âš¡n8n Workflow
Expert
Building RAG Chatbot for Movie Recommendations with Qdrant and OpenAI
Create a RAG-powered movie recommendation chatbot using n8n. This workflow automates fetching movie data, creating embeddings with OpenAI, storing them in a Qdrant vector store, and providing an interactive chat experience.
Overview
Creates a sophisticated movie recommendation chatbot by leveraging a RAG architecture with OpenAI for embeddings and language modeling, and Qdrant as the vector store.
🧠Description & Use Case
This workflow builds a complete Retrieval-Augmented Generation (RAG) chatbot designed for movie recommendations. It leverages Qdrant as a vector database and OpenAI for generating embeddings and powering the chat agent. The workflow is split into two main parts: data indexing and the interactive chatbot.
🔄 How It Works:
Part 1: Data Indexing (Populating the Knowledge Base)
- Trigger: The process starts manually with the
When clicking ‘Test workflow’node. - Fetch Data: The
GitHubnode retrieves a CSV file containing a list of top IMDB movies and their descriptions. - Process and Chunk: The
Extract from Filenode parses the CSV. TheDefault Data Loaderthen prepares each movie's description and attaches metadata (name, release year). - Create Embeddings: The
Embeddings OpenAInode converts the movie descriptions into numerical vector representations using thetext-embedding-3-smallmodel. - Store in Vector DB: The
Qdrant Vector Storenode inserts these embeddings and their associated metadata into a Qdrant collection, making the movie data searchable.
Part 2: Chatbot Interaction & Recommendation
- Chat Trigger: A user interacts with the chatbot through the
When chat message receivedtrigger. - AI Agent: The
AI Agentnode, powered by OpenAI'sgpt-4o-minimodel, receives the user's message. It's designed to understand the user's request for movie recommendations (e.g., "I want a romantic comedy but not a horror movie"). - Custom Tool Execution: The agent uses the
Call n8n Workflow Toolnamedmovie_recommender. This tool's job is to query the Qdrant database based on the user's request. - Vector Query:
- The tool takes the user's positive and negative preferences, generates embeddings for them using
OpenAI, and sends a recommendation query to theQdrant Recommendation API. - This allows for nuanced searches, finding movies similar to the "positive" examples while being dissimilar to the "negative" ones.
- The tool takes the user's positive and negative preferences, generates embeddings for them using
- Retrieve & Format: The workflow retrieves the movie details (name, description, year) for the recommended movie IDs from Qdrant, formats the data cleanly, and aggregates it into a single response.
- Final Response: The formatted list of movies is returned to the AI Agent, which then presents the top 3 recommendations to the user in a natural, conversational way, as instructed by its system prompt.
✅ Real-World Use Cases:
- Personalized Recommendations: Create a chatbot that gives users movie suggestions based on their specific tastes, including what they like and dislike.
- Semantic Search Engine: Adapt the workflow to build a search engine for any knowledge base, such as internal company documents, product catalogs, or support articles.
- Advanced AI Agents: Use this as a template for building complex AI agents in n8n that can interact with external APIs and databases via custom tools.
- Educational Tool: A practical example for learning how to implement the RAG (Retrieval-Augmented Generation) pattern from scratch.
This is a comprehensive blueprint for creating intelligent, context-aware chatbots that can query and reason over your custom data. 🚀
Instructions
Basic steps
- Download workflow JSON
- Import into n8n
- Configure node parameters as needed
- Test and enable the workflow
💡 Tips
After importing, verify all connectors and credentials match your environment.
