Summary

"Get It." is a desktop study companion that transforms PDFs into a "measurable mastery map," moving beyond simple summaries to foster deep comprehension. It targets students and researchers needing to prove understanding beyond rote recall, offering a unique approach to assess and improve knowledge retention for unseen questions.

How It Works

This application processes text-based PDFs, identifying key concepts and automatically generating visualizations (3D, 2D animations, formulas, graphs) via integrated agents. A knowledge graph maps concept relationships, while four study tools (Chat, Flashcards, Quizzes, Feynman) feed a journal. An evaluator agent continuously updates concept mastery scores (memory, comprehension, structure, application) on a 0-100 scale, ensuring progress. Crucially, "Get It." leverages user-provided OpenAI accounts (ChatGPT Plus recommended) via the Codex CLI, meaning no separate AI subscription or data sharing with the project. All interaction data remains local.

Quick Start & Requirements

Install via native desktop installers (DMG, EXE, AppImage) available on the project's Releases page. Requires an existing OpenAI account (ChatGPT Plus or API key recommended for sufficient model usage headroom; free tier is limited). macOS builds are notarized; Windows builds may trigger a SmartScreen warning on first launch due to lack of code signing. A setup wizard guides users through Codex CLI verification and OAuth sign-in.

Highlighted Details

• BYO-AI Model: Integrates with user's existing ChatGPT/OpenAI API subscription via Codex CLI, avoiding project-specific AI costs or data privacy concerns.
• Local-First Data: All user data, including PDFs, generated graphs, and study journals, is stored exclusively on the user's machine.
• Dynamic Concept Visualization: Concepts are tagged for rendering as interactive 3D scenes, 2D animations, formula walkthroughs, or data plots.
• Monotonic Progress Tracking: Study metrics are designed to be non-decreasing, reinforcing learning and preventing regression.
• Self-Healing Agents: AI agents monitor and automatically repair rendering sandbox crashes.

Maintenance & Community

Developed by Mattia Beltrami, Matteo Impieri, Filippo Difronzo, and Luca Feggi, originating from the GDG AI Hack Milan 2026. Post-hackathon polish has been applied. No specific community channels (Discord, Slack) or roadmap links are provided in the README.

Licensing & Compatibility

Licensed under the Apache License 2.0. Source code is open, and contributions are welcomed. The license permits commercial use and integration with closed-source projects.

Limitations & Caveats

The application only processes digital, text-based PDFs, rejecting scanned or image-only documents. A 150-page document limit is implied for affordability on standard ChatGPT plans. Windows builds lack code signing, necessitating user override of SmartScreen warnings. Free OpenAI tiers offer limited Codex usage, making paid subscriptions practically necessary for extended sessions.