📖 Ask questions about a PDF using AI
⚡ 142,454 views · 📖 Internal Wiki & Knowledge Base
Description
The workflow first populates a Pinecone index with vectors from a Bitcoin whitepaper. Then, it waits for a manual chat message. When received, the chat message is turned into a vector and compared to the vectors in Pinecone. The most similar vectors are retrieved and passed to OpenAI for generating a chat response.
Note that to use this template, you need to be on n8n version 1.19.4 or later.
🔗 Nodes Used
Google Drive, Question and Answer Chain, Binary Input Loader, Embeddings OpenAI, OpenAI Chat Model, Vector Store Retriever
📥 Import
Download workflow.json and import into n8n:
Workflow menu → Import from File