BookSim

Inspiration

We’ve all been there: you finish a brilliant non-fiction book like Atomic Habits or Rich Dad Poor Dad, feeling motivated and ready to change your life. But two weeks later? You’ve forgotten 90% of it. The problem isn't the content; it's the medium. Reading is passive. Learning, however, requires action.

This happends to me all the time. So I asked myself: What if a book wasn't just something you read, but a world you could enter? What if you could test your understanding of The Art of War by commanding an army, or practice Never Split the Difference by negotiating a high-stakes hostage situation?

That was the spark for BookSim.

What it does

BookSim is an interactive roleplay engine that instantly converts any non-fiction book into a text-based adventure game.

Instead of summarizing a book, it simulates it:

1. The Librarian: You enter a title (e.g., Sapiens). The app verifies it's a valid non-fiction work (filtering out fiction/fantasy).

2. The Architect: It analyzes the book’s core philosophy to generate a unique game framework. If you play a business book, your health bars might be "Cashflow" and "Assets." If you play a stoic philosophy book, they might be "Willpower" and "Clarity."

3. The Simulation: You are dropped into a roleplay scenario. You make choices, and the AI Game Master reacts based strictly on the book's principles.

4. Gamified Learning:

• Knowledge Cards: When you successfully apply a concept, you "unlock" it as a collectible card (Common/Rare/Legendary).

• Notebook: You can save your insights by leaving your personal note in the digital notebook.

• Final Report: At the end of a session, you receive a graded assessment (S/A/B/C) analyzing your decision-making style against the author's advice.

How we built it

BookSim's prototype is built within Google AI Stuido, the "brain" is powered entirely by the Google Gemini API.

• The Logic Engine (gemini-3-flash-preview): I utilized the reasoning capabilities of the Flash model to act as the Game Master. It handles the complex state management of the game—tracking inventory, updating stats based on user choices, and ensuring the narrative stays true to the book's specific worldview.

• The Visual Engine (gemini-2.5-flash-image): To make the experience immersive without infringing on copyright, we use the Nano Banana model to generate abstract, geometric cover art that captures the vibe of the book without reproducing the official cover.

• Guardrails & Safety: I implemented a "Transformative Use" layer. The AI is instructed strictly not to output verbatim text from the book. Instead, it synthesizes the methodologies into original scenarios, ensuring the app is a companion tool, not a replacement for buying the book.

Challenges we ran into

The process was smooth and fun. The only challenge was that I couldn't stop testing it—I simulated 40+ books that I’d never read before in a couple of days.

Accomplishments that we're proud of

• The "Universal Game Engine": I'm proud that BookSim works on any non-fiction book. You can type in an obscure gardening book or a best-selling biography, and the system intelligently extracts the core "winning conditions" of that specific text.

• The "Knowledge Card" System: We successfully gamified the "Aha!" moment. The dopamine hit of unlocking a "Legendary" concept card makes the learning process addictive.

• Performance: The speed of Gemini 3 Flash allows the game to feel real-time. There is almost no friction between making a choice and seeing the consequence.

What we learned

The Difference Between "Textual Fidelity" and "Philosophical Fidelity": Usually, developers try to prevent AI hallucination. In BookSim, I realized that to simulate a book, the AI shouldn't quote the text (Copyright infringement); it needs to "hallucinate" new scenarios that strictly adhere to the author's logic (Transformative Use).

What's next for BookSim

• Scale the prototype into a cross-platform application (Web, iOS, and Android).
• Partner with bookstores and libraries to explore integrating the app into their daily operations.

Built With

• googleaistudio