n8n Workflow: Index Local Docs for RAG with Ollama & Qdrant

Index your local documents for RAG—automatically

This n8n workflow reads a document from your local disk, turns it into vector embeddings using Ollama (model: nomic-embed-text), and stores the results in a Qdrant collection—ready for on-prem Retrieval-Augmented Generation (RAG).

What this workflow does

• Runs manually when you click Execute workflow in n8n.
• Reads a local file from the n8n host/container filesystem using the Read/Write Files from Disk step.
• Extracts text from the file (configured for PDF extraction).
• Splits extracted text into overlapping chunks with a recursive character splitter.
• Generates embeddings for each chunk via Ollama using nomic-embed-text.
• Upserts vectors into Qdrant in a specified collection (e.g., handbook).

Use cases

• Index a PDF handbook or internal policy docs into Qdrant so your RAG assistant can cite relevant sections.
• Create an on-prem document knowledge base for SaaS ops teams without sending data to external services.
• Refresh embeddings for updated documents on demand—ideal for periodic knowledge maintenance.

Technical details

• Required environment: Self-hosted n8n. This workflow depends on the local-filesystem Read/Write Files from Disk node and will not run on n8n Cloud.
• Ollama integration: add an n8n credential pointing to your Ollama instance and ensure you’ve pulled nomic-embed-text (e.g., ollama pull nomic-embed-text).
• Qdrant integration: configure Qdrant API credentials (URL + API key) and set the collection name.
• Nodes / tech stack: Manual Trigger, Read/Write Files from Disk, file text extraction, n8nn8n-nodes-langchain Ollama Embeddings and Qdrant Vector Store nodes.