Build a local question-answering interface for your notes
At a glance
A developer's RAG Builder indexes personal notes locally and returns model answers with source attribution. This workflow is listed under Local AI, RAG, Personal Knowledge.
- Agents / tools
- RAG Builder, Ollama, ChromaDB, Hugging Face
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Install Node.js 14 or later, Ollama and ChromaDB. Clone ps011/rag-builder from GitHub and run npm install in its directory. 2) Pull the llama3 model with ollama pull llama3 and start ollama serve. Start ChromaDB with chroma run --host localhost --port 8000. 3) Create a project .env with OLLAMA_HOST=http://localhost:11434, OLLAMA_MODEL=llama3, CHROMA_URL=http://localhost:8000, SOURCE_TYPE=local and LOCAL_PATH pointing to your notes directory. 4) Run npm run web and open localhost:3000. Use Settings to add the local source and choose file extensions, then click Refresh to build the index. 5) Ask a question about your notes. Read the generated answer and source attribution; refresh the index after changing the source collection.
Worth knowing
The first embedding-model download requires internet access. The article's older module names differ from the repository's current layout.