How Tangent works

Tangent is a cross-media recommendation engine. It connects books, films, TV, and podcasts by taste — so a book you loved can lead you to a film, a series, or a podcast that shares its themes. Most tools recommend within one medium; Tangent recommends across all four.

How it works

One taste space across every medium

Every title is embedded in a single 768-dimensional vector space (Google Gemini embeddings), so the similarity between a book and a podcast is directly computable — not just book-to-book or film-to-film.

A canonical-works graph

Tangent maintains a graph that links a work to its versions across media — a novel to its film, series, and audio adaptations — so "where should I start?" has a real, sourced answer instead of a guess.

Thematic taxonomy, not popularity

Every title is classified into a multi-level genre taxonomy and enriched with themes ("Southern Gothic," "moral courage," "unreliable narrator"), so matches are about substance, not what happens to be trending.

By the numbers

Tangent's catalog covers 29,162 titles: 11,423 books, 9,912 films, 4,983 TV series, and 2,844 podcasts. Every title is embedded for cross-media similarity (100% coverage) and 99% carry full descriptions. The canonical-works graph links 2,517 works to 4,237 versions and adaptations across the four media.

Where the data comes from

Film and TV metadata comes from TMDB, books from Google Books, and podcasts from Taddy. Tangent layers its own cross-media identity, taste embeddings, and thematic enrichment on top.

Common questions

How is Tangent different from Goodreads, Letterboxd, or IMDb?

Those each track one medium — books, films, or screen. Tangent connects all four: books, films, TV, and podcasts. It is built to recommend the next thing you will love in a different medium, not just more of the same one.

How does Tangent make recommendations?

It places every title in a shared taste space and finds the nearest matches across media, then re-ranks them by thematic overlap and acclaim. Recommendations are grounded in a real catalog and a canonical-works graph, so they do not invent titles or adaptations.

Does Tangent use AI?

Yes — for taste embeddings and for enriching each title with themes and genres. Recommendations are always grounded in the real catalog rather than generated freely, which keeps them accurate and free of fabricated titles.

Why cross-media?

Taste does not stop at one format. If you loved a literary thriller, the best next thing might be a film, a limited series, or a narrative podcast. Single-medium tools cannot make that jump; Tangent is built for it.

Is Tangent free?

Yes. You can get cross-media recommendations for free, with no account required to try it.

Explore

See it on your own taste — across every medium.

Try Tangent

Free, 30 seconds, no account.