Your Personal
Knowledge Operating System
Upload documents, research papers, and notes. Explore connections, retrieve knowledge instantly, and interact with your information through AI.
Features
Six capabilities that turn a folder of files into a connected, queryable knowledge system.
Knowledge Graph Visualization
Force-directed graph of documents, concepts, and citations rendered with React Flow.
Multi-Document Retrieval
Query across the entire library in one pass — chunks are ranked and merged before generation.
Semantic Search
Embeddings + pgvector similarity search surface meaning rather than keyword matches.
Source-Grounded Responses
Every answer is linked back to the originating chunk, page, and document.
Cross-Document Reasoning
Synthesize across multiple sources to surface agreements, contradictions, and gaps.
Interactive Research Workspace
Side-by-side graph, chat, and document viewer for exploring an idea end-to-end.
From raw files to living knowledge
Upload Documents
PDFs, papers, notes, bookmarks — drop them in.
Create Embeddings
Each chunk is vectorized for meaning-based recall.
Build Knowledge Graph
Concepts and citations are linked automatically.
Ask Questions
Query across everything in natural language.
Answers With Citations
Every claim grounded in the source.
Engineering Challenges Solved
The four core problems behind making a personal knowledge OS feel effortless.
Document Processing
Extract and normalize information from PDFs, notes, and markdown files into clean, chunkable text.
Semantic Retrieval
Retrieve relevant context using embeddings and vector similarity search across the full library.
Knowledge Graph Construction
Visualize relationships between documents, concepts, and retrieved information as a live graph.
Citation Grounding
Generate responses backed by traceable source references — every claim tied to a chunk.
Built with modern tooling
A pragmatic stack chosen for performance, developer experience, and AI-native workflows.
- React
- Tailwind CSS
- Framer Motion
- React Flow
- FastAPI
- Python
- Gemini API
- Embeddings
- RAG Pipeline
- Supabase
- pgvector
Why I Built SecondBrain
Most valuable information gets trapped inside PDFs, notes, documentation, and research papers. Finding connections between ideas often requires manually searching across dozens of files.
SecondBrain was built to transform isolated documents into a connected knowledge network that can be explored visually and queried naturally using AI.