What Happens When AI Meets 900 Daily App Switches?
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: What Happens When AI Meets 900 Daily App Switches? on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

A Harvard Business School preprint analyzed 103 million application events from 1,017 office workers in India and found an average of 908 app switches per workday. Workers switched less in the short sequence of app uses following AI use, but made about 200 more switches on AI-use days overall; the observational data cannot establish why.

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 workday, while days with AI use involved about 200 more switches than days without it. The findings complicate claims that AI necessarily reduces digital distraction: workers switched less in the app sequence immediately after using AI, but the observational study cannot show that AI caused either pattern.

The paper, titled 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. The records cover November 2025 through January 2026 and include a median of 58 workdays per person. Across 62,693 person-days, the researchers counted 56.9 million switches between applications. The mean was 908.3 switches per day; the median employee made 814, and the middle half ranged from 547 to 1,130.

The analysis also found that an application remained active for a mean of 27.8 seconds before a switch, with a median of 3.3 seconds. In the ten app uses after an AI use, workers used fewer apps, switched less, stayed about half a second longer in each app and followed more predictable sequences. Yet on days when workers used AI, they made about 200 additional switches per day on average, and days with more AI minutes were more fragmented.

The researchers stress that the data show associations, not cause and effect. The authors suggest several possible explanations: AI may reorganize the workday, workers may use AI during days that are already becoming more consolidated, or both patterns may be occurring. The study measured app activity, not whether work improved, and did not test whether AI made employees more productive.

At a glance
reportWhen: Preprint posted July 7, 2026, revised A…
The developmentA preprint analysis of workplace software logs finds that workers averaged 908 app switches per workday, while AI use was associated with both less switching immediately afterward and more switching across the day.

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.

10 app uses before AI median 7 seconds 10 app uses after Whole AI day: about 200 more switches
  1. 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. 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. 3 · The next ten app usesFewer different apps, fewer switches, about half a second longer per app, more predictable. Real, but small.
  4. 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 Use and the Shape of Workdays

The findings matter because claims about AI and productivity often rely on whether a tool appears to save time or reduce interruptions. This study suggests that the relationship may depend on the time window examined: the immediate sequence after AI use looked less fragmented, while the day as a whole was more fragmented when AI was used.

That distinction cautions employers against treating AI adoption as a proven focus intervention. The study does not establish that AI creates extra task switching, nor that it improves concentration after each use. It does show why short-term interaction patterns and whole-day measures can tell different stories. For workers and managers, the data point to daily workload and communication patterns as possible sources of fragmentation that deserve attention alongside individual tool use.

The authors found that 44.6% of variation in fragmentation came from day-to-day changes within the same person, compared with 35.8% from differences between people and 19.6% from differences between organizations. That breakdown suggests that a worker’s experience may shift substantially from one day to another, rather than being explained only by their habits or employer.

Samsung 55-Inch S90H OLED 4K UHD Smart TV Bundle with Full-Motion Mount and Surge Protector

Samsung 55-Inch S90H OLED 4K UHD Smart TV Bundle with Full-Motion Mount and Surge Protector

As an affiliate, we earn on qualifying purchases.

What the Workplace Logs Measure

The study draws on software records collected by Flowace, a commercial workforce-analytics service used by all eight participating organizations. Its software recorded when the active application changed, along with app names and other information described in the paper. Employers determined which applications counted as “productive.” Researchers received anonymized data under a data-use agreement, and the paper says Harvard’s human-subjects committee approved the protocol.

The sample covered organizations in India: two software firms, two accounting firms, and one each in financial services, e-publishing, law and marketing. The paper reports that Flowace did not fund the research or take part in study design, analysis or interpretation. It also states that one author holds equity in a private company unrelated to Flowace that uses AI to create enterprise digital twins. The authors report no external funding.

The paper first appeared on arXiv on July 7, 2026, with a revision posted August 7. It is a preprint, not a peer-reviewed publication. The manuscript itself warns that its findings may change during review. The study’s average of 908.3 switches is a count of changes to a different application; moving between two documents in the same app counts as one use, according to the source account.

“The manuscript is currently under review, and the findings may change based on the peer review process.”

— Vaid and Whillans, in the preprint

Limits of the AI Comparison

The study is observational, so it cannot establish whether AI use led to more switches or whether workers were more likely to use AI on already-busy days. The short-term and full-day patterns coexist in the data, but the direction of influence remains unknown.

The sample includes office workers at eight organizations in India, tracked through a particular workforce-monitoring service. It does not establish whether the same patterns apply in other countries, on phones, or in jobs outside the sampled industries. The researchers counted application activity, not the quality or output of work, so the study does not show whether switching affected performance, well-being or productivity.

The reported app-focus durations also need care in interpretation. The paper compares them with results from earlier attention research, but the source account notes that methods differ. The figures can suggest a possible trend, not a precise, directly comparable decline. The study also does not show that AI tools caused the reported differences between days.

Peer Review and Employer Decisions

The paper remains under peer review, and its results may change as reviewers assess the methods and findings. The next useful evidence would include a peer-reviewed account, clearer tests of how AI use relates to work patterns over time, and studies measuring outcomes such as work quality rather than app activity alone.

Until then, the study’s authors and the reported findings do not establish AI as a focus aid. Employers considering the results can distinguish between short periods after AI use and the broader shape of a workday, while treating the numbers as an account of the sampled workplaces—not a universal measure of office work. The source report suggests protecting concentrated morning time and paying attention to midweek communication load, but these are practical interpretations, not interventions tested by the study.

Key Questions

Did workers really switch apps about 900 times a day?

In the study sample, workers made an average of 908.3 switches per workday. The median was 814. These figures come from logged changes between applications, not movements between documents within the same app.

Did AI reduce app switching?

The results differed by time window. In the ten app uses after AI use, workers switched less and used fewer apps. Across the full day, however, days with AI use involved about 200 more switches on average. The study cannot show that AI caused either pattern.

Was the study peer-reviewed?

No. The paper was posted on arXiv and is under peer review. The manuscript warns that findings may change during that process.

Who was included in the study?

The researchers analyzed records from 1,017 employees at eight organizations in India, across software, accounting, financial services, e-publishing, law and marketing. The records covered November 2025 through January 2026.

Does the study show that AI makes workers more productive?

No. It measured application activity and switching, not the quality or amount of work completed. It does not establish whether AI improved productivity, focus or well-being.

Source: ThorstenMeyerAI.com

You May Also Like

Is The Future Of AI Operations In Data Center REITs? Trends Uncovered

Exploring whether data center REITs are becoming central to AI operations, based on recent signals and industry shifts.

Understanding Anthropic’s Early Self-Improving AI And Its Potential Impact

Anthropic has shown an early version of a self-improving AI, raising questions about autonomy, safety, and impact on AI development timelines.

Six Ideas From My Book To Help You Understand This Month In AI

Thorsten Meyer uses six ideas from his new book to interpret AI stories, from bank-account agents to falling model costs and workplace changes.

DC Appeals Court Sides With Pentagon On Blacklist Of Anthropic – The Hill

The Hill reports a DC appeals court ruled for the Pentagon in its dispute over a blacklist of Anthropic; the ruling’s details and effects remain unclear.