Data snapshot: July 2026
The Glasp Reading Report 2026
Glasp is a social highlighting platform used by 1,000,000+ readers. This report shares what that community actually reads, highlights, and asks AI about, as anonymized aggregates. Methodology at the bottom.
Atomic Habits
is the most-highlighted Kindle book on Glasp, ahead of Building a Second Brain
22.4%
of AI summarizing targets research and academic sources, the largest curated category
57%
of language-identified YouTube video summaries are for videos in languages other than English
The most-highlighted Kindle books
Ranked by how many Glasp readers highlighted each book, per edition language. The English list is a portrait of what serious readers underline: habit formation, note-taking systems, money psychology, and Stoicism. The Japanese list adds a distinct canon of business and philosophy titles that barely overlaps with it, and the Spanish and Portuguese lists are led by translated classics alongside homegrown authors. Each title links to its highlights page on Glasp.
| # | Title | Author | |
|---|---|---|---|
| 1 | Atomic Habits | James Clear | Amazon |
| 2 | Building a Second Brain | Tiago Forte | Amazon |
| 3 | $100M Offers | Alex Hormozi | Amazon |
| 4 | The Almanack of Naval Ravikant | Eric Jorgenson | Amazon |
| 5 | Never Split the Difference | Chris Voss | Amazon |
| 6 | The Daily Stoic | Ryan Holiday | Amazon |
| 7 | Thinking, Fast and Slow | Daniel Kahneman | Amazon |
| 8 | How to Take Smart Notes | Sönke Ahrens | Amazon |
| 9 | Sapiens | Yuval Noah Harari | Amazon |
| 10 | Deep Work | Cal Newport | Amazon |
| 11 | Tools of Titans | Tim Ferriss | Amazon |
| 12 | The Psychology of Money | Morgan Housel | Amazon |
| 13 | The Subtle Art of Not Giving a F*ck | Mark Manson | Amazon |
| 14 | The 7 Habits of Highly Effective People | Stephen R. Covey | Amazon |
| 15 | The Obstacle Is the Way | Ryan Holiday | Amazon |
| 16 | Four Thousand Weeks | Oliver Burkeman | Amazon |
| 17 | Essentialism | Greg McKeown | Amazon |
| 18 | Zero to One | Peter Thiel | Amazon |
| 19 | The ONE Thing | Gary Keller | Amazon |
| 20 | Man's Search for Meaning | Viktor E. Frankl | Amazon |
As an Amazon Associate, Glasp earns from qualifying purchases.
What people ask AI to summarize
Shares of hand-curated categories among the top pages users summarize with Glasp's AI (excluding video, entertainment, and auto-triggered pages; see methodology). The picture is serious: research papers are the largest named category, and the second is AI tools themselves. People summarize what AI told them.
journal articles, preprints, and academic databases
outputs from AI chat and research tools, summarized again
discussion threads and professional networks
shared documents, wikis, and internal knowledge bases
code hosting and technical Q&A
tax agencies, ministries, and legislatures across seven countries
news sites and magazines
course platforms and study resources
The languages of video learning
Language shares of the YouTube videos people summarize with Glasp, among summaries whose video language could be identified. English leads, but over 57% happens in other languages, led by Korean, Japanese, and Spanish. AI summarization is quietly a global study tool.
Methodology
- All figures are anonymized aggregates from Glasp platform data, snapshotted July 28, 2026. No individual-level data was used at any step.
- Book rankings: number of distinct readers who highlighted each book, computed separately per edition language and combining same-language Kindle editions of the same title (editions with at least 5 highlighting readers; ties broken by the largest single edition, remaining ties keep catalog order). A reader who highlighted two editions may be counted twice. The languages shown are those with enough distinct highlighted books: Top 20 for English and Japanese, Top 10 for Spanish and Portuguese. Author names added editorially; titles link to each book's community highlights page on Glasp, and Amazon links go to a verified product page on the edition's home marketplace or to an Amazon search for the title.
- AI summary categories: the most-summarized pages over the last 12 months were hand-classified into categories. Video platforms, entertainment domains, design canvases, and pages consistent with automatic (rather than user-initiated) summarizing were excluded from the percentage base. Shares are of the curated set, not of all activity, and an "Other" long tail (24.8%) is not shown.
- Video languages: share of YouTube summary events by video language over the last 12 months, normalized over events whose language labels could be reliably mapped (about 90% of the total).
- We publish rankings and shares, and deliberately do not publish absolute platform activity volumes, per-book reader counts, per-channel or per-video video rankings, or raw domain lists.
Privacy questions, answered
Is any individual user's data in this report?
No. Every number is an aggregate across many users. Book rankings count how many readers highlighted a book, never who. Category shares are computed over distinct users per category, and language shares over anonymized, aggregated usage. No highlight text, usernames, or individual reading histories appear anywhere.
What data does the report use?
Three aggregate sources: counts of readers per Kindle book, hand-curated category shares of the pages users summarize with AI, and the language distribution of YouTube video summaries. All three are aggregate, de-identified statistics; no individual accounts or content were examined.
Why publish this at all?
Reading behavior at scale is usually locked inside companies. Amazon stopped publishing Kindle popular highlights years ago, and most public statistics on reading and AI use come from surveys rather than behavior. Publishing aggregates, with the methodology in the open, is our attempt to fill that gap. Our research papers at glasp.co/research follow the same principle.
Go deeper
The behavioral research behind Glasp, including six arXiv papers on how people highlight, lives at glasp.co/research. For the statistics on whether highlighting even works, see Highlighting Statistics.