Preview
Workflow Visualization
27 nodes
100%
Nodes
1
When clicking ‘Test workflow’
manualTrigger
v1
2
GitHub
github
v1
3
Extract from File
extractFromFile
v1
4
Embeddings OpenAI
@n8n/langchain.embeddingsOpenAi
v1
5
Default Data Loader
@n8n/langchain.documentDefaultDataLoader
v1
6
Token Splitter
@n8n/langchain.textSplitterTokenSplitter
v1
7
Qdrant Vector Store
@n8n/langchain.vectorStoreQdrant
v1
8
When chat message received
@n8n/langchain.chatTrigger
v1.1
9
OpenAI Chat Model
@n8n/langchain.lmChatOpenAi
v1
10
Call n8n Workflow Tool
@n8n/langchain.toolWorkflow
v1.2
11
Window Buffer Memory
@n8n/langchain.memoryBufferWindow
v1.2
12
Execute Workflow Trigger
executeWorkflowTrigger
v1
13
Merge
merge
v3
14
Split Out
splitOut
v1
15
Split Out1
splitOut
v1
16
Merge1
merge
v3
17
Aggregate
aggregate
v1
18
AI Agent
@n8n/langchain.agent
v1.6
19
Embedding Recommendation Request with Open AI
httpRequest
v4.2
20
Embedding Anti-Recommendation Request with Open AI
httpRequest
v4.2
21
Extracting Embedding
set
v3.4
22
Extracting Embedding1
set
v3.4
23
Calling Qdrant Recommendation API
httpRequest
v4.2
24
Retrieving Recommended Movies Meta Data
httpRequest
v4.2
25
Selecting Fields Relevant for Agent
set
v3.4
26
Sticky Note
stickyNote
v1
27
Sticky Note1
stickyNote
v1
Statistics
Total nodes:27
Disabled nodes:0
Node types:19
Connections:22
View full JSON structure
{
"nodes": [
{
"id": "06a34e3b-519a-4b48-afd0-4f2b51d2105d",
"name": "When clicking ‘Test workflow’",
"type": "n8n-nodes-base.manualTrigger",
"position": [
4980,
740
],
"parameters": {},
"typeVersion": 1
},
{
"id": "9213003d-433f-41ab-838b-be93860261b2",
"name": "GitHub",
"type": "n8n-nodes-base.github",
"position": [
5200,
740
],
"parameters": {
"owner": {
"__rl": true,
"mode": "name",
"value": "mrscoopers"
},
"filePath": "Top_1000_IMDB_movies.csv",
"resource": "file",
"operation": "get",
"repository": {
"__rl": true,
"mode": "list",
"value": "n8n_demo",
"cachedResultUrl": "https://github.com/mrscoopers/n8n_demo",
"cachedResultName": "n8n_demo"
},
"additionalParameters": {}
},
"credentials": {
"githubApi": {
"id": "VbfC0mqEq24vPIwq",
"name": "GitHub n8n demo"
}
},
"typeVersion": 1
},
{
"id": "9850d1a9-3a6f-44c0-9f9d-4d20fda0b602",
"name": "Extract from File",
"type": "n8n-nodes-base.extractFromFile",
"position": [
5360,
740
],
"parameters": {
"options": {}
},
"typeVersion": 1
},
{
"id": "7704f993-b1c9-477a-8b5a-77dc2cb68161",
"name": "Embeddings OpenAI",
"type": "@n8n/n8n-nodes-langchain.embeddingsOpenAi",
"position": [
5560,
940
],
"parameters": {
"model": "text-embedding-3-small",
"options": {}
},
"credentials": {
"openAiApi": {
"id": "deYJUwkgL1Euu613",
"name": "OpenAi account 2"
}
},
"typeVersion": 1
},
{
"id": "bc6dd8e5-0186-4bf9-9c60-2eab6d9b6520",
"name": "Default Data Loader",
"type": "@n8n/n8n-nodes-langchain.documentDefaultDataLoader",
"position": [
5700,
960
],
"parameters": {
"options": {
"metadata": {
"metadataValues": [
{
"name": "movie_name",
"value": "={{ $('Extract from File').item.json['Movie Name'] }}"
},
{
"name": "movie_release_date",
"value": "={{ $('Extract from File').item.json['Year of Release'] }}"
},
{
