Gandzel open sources AI player to turn classic novels into immersive cinema

The fallacy of the perfect refactor

In software development, the urge to clean up messy code is often a trap. Gandzel advocates for ruthless prioritization when iterating with AI. He suggests a specific formula for refactoring: the product of how bad the code is, how painful it is to maintain, and how easy it is to fix. If a piece of OpenAI audio processing code is ugly but functional and rarely touched, let it stay ugly. Throwing away AI-generated code is emotionally easier than discarding manual work, allowing for faster experimental cycles in the early stages of a project.

Natural language and semantic search

Standard keyword searches are useless for the way human brains actually recall stories. We remember themes and interactions, not specific page numbers. By using embeddings, Books Reimagined allows readers to perform semantic searches, such as finding every instance where two characters met or researching party propaganda in 1984. This deep research capability reads the entire book up to the user's current progress, ensuring the AI never provides a spoiler for events yet to come.

Automation with a human touch

Pure AI generation often lacks soul. Gandzel’s workflow uses LLMs for the heavy lifting—scene analysis, character detection, and mood mapping—but keeps a human in the loop for the final curation. This hybrid model ensures that the Victorian London vibe of Sherlock Holmes isn't ruined by a mismatched musical score or a generic character avatar. By open-sourcing the player, Gandzel is giving developers the tools to create these "Netflix-style" literary experiences for any title in the public domain.