Memoir: A Second brain

In today’s fast-paced world, we consume endless information—but how much do we actually retain? Memoir organizes your knowledge intelligently, making it easy to recall and use when it matters most.

Comment

Memoir: Intelligent Knowledge Organization and Recall

🚀 Inspiration

In today’s digital age, we consume vast amounts of information—articles, videos, podcasts, research papers—but retaining and recalling this knowledge remains a challenge. Traditional bookmarking tools fall short, leaving users lost in an ocean of saved links with no intuitive way to organize or retrieve information when needed.

We wanted to build Memoir—a tool that doesn’t just store content but intelligently connects, structures, and retrieves knowledge dynamically based on user interactions.

💡 What We Learned

Throughout this journey, we explored cutting-edge AI and NLP techniques, diving deep into:

  • BERTopic for topic modeling to group related concepts.
  • Vector embeddings using pgvector for semantic search and contextual recall.
  • Gemini & Prompt Engineering to structure information hierarchically and improve knowledge retrieval.
  • Mindmap Visualization to present stored information in an intuitive, non-linear format.

Beyond the technical aspects, we learned how to design user-centric workflows, making Memoir as seamless and intuitive as possible.

🛠️ How We Built It

Memoir integrates multiple technologies to deliver an adaptive knowledge management experience:

Tech Stack

  • AI & NLP: BERTopic for topic modeling, Gemini for structuring topics.
  • Frontend: Chrome extension, Chatbot with ReactJs.
  • Graph: GIS -> GeoPandas, Leaflet.js, EPSG Projection, Shapely
  • Backend: FastAPI (Python) for efficient processing.
  • Database: pgvector for semantic search and contextual recall.

Core Features

  • Dynamic Knowledge Graph: Automatically links related topics as users add content.
  • Intelligent Memory Recall: Semantic search enables retrieval based on meaning, not just keywords.
  • Context-Aware Retrieval: Users can query Memoir with natural language (e.g., "Show me all articles on AI ethics").
  • Seamless UI & Quick-Save: A visually intuitive interface with one-click saving via a Chrome extension.

⚡ Challenges We Faced

  • Steep Learning Curve: We were new to BERTopic, vector embeddings, and Gemini, so mastering these tools took time.
  • Efficient Mindmap Visualization: Structuring a growing knowledge graph dynamically while keeping the UI responsive was tricky.
  • Optimizing Search & Recall: Tuning semantic search to provide the most relevant results required extensive testing and refinement.

🎯 Impact & Future Enhancements

Memoir transforms knowledge management, making learning more effective and information recall effortless. We’re excited about future enhancements, including:

  • AI-Powered Summarization of stored content.
  • Integration Beyond Chrome, expanding Memoir to multiple platforms.
  • Personalized Content Recommendations based on user interests.
  • Multi-User Collaboration for shared knowledge graphs.
  • Multi-Lingual Support to make knowledge organization accessible globally.

✨ Final Thoughts

Building Memoir was an incredible journey, pushing us to explore AI, NLP, and information retrieval in ways we hadn’t before. The challenge of creating an intelligent, dynamic knowledge assistant taught us not only technical skills but also the importance of designing intuitive and user-friendly experiences.

We’re proud of what we’ve built and excited to keep improving Memoir—because knowledge should be more than just stored; it should be alive, connected, and instantly accessible. 🚀

Built With

Share this project:

Updates