🔍 Read the full analysis: Workers Switch Apps 900 Times A Day. Here's What Harvard Found About AI on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
A Harvard Business School preprint analyzed 103 million application events from 1,017 office workers at eight companies in India. Workers switched apps an average of 908 times per day; although switching eased in the ten app uses after AI use, AI-use days had about 200 more switches overall. The observational data cannot show that AI caused either pattern.
A Harvard Business School preprint analyzing 103 million application events from 1,017 office workers found that employees switched apps an average of 908 times per day, with a mixed pattern around generative AI: switching eased immediately after AI use but was higher on days when workers used AI. The study is observational and does not establish that AI caused either change, a distinction that matters as employers weigh AI tools and workplace monitoring as ways to improve focus.
The paper, “Digital Fragmentation and Generative AI Use Across 103 Million Application Events,” was written by Sumer S. Vaid and Ashley V. Whillans of Harvard Business School. It examined anonymized computer-activity records from workers at eight organizations in India, covering software, accounting, financial services, e-publishing, law and marketing. The data span November 2025 through January 2026. The manuscript describes itself as a preprint under review and says its findings may change through peer review.
Across 62,693 person-days, researchers counted 56.9 million switches between applications. The mean was 908.3 switches per day; the median employee had 814, and the middle half of employees ranged from 547 to 1,130. An application remained active for a mean of 27.8 seconds before a switch, while the median interval was 3.3 seconds. The authors’ comparison with earlier attention studies is qualified: differences in methods mean the figures should be treated as a trend signal, not a precise measure of change over time.
The AI results depend on the time frame. Looking at the ten application uses after an AI use, workers used fewer apps, switched less, stayed about half a second longer per app and moved in a more predictable sequence. But on days workers used AI, they made about 200 more switches per day; days with more AI-use minutes were also more fragmented. The researchers did not assign AI use or control workers’ schedules, so the data cannot determine whether AI contributed to more switching, workers used AI on already fragmented days, or both.
App switching and AI · one moment
What happens around one AI query?
The study zoomed in on every AI use: the ten app uses before it, and the ten after. Up close, work calms down a little. Zoom out to the whole day, and the picture changes.
- 1 · A workday of switchingWorkers changed apps 908 times a day on average. One app held their focus for 27.8 seconds on average, 3.3 seconds at the median.
- 2 · AI comes in at a busy momentPeople used AI a median of 13 times a day, for a median of 7 seconds. Fragmentation peaked in the hour of AI use.
- 3 · The next ten app usesFewer different apps, fewer switches, about half a second longer per app, more predictable. Real, but small.
- 4 · Zoom out: the whole dayOn days with AI use, the same people made about 200 more switches. Whether AI calms work or busy moments bring AI: unknown.
Source: Sumer S. Vaid, Ashley V. Whillans (Harvard Business School), “Digital Fragmentation and Generative AI Use Across 103 Million Application Events”, arXiv:2607.06681v2, preprint under review. Blocks are schematic, not to scale.
Graphic: Powered by Thorsten Meyer AI
AI’s Mixed Link to Focus
The results complicate claims that adding generative AI automatically reduces interruptions or helps employees concentrate. The immediate post-use pattern points to fewer switches across the next ten app uses, while the full-day association points the other way. Those findings can coexist because they measure different windows: a short sequence after AI use versus the whole workday.
For managers, the study is a reason to be cautious about treating AI adoption as a focus intervention. It does not measure whether workers completed better work, saved time, or felt less interrupted. Its findings also suggest that the day’s conditions matter: 44.6% of variation in fragmentation was attributed to day-to-day changes within the same person, compared with 35.8% between individuals and 19.6% between organizations. These proportions describe variation in this dataset, not proof of what causes it.
The results also raise practical questions about workplace analytics. The records came from software installed on employees’ computers, making measurement dependent on employer monitoring systems and their definitions of productive apps. The paper offers information about application switching, not a complete account of attention, employee experience or work quality.
Products Worth Considering

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

Bose QuietComfort Wireless Bluetooth Noise-Cancelling Headphones, Cypress Green

DJI Mini 4K Camera Drone Combo with 2 Batteries
As an affiliate, we earn on qualifying purchases.
How the Activity Was Recorded
Flowace, a commercial workforce-analytics service used by the participating organizations, supplied the records. Its monitoring software logged when an employee changed the application in active use, along with details such as the application name and start and end times. Employers decided which apps counted as productive. The researchers received anonymized data under a data-use agreement, and the study protocol received approval from Harvard’s human-subjects committee.
The sample included 1,017 workers across eight firms, with a median of 58 workdays per person. The paper says the research had no external funding and that Flowace provided no commercial support or role in study design, analysis or interpretation. One author disclosed equity in a private company unrelated to Flowace that uses AI to create enterprise digital twins. The AI applications counted included products such as Anthropic’s Claude.
The headline that spread on November 18 captured the average switching count but presented the AI relationship too simply. The paper’s own results distinguish what happens just after AI use from how switching compares across entire days. It was first posted to arXiv on July 7, 2026, and revised on August 7; the manuscript identifies itself as version 4.0 dated August 4 and says it remains under review.
“The manuscript is currently under review, and the findings may change based on the peer review process.”
— The authors, in the preprint
What the Data Cannot Establish
The study does not establish whether AI use caused more or fewer app switches. Workers were not randomly assigned to use AI, and the researchers note that AI may reorganize a workday, workers may turn to AI when their day is already consolidating, or both. The dataset shows associations, not the direction of cause and effect.
The sample covers office workers at eight organizations in India, not a representative cross-section of workers worldwide. The analysis does not cover phone use and does not establish whether switching affected work quality, productivity or employee well-being. App-change records are also a limited proxy for attention: a switch does not by itself reveal why it happened or whether a worker was interrupted.
The paper remains under peer review, and the authors warn its findings may change. The source material does not provide a final peer-reviewed assessment or establish whether the patterns hold in other countries, workplaces or types of work.
Peer Review and Workplace Decisions
The next formal step reported in the paper is peer review; the authors say the findings could change during that process. No decision date is provided. Further studies would need to test whether AI itself changes switching behavior and whether any change improves outcomes workers and employers care about, such as concentration, completion time or work quality.
Until then, the authors’ findings support caution rather than a claim that AI is a proven focus tool. Teams considering the results can examine when interruptions cluster during the week and protect time for concentrated work, but those are practical responses—not interventions tested by this study. The paper does not establish that buying an AI product will reduce app switching.
Key Questions
What did the Harvard study find about app switching?
In the analyzed records, workers switched between applications a mean of 908.3 times per day. The median was 814 switches per workday. The count refers to changes between applications, not every change of document or window.
Did AI reduce app switching?
The result differed by time frame. Switching was lower across the ten app uses after an AI use, but workers had about 200 more switches on AI-use days overall. The observational study cannot show that AI caused either pattern.
Was the study peer-reviewed?
No. The paper describes itself as a preprint under review and says its findings may change during peer review.
Who was included in the study?
The analysis covered 1,017 employees at eight organizations in India, across software, accounting, financial services, e-publishing, law and marketing. It used workplace computer-activity records collected from November 2025 through January 2026.
Does the research show AI improves productivity?
No. The study measured application activity and switching, not whether work became faster, better or more productive. It also did not establish the effects on employee well-being.
Source: ThorstenMeyerAI.com
