Synapse

Synapse is an AI-powered learning platform that transforms user-uploaded PDFs or topics into structured, video-enriched, quiz-powered learning experiences with spaced repetition scheduling.

The platform extracts key concepts, matches relevant YouTube videos, generates quizzes, and provides revision tools.

an intelligent learning platform that transforms any PDF document or typed topic into a structured video course with AI-generated quizzes, flashcards, and spaced repetition review. The platform bridges the gap between passive reading and active recall by automatically extracting key concepts, matching educational video content, and generating personalized study instruments.

Core Value Proposition: Upload a textbook chapter, lecture notes, or research paper and receive a complete video course with MCQ quizzes, viva-style questions, mnemonic images, and flashcards optimized by the FSRS spaced repetition algorithm - all within 60 seconds.

Tech Stack Layer Technology Frontend React 18 + Vite + TypeScript Styling Tailwind CSS + shadcn/ui Backend Supabase (PostgreSQL + Auth + Storage + Edge Functions) State & Routing React Router DOM + Zustand-ready architecture Build Tool Vite Linting Biome MeDo Skills & Plugins Utilized The following MeDo platform skills were integrated into the application:

1. OpenMAIC Classroom Purpose: Multi-agent interactive learning environment Integration: Embedded AI teacher and AI classmates for immersive classroom discussions, slide generation, interactive quizzes, and whiteboard simulations Endpoint: Edge function proxy to OpenMAIC base URL with access code authentication
2. Large Language Model (Gemini 2.5 Flash) Purpose: Core AI reasoning engine Integration: Supabase Edge Function extract-concepts uses the LLM gateway to: Parse uploaded PDF content and extract key concepts Generate 5 multiple-choice questions + 2 viva questions per lesson Create structured course curricula with lesson hierarchies Remove em dashes and format text for clean UI rendering
3. Image Generation Purpose: Visual learning aids Integration: Supabase Edge Function generate-mnemonic uses the Image Generation API to: Generate mnemonic images from concept descriptions Upload base64-generated images to Supabase Storage (lesson-assets bucket) Attach image URLs to lesson records for memory reinforcement
4. Text-to-Speech (LemonFox) Purpose: Audio narration of lesson content Integration: Supabase Edge Function generate-narration uses the LemonFox TTS API to: Accept optional custom text input or auto-extract lesson content Truncate text to 4000 characters for API limits Generate MP3 audio using the heart voice profile Upload audio files to Supabase Storage and link to lessons
5. Speech-to-Text (Whisper v3) Purpose: Audio transcription for accessibility Integration: Available via Edge Function for converting spoken content or uploaded audio into text transcripts (supports 100+ languages and speaker diarization)
6. Web Search Purpose: Research and data validation Integration: Used during the research phase to gather competitive intelligence, verify educational video landscape, and inform the product roadmap
7. AI Search Purpose: Intelligent web-aware Q&A Integration: Available for real-time information retrieval and fact-checking within the platform
8. Login (Supabase Auth) Purpose: User authentication and identity management Integration: Complete auth system with username/email registration, password reset, email confirmation, and role-based access (student / admin roles) Security: Row Level Security (RLS) policies on all database tables with 16 active policies
9. The News API Purpose: Real-time news integration (available, reserved for future feature expansion) Architecture: How Plugins Extend Functionality PDF-to-Course Pipeline (Edge Functions) The core learning engine is powered by 9 Supabase Edge Functions that chain together platform skills:

Edge Function Skill Used What It Does extract-concepts LLM (Gemini 2.5 Flash) Parses PDF, extracts concepts, generates 5 MCQ + 2 viva questions per lesson generate-mnemonic Image Generation Creates visual memory aids, uploads to Storage, updates lesson record generate-narration Text-to-Speech (LemonFox) Creates audio MP3 of lesson text, uploads to Storage, links to lesson match-videos Web Search / YouTube Matches extracted concepts to relevant educational videos create-flashcards LLM Generates FSRS-optimized flashcards from lesson content schedule-reviews FSRS Algorithm Calculates optimal review intervals for spaced repetition generate-flowchart Image Generation Creates visual concept maps and flowcharts transcribe-audio Speech-to-Text (Whisper) Converts uploaded audio to text for course creation ai-classroom OpenMAIC Classroom Launches multi-agent interactive classroom sessions Data Flow User uploads PDF or types a topic extract-concepts (LLM) processes the content and structures a curriculum match-videos finds relevant educational videos for each concept generate-mnemonic (Image Generation) creates visual memory aids generate-narration (TTS) produces audio narration create-flashcards (LLM) generates review cards with FSRS scheduling metadata User studies through the lesson page (video + text + quiz) schedule-reviews calculates the next review date using the FSRS algorithm Dashboard shows due reviews, study streak, and progress stats Key Features PDF Upload & Topic Input: Two modes for course creation - upload any document or type a topic Auto-Curriculum Generation: AI structures the content into logical lessons with concept hierarchies Video Matching: Each concept is paired with a relevant educational video Dual Quiz System: 5 multiple-choice questions + 2 viva (oral exam) questions per lesson AI Mnemonic Images: Visual memory aids generated for difficult concepts Audio Narration: Full TTS narration of lesson content for auditory learners FSRS Spaced Repetition: Algorithm-powered flashcard scheduling for optimal retention Study Planner: Calendar view of upcoming reviews and study sessions Progress Dashboard: Stats cards showing concepts learned, quiz accuracy, day streak, and study hours Admin Panel: Course and user management for administrators OpenMAIC Classroom: Multi-agent interactive classroom with AI teacher and classmates Database Schema 8 core tables managed by Supabase PostgreSQL:

courses - Course metadata and configuration lessons - Individual lesson content and video references concepts - Extracted key concepts with linked videos flashcards - FSRS-scheduled flashcards with difficulty and interval tracking quiz_questions - MCQ and viva questions per lesson user_progress - Per-user lesson completion and quiz scores user_settings - User preferences and FSRS parameters mnemonic_images - Generated visual aids linked to concepts Design System Primary Background: Deep navy #0a0f1e Accent Color: Warm amber/gold #f59e0b Card Style: Solid background with 1px border (cc-card utility class) No Glassmorphism: Explicitly removed all backdrop-blur, gradient text, and glow effects Typography: Clean, tight tracking, text-balance for readable headings Aesthetic: Linear.app / Raycast-inspired minimalism with zero decorative animations Deployment Frontend: Static build via Vite Backend: Supabase (PostgreSQL + Edge Functions + Auth + Storage) Edge Functions: 9 functions deployed to Supabase, all TypeScript/Deno Storage Buckets: lesson-assets for generated images and audio files RLS Policies: 16 active security policies protecting all user data Built with the MeDo platform using React + Supabase + integrated AI skills.

Built With

• medo-plugin