🔍 Read the full analysis: App Switching At Work: What The Harvard Study Says About AI on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
A Harvard Business School preprint analyzed 103 million application events from 1,017 office workers in India. Workers switched apps an average of 908.3 times per day; around AI use, switching fell in the next ten app uses, but days with AI use had about 200 more switches on average. 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 at eight companies in India found that workers switched apps an average of 908.3 times per day. Its AI results are mixed: workers switched less in the ten app uses after an AI use, but days when they used AI involved about 200 more switches per day on average. The study is observational and does not establish 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 came from Flowace, a commercial workforce-analytics service used by the participating employers. The researchers analyzed 56.9 million app switches across 62,693 person-days.
The average was 908.3 switches per day; the median worker made 814, and the middle half of workers ranged from 547 to 1,130. An app remained active for a mean of 27.8 seconds before a switch, although the median duration was just 3.3 seconds. These measures describe changes between applications, not every movement between documents or tasks within one app.
The AI findings differ by timescale. In the ten application uses after an AI use, employees used fewer apps, switched less, remained in each app about half a second longer and followed a more predictable sequence. At the whole-day level, however, days involving AI use had about 200 additional switches, and more minutes using AI were associated with greater fragmentation. The paper does not establish why those patterns occur or whether AI use preceded the more fragmented 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.
- 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 Use Shows Two Switching Patterns
The findings caution against treating AI as an automatic focus aid. A short-term sequence after using an AI tool looked less fragmented, while the broader workday on AI-use days looked more fragmented. Those observations can coexist: a local change after one use does not establish that the overall day becomes calmer.
For workers and employers, the distinction matters when evaluating claims that AI saves attention or reduces interruptions. The authors say the data cannot tell whether AI use reorganizes a workday, whether workers are more likely to turn to AI on days already becoming consolidated, or whether both processes are involved. The study also did not measure whether work quality or output improved. It is not evidence that buying an AI tool will improve concentration.
The paper reports that 44.6% of variation in fragmentation came from day-to-day differences within the same person, compared with 35.8% from differences between workers and 19.6% between organizations. That points to daily conditions as an important part of the measured pattern, though it does not identify which conditions caused it. The authors also report that fragmentation peaked on Tuesdays and eased after weekends and holidays.
Popular right now

Bose QuietComfort Ultra Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Driftwood Sand

Shark Pet Cordless Stick Vacuum, Up to 40-Minute Runtime, Grey, IX141

Vitamix VX1 Professional-Grade Blender, 64 oz, Black
As an affiliate, we earn on qualifying purchases.
What the Tracking Data Captured
The study used records generated by Flowace monitoring software on employees’ computers. The paper says the software recorded when the active application changed, along with details including app name and start and end time; in some cases, it also recorded window titles and web addresses. Employers determined which applications counted as productive. The researchers received anonymized records under a data-use agreement, and the study protocol was approved by Harvard’s human-subjects committee.
The eight participating organizations operated in software, accounting, financial services, e-publishing, law and marketing. The sample covered a median of 58 workdays per person. The AI tools counted included Anthropic’s Claude, which was the smallest of five tools represented in the sample. These boundaries matter: the records concern work computers at a particular group of Indian employers, not workers in every country or app use on phones.
The paper was posted on arXiv on July 7, 2026, and revised on August 7. It identifies itself as a preprint under peer review and warns that its findings may change during review. The authors say the research had no external funding; Flowace provided no commercial support and had no role in the design, analysis or interpretation, according to the paper.
“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 preprint
Limits on the AI Findings
The analysis shows associations, not cause and effect. It cannot establish whether AI use led to more switching across a day, whether workers chose AI on days that were already more fragmented, or whether other factors shaped both. The short sequence after AI use likewise does not prove that AI reduces distraction.
The sample is limited to 1,017 workers at eight organizations in India, observed through employer-used monitoring software. The paper does not cover phone use, establish whether app switching harmed productivity, or test whether the work became better. Its findings remain subject to peer review, and the authors say they may change. The reported daily count and app-duration figures also describe activity as captured by this system, rather than a universal measure of attention.
Peer Review and Workplace Decisions
The next formal development is peer review, which could lead to changes in the manuscript or its conclusions. Until then, the results are best treated as an observational report from a specific group of workplaces, not a general finding about all employees or AI tools.
Employers considering the findings can distinguish between app switching immediately after AI use and the pattern across an entire day, while avoiding claims that the study proves an AI product improves focus. The authors’ data suggest that fragmentation varies substantially from day to day, but the study does not test workplace interventions such as protected focus periods or reduced meeting loads. Whether such changes would reduce switching, and whether doing so would improve work, remains untested here.
Key Questions
What did the Harvard study find about app switching?
Workers in the sample switched applications an average of 908.3 times per day, based on computer records from eight organizations in India between November 2025 and January 2026.
Did AI reduce app switching?
The results differed by timescale. Switching fell in the ten app uses after an AI use, but days with AI use had about 200 more switches per day on average. The observational study cannot show that AI caused either result.
Is this a peer-reviewed study?
No. The paper is a preprint under peer review. Its authors say the findings may change during that process.
Who was included in the research?
The study tracked 1,017 employees at eight organizations in India, across industries including software, accounting, financial services, publishing, law and marketing. It used application-event records from Flowace monitoring software.
Does the study show that AI improves productivity?
No. It measured application activity and switching, not whether work quality or output improved. It does not establish that using AI makes workers more productive or more focused.
Source: ThorstenMeyerAI.com