"name": "movie_description",
"value": "={{ $('Extract from File').item.json.Description }}"
}
]
}
},
"jsonData": "={{ $('Extract from File').item.json.Description }}",
"jsonMode": "expressionData"
},
"typeVersion": 1
},
{
"id": "f87ea014-fe79-444b-88ea-0c4773872b0a",
"name": "Token Splitter",
"type": "@n8n/n8n-nodes-langchain.textSplitterTokenSplitter",
"position": [
5700,
1140
],
"parameters": {},
"typeVersion": 1
},
{
"id": "d8d28cec-c8e8-4350-9e98-cdbc6da54988",
"name": "Qdrant Vector Store",
"type": "@n8n/n8n-nodes-langchain.vectorStoreQdrant",
"position": [
5600,
740
],
"parameters": {
"mode": "insert",
"options": {},
"qdrantCollection": {
"__rl": true,
"mode": "id",
"value": "imdb"
}
},
"credentials": {
"qdrantApi": {
"id": "Zin08PA0RdXVUKK7",
"name": "QdrantApi n8n demo"
}
},
"typeVersion": 1
},
{
"id": "f86e03dc-12ea-4929-9035-4ec3cf46e300",
"name": "When chat message received",
"type": "@n8n/n8n-nodes-langchain.chatTrigger",
"position": [
4920,
1140
],
"webhookId": "71bfe0f8-227e-466b-9d07-69fd9fe4a27b",
"parameters": {
"options": {}
},
"typeVersion": 1.1
},
{
"id": "ead23ef6-2b6b-428d-b412-b3394bff8248",
"name": "OpenAI Chat Model",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
5040,
1340
],
"parameters": {
"model": "gpt-4o-mini",
"options": {}
},
"credentials": {
"openAiApi": {
"id": "deYJUwkgL1Euu613",
"name": "OpenAi account 2"
}
},
"typeVersion": 1
},
{
"id": "7ab936e1-aac8-43bc-a497-f2d02c2c19e5",
"name": "Call n8n Workflow Tool",
"type": "@n8n/n8n-nodes-langchain.toolWorkflow",
"position": [
5320,
1340
],
"parameters": {
"name": "movie_recommender",
"schemaType": "manual",
"workflowId": {
"__rl": true,
"mode": "id",
"value": "a58HZKwcOy7lmz56"
},
"description": "Call this tool to get a list of recommended movies from a vector database. ",
"inputSchema": "{\n\"type\": \"object\",\n\"properties\": {\n\t\"positive_example\": {\n \"type\": \"string\",\n \"description\": \"A string with a movie description matching the user's positive recommendation request\"\n },\n \"negative_example\": {\n \"type\": \"string\",\n \"description\": \"A string with a movie description matching the user's negative anti-recommendation reuqest\"\n }\n}\n}",
"specifyInputSchema": true
},
"typeVersion": 1.2
},
{
"id": "ce55f334-698b-45b1-9e12-0eaa473187d4",
"name": "Window Buffer Memory",
"type": "@n8n/n8n-nodes-langchain.memoryBufferWindow",
"position": [
5160,
1340
],
"parameters": {},
"typeVersion": 1.2
},
{
"id": "41c1ee11-3117-4765-98fc-e56cc6fc8fb2",
"name": "Execute Workflow Trigger",
"type": "n8n-nodes-base.executeWorkflowTrigger",
"position": [
5640,
1600
],
"parameters": {},
"typeVersion": 1
},
{
"id": "db8d6ab6-8cd2-4a8c-993d-f1b7d7fdcffd",
"name": "Merge",
"type": "n8n-nodes-base.merge",
"position": [
6540,
1500
],
"parameters": {
"mode": "combine",
"options": {},
"combineBy": "combineAll"
},
"typeVersion": 3
},
{
"id": "c7bc5e04-22b1-40db-ba74-1ab234e51375",
"name": "Split Out",
"type": "n8n-nodes-base.splitOut",
"position": [
7260,
1480
],
"parameters": {
"options": {},
"fieldToSplitOut": "result"
},
"typeVersion": 1
},
{
"id": "a2002d2e-362a-49eb-a42d-7b665ddd67a0",
