GitHub radar
Utopia: A Self-updating Knowledge Base that works Offline
A bitemporal knowledge graph built by DeepLethe which indexes your company’s documents and allows you to query it using an AI agent offline.
Utopia is an open source enterprise knowledge graph built by DeepLethe in Rust. It parses documents, creates a bitemporal knowledge graph (meaning each fact has a “valid time” and “stated time”), and allows you to query the knowledge graph using an AI agent. It supports PDF, DOCX, PPTX, XLSX, CSV, Markdown, HTML, Notion, GitHub, Jira, WebDAV, RSS, and S3. It uses full text search (Tantivy), vector search (pgvector), graph traversal (RRF Fusion), and streaming results with inline citations. It has multi-user access control, an append only audit log, an ontology workbench, and forward chaining reasoning. It was released just one month ago but already has over 4000 GitHub stars. It’s currently at version 0.1 so the database schema will change with forward only migrations, so back it up first!
Why a vibe-coder should care
Do you have knowledge scattered throughout various Notion pages, GitHub repos, and random PDF files? Wouldn’t it be nice if you could just query all this information from a central location and get back citations for your answers? With Utopia, you can deploy this system in minutes using Docker Compose and run it entirely on your own hardware. You can even use the built in AI agent to query your documents and navigate the knowledge graph! And because it’s bitemporal, you can query the knowledge graph to understand what was known at any given point in time.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Deploy Utopia from https://github.com/deeplethe/utopia — follow the README (docker compose --profile app up -d), open localhost:1516, create an admin account, and show me how to upload the first documents. Ask if you need any credentials or keys.
Runs on a regular laptop — only Docker needed.
Open on GitHub▌ More finds