Ether Notes

🌟 Inspiration

We were frustrated by the endless cycle of saving bookmarks, taking scattered notes, and losing track of important research across different websites. Traditional note-taking apps require manual organization, while bookmarks become digital graveyards. We envisioned an AI-powered system that could understand context, automatically organize knowledge, and make research truly intelligent - like having a personal research assistant that never forgets.

🚀 What it does

Ether Notes is an intelligent knowledge management system that transforms how you capture and organize web content:

Smart Text Capture: Highlight any text on any website and Ether automatically saves it with intelligent categorization
AI-Powered Organization: No manual folders - AI understands your content and creates topics automatically (e.g., "Machine Learning Research", "Cooking Recipes")
Intelligent Routing: AI decides whether to append to existing notes, merge similar content, or create new notes
Context-Aware Chat: Ask questions about your saved research and get intelligent responses based on your actual notes
Seamless Integration: Chrome extension + web dashboard for access anywhere

🛠️ How we built it

Architecture: Microservices with AI-first design

Chrome Extension: JavaScript/ES6 with Webpack bundling for text selection and real-time saving
Backend API: Node.js/Fastify handling authentication, note management, and routing decisions
LLM Service: Python/FastAPI with OpenAI integration for content analysis and categorization
Database: PostgreSQL with pgvector for semantic similarity search using embeddings
Frontend: Next.js dashboard for note management and advanced features

Key Innovation - Phase 3 Combined LLM Architecture:

Single AI call handles both topic assignment AND content routing (50% API reduction)
User intent weighted as primary signal (70%) over pure content similarity
Binary decision framework - AI makes clear choices without confidence thresholds

💪 Challenges we ran into

Complex Content Routing: Deciding when to append vs. create new notes required sophisticated prompt engineering and multiple LLM decision frameworks
Real-time Performance: Balancing AI intelligence with speed - optimized from 3 API calls to 1 combined call
Cross-Website Compatibility: Making text selection work reliably across different website structures and CSS frameworks
Semantic Understanding: Teaching AI to understand user intent vs. just content similarity - breakthrough came with weighted decision making
Chrome Extension Complexity: Managing state between content scripts, background workers, and popup interfaces

🏆 Accomplishments that we're proud of

AI-First Architecture: Built a system where AI makes intelligent decisions without human intervention
50% API Efficiency Gain: Optimized from multiple LLM calls to single combined analysis
Universal Compatibility: Works on any website with intelligent text selection and positioning
Smart Evolution: Topics can evolve (e.g., "Apple Stock" → "Tech Stock Analysis") based on user behavior
Production-Ready: Full authentication, error handling, and scalable microservices architecture
Intelligent Chunking: Handles large content with semantic splitting and cached embeddings

📚 What we learned

Prompt Engineering is Critical: Spent significant time crafting prompts that balance user intent with content analysis
User Intent > Similarity: Pure semantic similarity often misses what users actually want - context is king
Chrome Extension Complexity: Modern web extensions require sophisticated state management and cross-script communication
AI Decision Frameworks: Binary decisions work better than confidence scores for user-facing features
Performance vs. Intelligence Trade-offs: Every AI call adds latency - optimization is crucial for user experience

🔮 What's next for Ether Notes

Immediate Roadmap:

Smart Connections: Automatically link related notes across different topics
Advanced Search: Natural language search across all saved content
Collaboration Features: Share topics and notes with team members
Mobile App: Native iOS/Android apps with offline sync

Vision:

Research Assistant: AI that proactively suggests connections and insights from your knowledge base
Auto-Summarization: Weekly/monthly summaries of your research patterns and key insights
Integration Ecosystem: Connect with Notion, Obsidian, and other productivity tools
Enterprise Features: Team knowledge bases with advanced analytics and insights

The Goal: Transform Ether Notes from a note-taking tool into an intelligent research companion that helps users discover insights they never knew existed in their own knowledge.

*Built with ❤️ and lots of coffee.