"name": "Split Out1",
"type": "n8n-nodes-base.splitOut",
"position": [
7140,
1260
],
"parameters": {
"options": {},
"fieldToSplitOut": "result.points"
},
"typeVersion": 1
},
{
"id": "f69a87f1-bfb9-4337-9350-28d2416c1580",
"name": "Merge1",
"type": "n8n-nodes-base.merge",
"position": [
7520,
1400
],
"parameters": {
"mode": "combine",
"options": {},
"fieldsToMatchString": "id"
},
"typeVersion": 3
},
{
"id": "b2f2529e-e260-4d72-88ef-09b804226004",
"name": "Aggregate",
"type": "n8n-nodes-base.aggregate",
"position": [
7960,
1400
],
"parameters": {
"options": {},
"aggregate": "aggregateAllItemData",
"destinationFieldName": "response"
},
"typeVersion": 1
},
{
"id": "bedea10f-b4de-4f0e-9d60-cc8117a2b328",
"name": "AI Agent",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
5140,
1140
],
"parameters": {
"options": {
"systemMessage": "You are a Movie Recommender Tool using a Vector Database under the hood. Provide top-3 movie recommendations returned by the database, ordered by their recommendation score, but not showing the score to the user."
}
},
"typeVersion": 1.6
},
{
"id": "e04276b5-7d69-437b-bf4f-9717808cc8f6",
"name": "Embedding Recommendation Request with Open AI",
"type": "n8n-nodes-base.httpRequest",
"position": [
5900,
1460
],
"parameters": {
"url": "https://api.openai.com/v1/embeddings",
"method": "POST",
"options": {},
"sendBody": true,
"sendHeaders": true,
"authentication": "predefinedCredentialType",
"bodyParameters": {
"parameters": [
{
"name": "input",
"value": "={{ $json.query.positive_example }}"
},
{
"name": "model",
"value": "text-embedding-3-small"
}
]
},
"headerParameters": {
"parameters": [
{
"name": "Authorization",
"value": "Bearer $OPENAI_API_KEY"
}
]
},
"nodeCredentialType": "openAiApi"
},
"credentials": {
"openAiApi": {
"id": "deYJUwkgL1Euu613",
"name": "OpenAi account 2"
}
},
"typeVersion": 4.2
},
{
"id": "68e99f06-82f5-432c-8b31-8a1ae34981a6",
"name": "Embedding Anti-Recommendation Request with Open AI",
"type": "n8n-nodes-base.httpRequest",
"position": [
5920,
1660
],
"parameters": {
"url": "https://api.openai.com/v1/embeddings",
"method": "POST",
"options": {},
"sendBody": true,
"sendHeaders": true,
"authentication": "predefinedCredentialType",
"bodyParameters": {
"parameters": [
{
"name": "input",
"value": "={{ $json.query.negative_example }}"
},
{
"name": "model",
"value": "text-embedding-3-small"
}
]
},
"headerParameters": {
"parameters": [
{
"name": "Authorization",
"value": "Bearer $OPENAI_API_KEY"
}
]
},
"nodeCredentialType": "openAiApi"
},
"credentials": {
"openAiApi": {
"id": "deYJUwkgL1Euu613",
"name": "OpenAi account 2"
}
},
"typeVersion": 4.2
},
{
"id": "ecb1d7e1-b389-48e8-a34a-176bfc923641",
"name": "Extracting Embedding",
"type": "n8n-nodes-base.set",
"position": [
6180,
1460
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "01a28c9d-aeb1-48bb-8a73-f8bddbd73460",
"name": "positive_example",
"type": "array",
"value": "={{ $json.data[0].embedding }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "4ed11142-a734-435f-9f7a-f59e2d423076",
"name": "Extracting Embedding1",
"type": "n8n-nodes-base.set",
"position": [
6180,
