← Back to EpubGraph
Software engineer and avid reader with a library of 70,000+ ebooks. EpubGraph was born from a personal frustration: having too many books and no good way to discover what to read next. Traditional library managers organize books but don't understand them. I wanted a system that could read between the lines — one that maps the hidden semantic connections between books using the same AI technology that powers modern language understanding.
The result is EpubGraph: a desktop application that transforms a flat ebook collection into an intelligent, navigable knowledge graph. Every book gets a neural embedding that captures its essence — themes, genre, tone, style — and the system uses these embeddings to surface recommendations you'd never find through keyword search alone.
The Vision
"Life is short. Your library is vast. The next great book you'll love is already on your shelf — you just haven't found it yet. EpubGraph is the map."
EpubGraph sits at the intersection of three ideas: that personal libraries deserve the same recommendation intelligence as streaming platforms; that AI should run locally, respecting your privacy and your data; and that the relationships between books are as interesting as the books themselves.
Built With
Open Source
EpubGraph is released under the MIT License. The entire source code — Rust backend, Svelte frontend, AI pipeline, and graph algorithms — is open for inspection, modification, and contribution. No telemetry, no tracking, no cloud dependencies. Your books, your data, your machine.