Find the most relevant workflow templates using RAG, Qdrant and Gemini

Go to Workflow
0 views
Built by ARofiqi Maulana ARofiqi Maulana
Created on June 11, 2026

Description

🤖 AI Workflow Recommender (RAG + Qdrant + Gemini)

This workflow helps users find the most relevant n8n templates using AI.

It combines Retrieval-Augmented Generation (RAG), vector search (Qdrant), and Gemini to understand user intent and recommend workflows based on meaning, not just keywords.

⚙️ How it works

Collect workflow templates from the n8n API using multiple search queries
Process and clean the data (split, format, deduplicate)
Convert workflows into embeddings using Gemini
Store embeddings in a vector database (Qdrant)
Accept user queries via chat interface
Convert queries into embeddings
Retrieve relevant workflows using semantic search
Generate AI-powered recommendations with explanations and template links

🚀 What this workflow does

Understands user intent (not just keywords)
Finds relevant workflows using semantic similarity
Recommends the best workflows with explanations
Provides ready-to-use template links

🧩 Setup steps

Set up Qdrant (Cloud or self-hosted)
Add Google Gemini API credentials
Run the Data Ingestion workflow to populate the database
Activate the RAG chatbot workflow

⚠️ Important

Make sure the vector database is populated before using the chatbot
Ensure embedding model and vector dimension match
Free-tier APIs may have rate limits

🎥 Tutorial

@youtube

Nodes Used (8)

AI Agent
@n8n/n8n-nodes-langchain.agent
Code
n8n-nodes-base.code
Default Data Loader
@n8n/n8n-nodes-langchain.documentDefaultDataLoader
Embeddings Google Gemini
@n8n/n8n-nodes-langchain.embeddingsGoogleGemini
Google Gemini Chat Model
@n8n/n8n-nodes-langchain.lmChatGoogleGemini
HTTP Request
n8n-nodes-base.httpRequest
Qdrant Vector Store
@n8n/n8n-nodes-langchain.vectorStoreQdrant
Simple Memory
@n8n/n8n-nodes-langchain.memoryBufferWindow