Turning Static Knowledge into Interactive Learning Ecosystems.
LEO LIVE BOOK is an AI-native engine designed to transform any book—academic, professional, or personal—into a structured, multi-dimensional learning experience. By leveraging Python, LLM AI Agents (Google Gemini), and high-performance vector databases, LEO LIVE BOOK automates the entire pedagogical pipeline: from raw text extraction to the generation of lecture slides, mindmaps, and auto-graded assessments.
The project is part of the broader LEO CDP Framework ecosystem, focusing on the intelligence layer of knowledge management.
https://leocdp.com/docs/index.html
LEO LIVE BOOK goes beyond simple summarization; it acts as a Virtual Professor and Content Architect:
-
Universal Book Ingestion: Parse PDFs, EPUBs, and text files into structured data.
-
AI Agent Intelligence: Automated extraction of core concepts, keywords, and semantic summaries.
-
Dynamic Lecture Material: Automatically generate structured lecture slides and presentation outlines from book chapters.
-
Visual Knowledge Maps: Generate Mermaid.js or Graphviz code to visualize book structures as interactive mindmaps.
-
Active Pedagogy: * Automatic generation of exercises (Multiple Choice, Q&A, Case Studies).
-
Auto-Grading Engine: Real-time evaluation of student answers with personalized feedback.
-
Life-Long Learning: A persistent knowledge graph that tracks student progress and adapts content based on mastery.
LEO LIVE BOOK combines a robust data architecture with cutting-edge AI:
- Extraction Layer (Python + LLM): Uses
PyPDF2andgoogle-genaito batch-process book content, maintaining context across thousands of pages. - Memory Layer (PostgreSQL + pgvector): Stores book sections as vector embeddings to support Retrieval-Augmented Generation (RAG) and semantic search.
- Reasoning Layer (AI Agents): Specialized agents for different tasks:
- The Analyst: Extracts keywords and summaries.
- The Designer: Creates slide decks and mindmaps.
- The Examiner: Crafts exercises and grades submissions.
- Ingest: The Python engine parses the PDF and creates a structured map of Chapters, Sections, and Exercises.
- Index: Content is vectorized and stored in PostgreSQL 16 for instant semantic retrieval.
- Generate: * Slides: Creates Markdown/PPTX-ready outlines for instructors.
- Visuals: Generates hierarchical mindmaps of the book's logic.
- Drills: Produces practice questions based on specific sections.
- Assess: Students submit answers; the AI Agent compares them against the book's "source of truth," provides a grade, and explains why the answer was right or wrong.
The system organizes data into four primary layers to support life-long learning:
- Source Material: Structured text, page references, and book metadata.
- Knowledge Assets: Keywords, summaries, slides, and mindmap nodes.
- Assessment Bank: Generated exercises, correct answers, and grading rubrics.
- Student Ledger: History of attempts, grades, and mastery levels for specific topics.
- Source Fidelity: Every AI-generated summary or slide is traceable back to a specific page and paragraph in the original book.
- Agentic Autonomy: AI Agents handle the heavy lifting of content creation, allowing human learners to focus on synthesis and application.
- Multi-Modal Learning: Information is presented via text (summaries), visuals (mindmaps), and activity (exercises).
- Framework Agnostic: While optimized for Gemini, the core logic is model-agnostic and schema-first.
- Corporate Training: Turn technical manuals into interactive certification courses.
- Higher Education: Automate the creation of study guides and lecture slides for textbooks.
- Self-Directed Learning: Upload any non-fiction book to receive a personalized curriculum and "final exam."
- Content Publishing: Help authors provide "Active Learning" versions of their books.
Most AI tools just summarize text. LEO LIVE BOOK builds a bridge between reading and mastery. It ensures that the knowledge contained within a book isn't just stored in a database, but is actively taught, tested, and visualized.
Don't just read the book. Live the book.
*For development and contribution, visit the core repository: leo-live-book*