← Back to our workVA TECHSAVVY / WORKFLOW LIBRARY

N8N RAG KNOWLEDGE BASE

RAG knowledge assistant

14 workflow steps · 12 connections · Drive + chat trigger

Problem

Business knowledge stored in documents needed to stay searchable and useful in chat.

Solution

Connected Google Drive changes, document loading, embeddings, Supabase vector storage, and an AI chat agent.

Outcome

A document-aware assistant that can answer from current knowledge sources.

For

Teams maintaining internal knowledge bases.

Automation workflow · n8n

Two parts

Keep knowledge fresh
  1. Drive file changesAdded, edited, deleted
  2. Read + embedVertex AI
  3. StoreSupabase vectors
Answer questions
  1. Question askedChat
  2. Search knowledgeSupabase
  3. AI answersFrom your documents

The actual build

Graph
File Created → Download file (main, output 1)Download file → Supabase Vector Store (main, output 1)Default Data Loader → Supabase Vector Store (ai_document, output 1)Embeddings Google Vertex → Supabase Vector Store (ai_embedding, output 1)Embeddings Google Vertex → Supabase Vector Store1 (ai_embedding, output 1)When chat message received → AI Agent (main, output 1)Supabase Vector Store1 → AI Agent (ai_tool, output 1)File Updated → Delete a row (main, output 1)Delete a row → Download file1 (main, output 1)Download file1 → Supabase Vector Store (main, output 1)File Deleted → Delete a row1 (main, output 1)OpenAI Chat Model → AI Agent (ai_languageModel, output 1)File CreatedgoogleDriveTriggerFile CreatedDownload filegoogleDriveDownload fileSupabase Vector …vectorStoreSupabaseSupabase Vector StoreDefault Data Loa…documentDefaultDataLoaderDefault Data LoaderEmbeddings Googl…embeddingsGoogleVertexEmbeddings Google VertexWhen chat messag…chatTriggerWhen chat message receivedAI AgentagentAI AgentSupabase Vector …vectorStoreSupabaseSupabase Vector Store1File UpdatedgoogleDriveTriggerFile UpdatedDelete a rowsupabaseDelete a rowDownload file1googleDriveDownload file1File DeletedgoogleDriveTriggerFile DeletedDelete a row1supabaseDelete a row1OpenAI Chat ModellmChatOpenAiOpenAI Chat Model

14 workflow steps · 12 connections. Teal lines show the main flow; purple lines connect AI models, memory, and tools. Drag to pan; use + to inspect the workflow.