Using LLMs as Recommendation Engines: Finding Books by Character Traits

devanshmehta · x · 2026-09-29

The author shares a novel LLM use case: after reading The Goldfinch, they asked an LLM to grasp the protagonist's "self-destructive dragon" and recommend another book whose protagonist has an "impatience dragon" instead. This kind of semantic matching on abstract character traits is impossible for traditional recommenders, highlighting LLMs as open-dimension recommendation engines.

Original post →

More from Apps

Apps channel →