Hosted BookSeerr: managed semantic book discovery for readers whose Calibre library has outgrown the alphabetical scroll.

Early Access

This app is a brand-new addition to ElfHosted, and is considered beta. It runs, we test it, and your trial lets you prove it fits before you commit, but our support team does not yet have deep experience with it. Expect some trial and error: lean on the upstream documentation, experiment freely, and share what you learn in Discord, where the community and team are figuring these new apps out together. If you need a battle-tested workflow today, our established apps are the safer pick.

What is BookSeerr?

BookSeerr sits on top of your Calibre library and turns it into a discovery engine. Instead of scrolling an alphabetical list, you browse by meaning: open any book and BookSeerr surfaces similar titles and reading suggestions drawn from your own shelves. It reads the library managed by your Calibre Web instance, builds semantic embeddings on-box (no external AI service required), and serves a fast browsing UI for searching by title, author, or tag and jumping between related books.

Key Features

• Semantic similarity: Open a book and instantly see titles close in theme and content, not just matching tags.
• On-box embeddings: Recommendations run locally on CPU, so discovery works out of the box with no API keys or per-query costs.
• Reads your Calibre library: Points at your existing Calibre Web library; every book you already manage becomes searchable. It only ever reads, never writes.
• Covers and metadata: A clean grid of cover art with titles, authors, and tags pulled straight from your library.
• Fast search: Filter the catalog by title, author, or tag as you type.
• Optional LLM suggestions: Bring your own OpenAI-compatible endpoint to layer richer, generated reading suggestions on top of the built-in similarity engine.

BookSeerr in Your Stack

BookSeerr requires a Calibre library to shine: enable Calibre Web in your ElfHosted subscription, add your books, and BookSeerr discovers them automatically. It complements a full library manager like Grimmory too, answering the question those apps do not: "I loved this, what on my shelves is like it?"

Why Run BookSeerr on ElfHosted?

Embedding models and library indexing want steady compute and a permanent home next to your Calibre library. ElfHosted supplies both, on the same internal network as the rest of your book stack.

• Managed access and support: Provisioned in minutes, kept updated, with humans on Discord when you need them.
• Plays well with other ElfHosted apps: Reads your Calibre Web library directly, no path or mount configuration required.
• Persistent config: Embeddings and settings survive restarts and upgrades, no re-indexing marathon after maintenance.
• Simple subscription: One flat monthly price with a 7-day trial, cancel anytime.

Works With

• Calibre Web: The library BookSeerr reads; effectively a prerequisite.
• Grimmory / BookLore: Full-featured library managers that pair well with BookSeerr's discovery layer.
• Your own LLM endpoint: Optional, for generated reading suggestions beyond the built-in similarity.

Frequently Asked Questions

Does BookSeerr include any books?
No. BookSeerr discovers within the library you already own and manage in Calibre Web. No library, nothing to discover.

Do I need an AI API key?
No. The core similarity engine runs entirely on your instance using local CPU embeddings. An OpenAI-compatible key is optional, only for the richer generated-suggestions layer.

Can BookSeerr damage my Calibre library?
No. It mounts your library read-only in spirit and in practice: it never writes to it, it only indexes and displays it.

How mature is BookSeerr?
It is a small, young upstream project, offered as an early-access app. The core browse-and-discover flow works well; expect the polish to keep improving.