dabit3/semantic-search-nextjs-pinecone-langchain-chatgpt

Implements a full document ingestion pipeline that vectorizes local text/markdown files through LangchainJS and OpenAI's embedding models, then stores them in Pinecone's vector database for retrieval-augmented generation (RAG) queries via GPT-3. The Next.js API routes handle embedding generation and index initialization with automatic retry logic, while the frontend enables conversational search over custom document collections. Designed as a starter template with pre-configured Lens protocol documentation, making it straightforward to swap in alternative datasets using tools like GPT Repository Loader for ingesting entire codebases.

762 stars.