0%
VS
back to projects

ai learning companion

/the idea

learnflow ai is an intelligent browser extension and learning platform that transforms passive educational content consumption into an active, personalized, and ai-assisted learning experience. it bridges the gap between simply watching a tutorial or reading an article and actually retaining the information. by integrating directly into the user's workflow, it eliminates the context switching normally required to take notes or build study materials.

/features

  • real-time note generation: automatically synthesizes key points while you watch videos or read articles.
  • smart quizzes & flashcards: instantly generates spaced-repetition materials based on the consumed content.
  • contextual tutor: ask questions about the current page, and the ai will explain concepts with full awareness of what you are looking at.
  • cross-platform sync: access your knowledge base and study materials from anywhere through the web platform.

/implementation

the project consists of a browser extension interacting with a robust backend architecture. the frontend is injected into the user's browser via content scripts, capturing context from videos (like youtube transcripts) and articles (parsing dom content). the backend leverages advanced llms and contextual ai to process this data in real-time. it utilizes vector databases for semantic search and retrieval-augmented generation (RAG) to ensure the ai tutor has perfect recall of the user's learning history.

/the tech stack

frontend

  • framework: react & browser extensions api
  • language: typescript
  • styling: tailwind css

backend

  • framework: python (fastapi)
  • ai integration: llms, vector dbs & contextual ai

/thoughts

building this platform reinforced the idea that learning should be interactive and frictionless. by bringing the ai directly into the browser where the content is consumed, the barrier to study and retain information drops significantly. the hardest technical challenge was ensuring low-latency responses for the contextual tutor while handling large transcripts and complex webpage structures, which was a great lesson in optimizing data pipelines and state management.