Organize your research
Group sources into notebooks. Each notebook is its own searchable knowledge base — keep projects, papers, and notes cleanly separated.
Upload anything
PDFs, websites, audio recordings, images. kvault parses and indexes every format into a unified semantic index you can query.
Ask in plain language
No query syntax. Ask questions the way you think them. kvault retrieves the most relevant chunks across all your sources and synthesizes an answer.
What are the key findings on sleep and memory consolidation?
Sleep plays a critical role in memory consolidation through hippocampal replay during slow-wave sleep. [1] REM sleep strengthens emotional and procedural memories. [2]
Every answer is traceable
Inline citations map each claim back to the exact passage it came from. Open the source and jump to the chunk — no guessing where the answer came from.
…hippocampal sharp-wave ripples during NREM drive the reactivation of recently encoded…
…emotional memory consolidation during REM is mediated by amygdala–hippocampal…
Built to understand
kvault is a portfolio project. The goal was to understand how semantic search, vector embeddings, and AI-synthesized answers actually work — end to end, without abstractions hiding the interesting parts.
Next.js + Python — full-stack from scratch.
pgvector on Neon Postgres for embeddings.
Chunking, retrieval, reranking — full RAG pipeline.
Citations as a first-class feature.