kiddo.ai 🤖📚

Kiddo.ai transforms your exam preparation by challenging you to become a teacher. The core of the experience revolves around explaining complex concepts to "Kiddo," your virtual study companion, as if he were a curious child. This process, rooted in the Feynman technique, forces you to simplify and deeply understand the material, truly cementing it in your mind.

But here's the catch: if you don't regularly explain things to Kiddo or help him review, he starts to "forget" what you've taught him! This unique dynamic is designed to motivate you, leveraging principles like Loss Aversion. Seeing Kiddo's understanding fade should spur you on to consistently engage, teach, and reinforce – ensuring both you and Kiddo master the concepts together. The system also integrates Spaced Repetition and Active Recall to further enhance learning and retention.

✨ Key Features

• Interactive Learning: Explain concepts to Kiddo, a curious avatar that wants to learn.
• Effective Techniques: Implements Feynman and "Teach Others" methods.
• Smart Study: Utilizes Spaced Repetition and Active Recall.
• Motivation: Includes gamification elements to maintain high engagement.
• Source-Based: Users provide study material (PDF) as a reference.

🔧 How it Works

The system is an API-based backend with two main pipelines:

1. Ingestion Pipeline:

   • The user creates a new "exam" by providing the date, study schedule, and the reference PDF.
   • The PDF is chunked and stored in a vector database.
   • An agent identifies macro-areas and key concepts.

2. Kiddo Pipeline (Learning Interaction):

   • The user selects a macro-area.
   • Kiddo initiates a question-driven dialogue for that set of concepts.
   • Concept Validation: The user's explanation is compared with the original content.
   • Knowledge Graph Update: If the explanation is accurate, the concept is integrated into Kiddo's internal knowledge graph. The graph nodes represent concepts and their learning status:
     • 🟢 Green: well-learned and stable.
     • 🟡 Yellow: correct but needs review (initial state).
     • 🔴 Red: incorrect concept.
     • ⚪ Gray: forgotten due to lack of review.

🚀 Setup

The application is dockerized. Ensure you have Docker installed.

1. Create two .env files following the template provided in .env.template.
2. Run the application with:

   docker compose up