AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Workers Switch Apps 900 Times A Day—What Does That Mean For AI? on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • 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 analyzing 103 million application events from 1,017 workers at eight companies in India found an average of 908 app switches per workday. Workers switched less in the ten app uses after AI use, but had about 200 more switches on AI-use days; the observational study cannot establish that AI caused either pattern.

A Harvard Business School preprint analyzing 103 million application events from 1,017 office workers found an average of 908 app switches per workday—and a mixed relationship between switching and AI use. Workers switched among fewer apps in the ten app uses after using AI, but on days when they used AI they made about 200 more switches overall. Because the research is observational, it does not show that AI caused either pattern.

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. The researchers analyzed records covering 62,693 person-days and 56.9 million switches. The mean was 908.3 switches a day; the median worker made 814. The middle half of workers ranged from 547 to 1,130 switches per workday.

The study defines a switch as a change to a different application, rather than a move between documents in the same app. The average time spent in an application before switching was 27.8 seconds, while the median was 3.3 seconds. Those figures describe recorded application activity, not necessarily uninterrupted attention or the quality of a worker’s output.

AI use showed different associations at different scales. In the ten application uses following an AI use, workers used fewer apps, switched less, stayed about half a second longer in each app and moved through apps more predictably. Yet on days when a worker used AI, the total was about 200 more switches; days with more AI minutes were also more fragmented. The authors say the data cannot establish whether AI contributes to fragmentation, workers use AI on already fragmented days, or both.

At a glance
reportWhen: Preprint posted July 7, 2026, revised A…
The developmentA Harvard Business School preprint reports that workers switched apps an average of 908 times per day, with different patterns immediately after AI use and across days when AI was used.

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’s Mixed Link to App Switching

The findings complicate claims that AI tools automatically reduce workplace distraction. The pattern immediately after AI use looks more focused, while the broader daily pattern goes the other way. These are not necessarily contradictory: a short sequence following a tool interaction and the full workday measure different things.

For employers, the paper is a reason to be cautious about treating AI as a focus intervention. It does not measure whether employees completed tasks faster, produced better work, or felt less interrupted. A lower switch count in a short window is not proof of improved productivity, and a higher daily count does not show that AI caused workers to lose focus.

The researchers also 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% between organizations. That points to daily conditions as a relevant part of the pattern, though it does not identify which conditions are responsible. The source account suggests protecting concentrated morning work and watching midweek communication load; those are practical interpretations, not interventions tested by the study.

Products Worth Considering

Dell 32 Plus 4K Monitor (S3225QS), 31.5-Inch VA, 120Hz, FreeSync Premium

Dell 32 Plus 4K Monitor (S3225QS), 31.5-Inch VA, 120Hz, FreeSync Premium

Apple MacBook Neo 13-inch Laptop with A18 Pro Chip, 8GB Memory, 256GB SSD, Liquid Retina Display, and Blush Finish

Apple MacBook Neo 13-inch Laptop with A18 Pro Chip, 8GB Memory, 256GB SSD, Liquid Retina Display, and Blush Finish

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

As an affiliate, we earn on qualifying purchases.

How the Workday Data Was Collected

The researchers used data collected from November 2025 to January 2026 by Flowace, a commercial workforce-analytics service used by the eight participating organizations. Its monitoring software records active application changes, including app names and timing; the paper says data provided to researchers were anonymized under a data-use agreement. The sample included workers in software, accounting, financial services, e-publishing, law and marketing, all at organizations in India.

The employers determined which applications counted as productive, according to the source material. The paper says Harvard’s human-subjects committee approved the protocol. It also states the research received no external funding and that Flowace had no role in the study’s design, analysis or interpretation. One author disclosed equity in a private company unrelated to Flowace that uses AI to create enterprise digital twins.

The paper was posted on arXiv on July 7, 2026, and revised on August 7. It identifies itself as a preprint under review and warns that its findings may change during peer review. Its app-switching estimates are also compared with earlier attention research, but the authors’ source material cautions that methods differ, so those figures should not be read as a precise measure of change over time.

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

— Sumer S. Vaid and Ashley V. Whillans, in the paper

Causation and Generalizability Remain Open

The study is observational, so it cannot show whether AI use led to more switching or whether workers turned to AI because their days were already fragmented. The short-window association after AI use does not resolve that question. The paper also does not establish whether the recorded changes in app use affected productivity, work quality, stress or employee experience.

The sample covers eight organizations in India and does not include phone activity, according to the source material. It may not represent workers in other countries, industries or workplace arrangements. Monitoring software records active application changes, but the available description does not show what workers were thinking or doing away from their computers. The paper remains a preprint under peer review, so its estimates and conclusions may change.

Peer Review and Further Evidence

The paper’s next stated milestone is peer review; the authors say the findings may change through that process. The source material does not give a publication date or identify a later study. Until additional evidence is available, readers should treat the results as a description of patterns in this specific dataset, not as a test showing that AI improves or harms focus.

For teams considering AI tools, the study offers no basis for buying them as a way to reduce app switching. Employers and workers can instead look at when interruptions occur in their own workflows and whether particular tools or routines help with specific tasks. Any conclusions about productivity would require measures beyond the app-switching counts reported here.

Key Questions

Did workers switch apps 900 times every day?

The study reported a mean of 908.3 switches per workday among 1,017 workers. The median was 814, and the middle half of workers ranged from 547 to 1,130. The number is an average for this study sample, not a universal figure for all workers.

Did AI reduce app switching?

The results were mixed. In the ten app uses after AI use, workers switched less among apps; across whole days, workers using AI made about 200 more switches than on days without AI use. The study cannot show that AI caused either pattern.

Who and where did the study cover?

The researchers studied 1,017 employees at eight organizations in India, across several industries. Flowace monitoring software recorded application activity from November 2025 to January 2026.

Has the paper been peer-reviewed?

No. The paper was posted as a preprint under peer review. Its authors warn that findings may change during that process.

Does the study show that AI improves productivity?

No. It measures application activity and switching patterns, not whether work became faster, better or more productive. The study does not establish that AI caused the observed changes.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

Six Patterns That Can Make AI Writing Less Pleasant To Read

A preprint comparing 14 language models with human writing identifies six patterns that may make AI-generated text harder to read.

What It Costs To Retire Around The World: I Fact-Checked The Viral $2 Million Graphic

A viral retirement graphic uses a consistent 25-times formula, but lacks cost sources and omits pensions, household details and early-retirement risks.

Inside The Setup: UGREEN NASync DXP6800 Pro For Our AI Fleet

A small AI publishing operation has bought a six-bay UGREEN NAS after backup-sync failures. The device is delivered but not yet configured or tested.

What Should The UK Consider About AI In Its Youth Employment Plans?

Alan Milburn’s review proposes legal support and AI training for young people who are NEET; the government has accepted its principles but not its benefit reforms.