AI-native spoiler-free reading companion
Readers stop mid-book all the time due to life interruptions. When they pick it back up days or weeks later, they've forgotten the characters, plot threads, and details about where they left off. The existing options: reread chapters (tedious), Google a summary (spoilers), or give up on the book entirely.
Roughly 27% of adults say they haven't finished a book in the past year. Many of those aren't people who don't want to read. They're people who lost their place and couldn't get back in.
Recapty lets you tell the app where you stopped in any book (by chapter, page number, percentage, or a camera photo of your page), choose a recap depth, and get a personalized summary of everything you've read so far. Nothing past your stopping point. No spoilers.
Three recap depths serve different needs: Snippet (quick refresher), Compass (moderate detail with character tracking), and Chronicle (full recap with plot threads, tensions, key figures, context notes, and reader questions to carry forward).
The "Learn More" feature lets readers ask questions about anything they've read so far. The AI answers with chapter-specific citations, staying within the reader's progress boundary.
Ideation, product strategy, front-end design and build (using Claude and Cursor AI), and the overall product experience. The app uses Claude for content generation logic, which my partner and I co-designed: the prompting architecture that generates spoiler-aware recaps at different depth levels while respecting the user's exact stopping point.
Multiple input methods for marking your place were critical. Readers use different formats (Kindle shows percentages, physical books have page numbers, audiobooks track by chapter). Supporting all of them means Recapty works regardless of how you're reading.
The three depth levels came from thinking about different use cases. Sometimes you just need a two-sentence reminder of where you are. Sometimes you need a full character refresher because you've been away for a month. One size doesn't fit.
The "Request Book" feature lets users submit titles the library doesn't yet support, which doubles as demand signal for content prioritization.