The Problem with Looking Things Up
You're 200 pages into a fantasy epic. A character appears and you think: "Wait, who is this again?" So you Google their name—and immediately see: "Character X dies in Chapter 47" in the search results.
Wikis, fandom sites, and even well-meaning forums are landmines of spoilers. That's why we built Recall Reader with a fundamentally different approach.
The Core Principle
Every feature in Recall Reader is bounded by your reading position. If you haven't read it yet, we can't reference it.
How It Works
On-device indexing
When you import a book, Recall Reader builds a local index on your device. We use full-text search, entity recognition, and semantic scoring to understand characters, locations, and narrative structure—all processed locally.
Position-bounded context
Every query is filtered through your current reading position. When you ask "Who is this character?", the system only considers text from the beginning up to where you are now. Everything after your position is excluded.
No external lookups
Recall Reader never searches the web, accesses Wikipedia, or queries book databases for your specific book. Answers come from the book itself—the text you've already read.
Transparent sourcing
Every response shows where it's coming from: "Based on Chapter 7" or "Based only on what you've read." You can verify that nothing is being pulled from content you haven't reached yet.
What can be referenced vs. what can't
Built for Privacy and Performance
Local-first processing
Heavy lifting happens on your device. Book indexing, text chunking, entity extraction, and relevance scoring are all performed locally. Your library stays yours.
Fast and offline-capable
Core reading features work completely offline. Smart features require a connection for response generation, but the context filtering is always done locally first.
Privacy by design
Your full EPUB files never leave your device. Smart features send only the excerpts needed to generate a response. Cloud Sync (Pro, optional) backs up your reading data with encryption. See our Privacy Policy for details.
Evolving capabilities
We're investing in on-device language models and advanced local processing. The goal: even smarter features with even less data leaving your device.
Common Questions
What if the answer is wrong?
Generated responses can occasionally be inaccurate—that's the nature of any automated system. However, because responses are bounded by text you've already read, any errors will reference content you can verify by flipping back. It cannot invent future plot points.
What about series where I've read earlier books?
By default, each book is treated independently. Pro users can enable cross-book memory for series—if you're reading book 3, "Who Is This?" can pull context from books 1 and 2 that you've already finished. It still respects your position in the current book, and never pulls from books ahead of where you are in the series.
How is this different from Googling?
Google indexes the entire internet, including wikis written by people who've finished the book. Recall Reader's index is built from your book alone, bounded by where you are in it. There's no path to spoiler content because that content simply isn't in the system.
Ready to read without fear of spoilers?
Get Recall Reader and experience context-aware reading that respects your journey.
Get it on Google Play