1660
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "01a28c9d-aeb1-48bb-8a73-f8bddbd73460",
"name": "negative_example",
"type": "array",
"value": "={{ $json.data[0].embedding }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "ce3aa9bc-a5b1-4529-bff5-e0dba43b99f3",
"name": "Calling Qdrant Recommendation API",
"type": "n8n-nodes-base.httpRequest",
"position": [
6840,
1500
],
"parameters": {
"url": "https://edcc6735-2ffb-484f-b735-3467043828fe.europe-west3-0.gcp.cloud.qdrant.io:6333/collections/imdb_1000_open_ai/points/query",
"method": "POST",
"options": {},
"jsonBody": "={\n \"query\": {\n \"recommend\": {\n \"positive\": [[{{ $json.positive_example }}]],\n \"negative\": [[{{ $json.negative_example }}]],\n \"strategy\": \"average_vector\"\n }\n },\n \"limit\":3\n}",
"sendBody": true,
"specifyBody": "json",
"authentication": "predefinedCredentialType",
"nodeCredentialType": "qdrantApi"
},
"credentials": {
"qdrantApi": {
"id": "Zin08PA0RdXVUKK7",
"name": "QdrantApi n8n demo"
}
},
"typeVersion": 4.2
},
{
"id": "9b8a6bdb-16fe-4edc-86d0-136fe059a777",
"name": "Retrieving Recommended Movies Meta Data",
"type": "n8n-nodes-base.httpRequest",
"position": [
7060,
1460
],
"parameters": {
"url": "https://edcc6735-2ffb-484f-b735-3467043828fe.europe-west3-0.gcp.cloud.qdrant.io:6333/collections/imdb_1000_open_ai/points",
"method": "POST",
"options": {},
"jsonBody": "={\n \"ids\": [\"{{ $json.result.points[0].id }}\", \"{{ $json.result.points[1].id }}\", \"{{ $json.result.points[2].id }}\"],\n \"with_payload\":true\n}",
"sendBody": true,
"specifyBody": "json",
"authentication": "predefinedCredentialType",
"nodeCredentialType": "qdrantApi"
},
"credentials": {
"qdrantApi": {
"id": "Zin08PA0RdXVUKK7",
"name": "QdrantApi n8n demo"
}
},
"typeVersion": 4.2
},
{
"id": "28cdcad5-3dca-48a1-b626-19eef657114c",
"name": "Selecting Fields Relevant for Agent",
"type": "n8n-nodes-base.set",
"position": [
7740,
1400
],
"parameters": {
"options": {},
"assignments": {
"assignments": [
{
"id": "b4b520a5-d0e2-4dcb-af9d-0b7748fd44d6",
"name": "movie_recommendation_score",
"type": "number",
"value": "={{ $json.score }}"
},
{
"id": "c9f0982e-bd4e-484b-9eab-7e69e333f706",
"name": "movie_description",
"type": "string",
"value": "={{ $json.payload.content }}"
},
{
"id": "7c7baf11-89cd-4695-9f37-13eca7e01163",
"name": "movie_name",
"type": "string",
"value": "={{ $json.payload.metadata.movie_name }}"
},
{
"id": "1d1d269e-43c7-47b0-859b-268adf2dbc21",
"name": "movie_release_year",
"type": "string",
"value": "={{ $json.payload.metadata.release_year }}"
}
]
}
},
"typeVersion": 3.4
},
{
"id": "56e73f01-5557-460a-9a63-01357a1b456f",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
5560,
1780
],
"parameters": {
"content": "Tool, calling Qdrant's recommendation API based on user's request, transformed by AI agent"
},
"typeVersion": 1
},
{
"id": "cce5250e-0285-4fd0-857f-4b117151cd8b",
"name": "Sticky Note1",
"type": "n8n-nodes-base.stickyNote",
"position": [
4680,
720
],
"parameters": {
"content": "Uploading data (movies and their descriptions) to Qdrant Vector Store\n"
},
"typeVersion": 1
}
],
