Book.sv

What it does

Book.sv is a recommendation engine that suggests books based on titles a user has already read. The system uses a machine learning model trained on over 3 billion Goodreads reviews. Users input at least three books they have read, and the engine returns personalized recommendations. The site requires books to meet a popularity threshold before they appear in results, though less popular titles can be used in other features.

The platform also offers a secondary tool called "intersect" that identifies Goodreads users who have read a specific set of books. Users can search by inputting book titles directly or import their entire Goodreads shelf using their user ID.

Who it is for

This tool targets voracious readers seeking their next book. It appeals to people already active on Goodreads who want algorithmic suggestions beyond basic genre or popularity filters. The intersect feature may interest book clubs or reading groups looking to find like-minded readers who share their exact reading history.

Pricing

The site does not show prices.

How it stands out

The scale of training data—3 billion Goodreads reviews—is substantial for a cold-start recommendation problem. Most book recommendation engines rely on collaborative filtering or basic metadata. Training on review text itself could capture subtle taste signals beyond simple "liked" or "rated" signals. The tool also offers an unusual secondary feature in intersect, which turns the recommendation problem sideways into a community-finding tool.

The ability to import directly from Goodreads shelves reduces friction for users who already maintain reading lists there.

What a founder should check

First, examine the incumbent landscape. Goodreads itself has recommendation algorithms, as do Amazon, StoryGraph, and Shepherd. Understand what these established players do well and where Book.sv claims differentiation beyond raw training data size.

Second, investigate data freshness and model quality. The homepage notes an update date of November 9, 2025, but it is unclear how often the recommendation model retrains or whether new Goodreads reviews are incorporated. Test whether recommendations degrade for niche or recently published books, and whether the popularity threshold meaningfully limits utility.

Third, consider the switching cost moat. Users can export recommendations, but they have no lock-in beyond convenience. Goodreads owns the shelf data and user relationships. Assess whether the quality advantage needs to be compelling enough to pull users away from free, integrated recommendations they already use.