BookSoul

Thanks to Kiro, I developed three branches of the BookSoul product

1. Stable version The most proven and stable version. The version here will be developed and played. URL: https://booksoul.me/ Repo: https://github.com/KFranciszek/BookSoul/tree/main

2. Development version Contains an advanced multi-LLM engine, may contain bugs. The version that can be considered KIRO PROD URL: https://booksoul-1.onrender.com/ Repo: https://github.com/KFranciszek/BookSoul/tree/multi-ai-recommendation

3. Test version - unstable. The version that can be considered KIRO DEV URL: https://booksoul-dev2.onrender.com A version containing the entire user module - login, learning, and individual recommendations for logged-in users. This version is currently in development, contains bugs, and is unstable.

Repo: https://github.com/KFranciszek/BookSoul/tree/user_module_v2 (KIRO FOLDER HERE!!!)

Visit my social media Instagram:https://www.instagram.com/booksoul_me/ TikTok:https://www.tiktok.com/@booksoul_me X: https://x.com/BookSoul_me

The inspiration for BookSoul came from a deeply personal frustration: the endless scrolling through generic book recommendation lists that felt completely disconnected from my actual emotional state and reading psychology. Traditional recommendation systems rely on cold algorithms that match genres or popularity metrics, but they miss the most crucial element – the human soul behind the reading desire. I realized that when we choose a book, we're not just selecting entertainment; we're seeking a psychological companion for our current life chapter. Sometimes we need escapism during stressful periods, other times we crave intellectual challenge when we're mentally energized, and occasionally we want emotional healing through carefully chosen narratives. This revelation sparked the idea: What if AI could understand not just what we like, but who we are in this moment?

BookSoul operates on a fascinating psychological principle I discovered during development: reading preferences are dynamic emotional fingerprints. The system implements a Multi-Agent AI Psychology Engine that mirrors how a human psychologist would analyze a patient

I have designed four different recommendation paths, each tailored to different user attitudes and psychological needs:

1. Quick Match For busy readers who want an instant psychological match based on their current mood, stress level, and reading goals. Perfect for finding your next read during your lunch break.
2. CineMatch A revolutionary translation of movies into books using narrative psychology. Users enter their favorite movies, and artificial intelligence analyzes cinematic narrative patterns, visual preferences, and emotional pacing to find books that capture the same psychological experiences as their beloved movies.
3. BookInspiration A “More Like This” mode for true book lovers. Users describe their favorite books and explain what they liked about them. Artificial intelligence conducts an in-depth literary analysis to understand specific narrative elements, writing styles, thematic preferences, and emotional connections, and then searches for books with similar psychological DNA.
4. Deep analysis Comprehensive psychological profiling to discover life-changing books. This mode examines tolerance for complexity, reading frequency, educational interests, motivational needs, and creates a complete psychological profile of the reader to match books that can change their life.

Built With

• kiro
• llm
• openai
• psychology
• react
• supabase