"connections": {
"Merge": {
"main": [
[
{
"node": "Calling Qdrant Recommendation API",
"type": "main",
"index": 0
}
]
]
},
"GitHub": {
"main": [
[
{
"node": "Extract from File",
"type": "main",
"index": 0
}
]
]
},
"Merge1": {
"main": [
[
{
"node": "Selecting Fields Relevant for Agent",
"type": "main",
"index": 0
}
]
]
},
"Split Out": {
"main": [
[
{
"node": "Merge1",
"type": "main",
"index": 1
}
]
]
},
"Split Out1": {
"main": [
[
{
"node": "Merge1",
"type": "main",
"index": 0
}
]
]
},
"Token Splitter": {
"ai_textSplitter": [
[
{
"node": "Default Data Loader",
"type": "ai_textSplitter",
"index": 0
}
]
]
},
"Embeddings OpenAI": {
"ai_embedding": [
[
{
"node": "Qdrant Vector Store",
"type": "ai_embedding",
"index": 0
}
]
]
},
"Extract from File": {
"main": [
[
{
"node": "Qdrant Vector Store",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Chat Model": {
"ai_languageModel": [
[
{
"node": "AI Agent",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Default Data Loader": {
"ai_document": [
[
{
"node": "Qdrant Vector Store",
"type": "ai_document",
"index": 0
}
]
]
},
"Extracting Embedding": {
"main": [
[
{
"node": "Merge",
"type": "main",
"index": 0
}
]
]
},
"Window Buffer Memory": {
"ai_memory": [
[
{
"node": "AI Agent",
"type": "ai_memory",
"index": 0
}
]
]
},
"Extracting Embedding1": {
"main": [
[
{
"node": "Merge",
"type": "main",
"index": 1
}
]
]
},
"Call n8n Workflow Tool": {
"ai_tool": [
[
{
"node": "AI Agent",
"type": "ai_tool",
"index": 0
}
]
]
},
"Execute Workflow Trigger": {
"main": [
[
{
"node": "Embedding Recommendation Request with Open AI",
"type": "main",
"index": 0
},
{
"node": "Embedding Anti-Recommendation Request with Open AI",
"type": "main",
"index": 0
}
]
]
},
"When chat message received": {
"main": [
[
{
"node": "AI Agent",
"type": "main",
"index": 0
}
]
]
},
"Calling Qdrant Recommendation API": {
"main": [
[
{
"node": "Retrieving Recommended Movies Meta Data",
"type": "main",
"index": 0
},
{
"node": "Split Out1",
"type": "main",
"index": 0
}
]
]
},
"When clicking ‘Test workflow’": {
"main": [
[
{
"node": "GitHub",
"type": "main",
"index": 0
}
]
]
},
"Selecting Fields Relevant for Agent": {
"main": [
[
{
"node": "Aggregate",
"type": "main",
"index": 0
}
]
]
},
"Retrieving Recommended Movies Meta Data": {
"main": [
[
{
"node": "Split Out",
"type": "main",
"index": 0
}
]
]
},
"Embedding Recommendation Request with Open AI": {
"main": [
[
{
"node": "Extracting Embedding",
"type": "main",
"index": 0
}
]
]
},
"Embedding Anti-Recommendation Request with Open AI": {
"main": [
[
{
"node": "Extracting Embedding1",
"type": "main",
"index": 0
}
]
]
}
}
}Actions
Share
Categories:
Tags:
manualtriggersetgithubmergehttprequeststickynotesplitoutexecuteworkflowtriggeraggregateextractfromfile
Technical Specs
Complexity:
Expert
Node Count:
27 nodes
Published:11 months ago
Updated:about 2 months ago
Compatibility
Platform:
N8N Cloud & Self-hosted
Min Version:
N8N v1.0.0+
Supports cloud and self-hosted deployment
License
This workflow template follows MIT license, you can freely use, modify and distribute.
Please comply with relevant third-party service terms when using.