Topic Discovery
Screvi's AI groups your highlights into themes that cut across books and sources, then lets you review or search each theme on its own.
Updated
What It Does
Topic Discovery reads your whole library and finds clusters of highlights that share a theme, even when they come from different books, articles, and podcasts. You might learn that your notes on psychology, philosophy, and business all circle "deciding under uncertainty".
It needs no tags. The grouping comes from the meaning of the highlights themselves.
Where to Find It
- Web: open Review. Under Custom review, the Suggested topics row lists the themes Screvi found. Pick one to review just those highlights.
- Mobile: the Search tab shows discovered topics. Tap one to search within it.
What You See
Screvi proposes between 5 and 20 topics depending on how large and varied your library is. Each topic has an AI-written label, a count of highlights, and a few representative passages. Open a topic to read everything in it.
Topics are cached so they load instantly. Refresh them after a big import to pick up new themes.
Ways to Use It
- Review by theme: a focused session on one topic beats a random mix when you are working on that subject.
- Find the through-line: see which ideas you keep returning to across years of reading.
- Spot gaps: a thin topic tells you where you have read little.
- Seed your tags: when a topic is useful, create a tag for it and let Suggest tags apply it.
How It Works
Every highlight already has an embedding for semantic search. Topic Discovery clusters those embeddings by similarity and asks an AI model to name each cluster from its highlights.
Related
Still have questions?