OctoStudy

OctoStudy — Learn Smarter, One Micro-Lesson at a Time.

Comment

Inspiration

OctoStudy started from a simple observation: rereading notes feels productive but rarely leads to long-term retention. I wanted a low-friction tool that turns passive notes into active learning — short, teacher-like micro-lessons and quick quizzes — and couples that with simple study mechanics (Pomodoro + spaced repetition) so learners actually practice recall.
The octopus metaphor fits naturally: many small tentacles (micro-lessons) working together to improve retention.

What it does

  • Converts raw study material (pasted text, PDFs, or web pages) into AI-generated micro-lessons.
  • Automatically creates active-recall quizzes (2 MCQs + 1 short answer) for each lesson.
  • Schedules reviews using a spaced repetition system (SRS) to optimize memory retention.
  • Runs focused study sessions with an integrated Pomodoro timer.
  • Tracks progress with simple, meaningful metrics (cards due, accuracy, streak).

How we built it

  • Frontend: Next.js (App Router) + TypeScript + Tailwind for a clean, responsive UI.
  • Backend: Next.js API routes handling content ingestion, LLM calls, and SRS updates.
  • AI layer: Openrouter-compatible LLM with carefully designed prompts that return strictly structured JSON.
  • Database: Postgres (Supabase) for production, with a lightweight SQLite/JSON fallback for local development.
  • Learning logic: A simplified SM-2 spaced repetition algorithm implemented in JavaScript and covered with unit tests.
  • Deployment: Vercel for hosting, Supabase for database.

Challenges we ran into

  • Ensuring LLM output consistency; early responses often broke JSON structure, requiring prompt refinement and validation retries.
  • Handling PDF and web content extraction, which varies widely in formatting and quality.
  • Designing an SRS system that is simple, explainable, and stable under repeated reviews.
  • Balancing scope and polish as a solo developer — choosing a flawless core workflow over extra features.

Accomplishments that we're proud of

  • A complete, demo-ready flow: import → generate → study → schedule reviews.
  • Reliable LLM prompts that consistently generate structured lessons and quizzes.
  • A fully implemented and tested spaced repetition engine.
  • A smooth, responsive study session UI with real-time feedback.
  • A development mode that allows testing without excessive API usage.

What we learned

  • Prompt engineering is as critical as traditional coding in AI-first applications.
  • Active recall and spaced repetition dramatically outperform passive review.
  • Smaller, well-polished features impress more than many unfinished ones.
  • Validating and testing AI outputs is essential for reliability.

What's next for OctoStudy

  • Add user accounts and syncing with Supabase Auth.
  • Improve lesson relevance using vector-based semantic search.
  • Enhance the SRS with response-time awareness.
  • Release a mobile-friendly PWA for offline study.
  • Add deeper analytics to visualize retention over time.

Built With

Share this project:

Updates