4. RAG & Vector Store Integration

Build knowledge base chatbots with RAG.

20m Lesson 4 of 5

Retrieval Augmented Generation (RAG) combines a knowledge base with AI to answer questions accurately based on your own data.

RAG Architecture

Embed: Convert documents into vector embeddings
Store: Save embeddings in a vector database (Pinecone, Qdrant, Supabase)
Retrieve: Find relevant documents for a user's question
Generate: Use retrieved context + AI to generate accurate answers

Implementation in n8n

n8n has built-in support for vector stores through the AI Agent and Vector Store nodes. You can build a complete RAG pipeline without writing code.

Ingestion Pipeline 1. Load documents (PDF, HTML, Markdown) 2. Split into chunks using the Text Splitter node 3. Generate embeddings via OpenAI Embeddings node 4. Store in Pinecone, Qdrant, or Supabase pgvector

Query Pipeline 1. Receive user question via webhook or chat 2. Embed the question 3. Search vector store for top-K similar chunks 4. Pass chunks as context to the LLM 5. Return the AI-generated answer

Code Examples

RAG Query via Code Node

javascript

// After retrieving relevant chunks from vector store:
const question = $json.user_question;
const chunks = $json.retrieved_chunks;

// Build context from retrieved documents
const context = chunks
  .map((c, i) => `[Source ${i + 1}: ${c.metadata.filename}]\n${c.text}`)
  .join("\n\n---\n\n");

const prompt = `Answer the question based ONLY on the provided context.
If the answer isn't in the context, say "I don't have enough information."

Context:
${context}

Question: ${question}

Provide a clear answer and cite which source(s) you used.`;

return [{ json: { prompt, sources: chunks.map(c => c.metadata.filename) } }];

Pro Tips

💡Chunk your documents into 500–1000 token segments for optimal retrieval accuracy
💡Always include the source document reference in AI responses for transparency
💡Use overlap between chunks (50-100 tokens) to avoid losing context at chunk boundaries

Comprehension Quiz

Answer 2 of 2 correctly to unlock lesson completion.

1. What does RAG stand for?

2. What is the recommended chunk size for RAG documents?

Pass the quiz above to unlock lesson completion

Previous Next Lesson

All lessons in this chapter

1. Connecting to LLM Providers15m 2. Content Generation & Summarization20m 3. Classification & Sentiment Analysis15m 4. RAG & Vector Store Integration20m 5. AI Agents & Tool Calling20m