Gogia, chief analyst at Greyhound Research, credited ActivTrak with a useful view of how work moves across applications. He also noted the company sells the measurement capability behind the findings.

Recorded use depends on what the instrumentation can detect, Gogia observed. An employee using AI at work can still appear in the non-user group. Work on an unmanaged personal device sits outside the measured estate. An API call or a background agent can complete work without producing a desktop session. Passive screen time, offline meetings captured through a calendar, and paid hours are distinct measures. Switching applications is easier to observe than switching attention within one application.

"An improvement per paid hour can conceal effort left off the timesheet," Gogia said. "The board should be able to say whose day improved and whose work made that improvement possible."