RAG Search

Ingest documents, embed, and answer with citations — search that explains itself.

Retrieval-augmented search as a product feature: an ingestion pipeline for files and URLs, chunking and embeddings into pgvector, hybrid keyword + vector retrieval, and streaming answers with inline citations users can check.

What's inside

• File and URL ingestion pipeline
• Chunking + embedding jobs
• Hybrid keyword + vector retrieval
• Streaming answers with inline citations
• Feedback loop on answer quality
• Eval set scaffold

Quickstart

$ git clone git@github.com:devai-io/rag-search.git my-search$ cd my-search && bun install$ cp .env.example .env   # add ANTHROPIC_API_KEY$ bun dev