Back to Templates

Provide ecommerce support with Qdrant RAG, WooCommerce, and human backup

Created by

Created by: Davide Boizza || n3witalia
Davide Boizza

Last update

Last update 6 hours ago

Categories

Share

This workflow is a complete AI-powered customer support automation for WooCommerce e-commerce websites.
It combines conversational AI, Retrieval-Augmented Generation (RAG), vector search, WooCommerce integration, and human escalation into a single intelligent support system.

The workflow allows customers to interact with an AI assistant through a website chat interface, providing real-time support, product discovery, policy assistance, and personalized shopping guidance.

Key Advantages

1. ✅ Intelligent Customer Support

The chatbot can answer customer questions instantly, reducing response times and improving customer satisfaction.

Benefits:

  • 24/7 automated assistance
  • Faster customer response times
  • Reduced support workload
  • Consistent customer experience

2. ✅ Real-Time WooCommerce Integration

The workflow connects directly to WooCommerce to retrieve live product data.

Capabilities:

  • Product search
  • Availability checks
  • Price retrieval
  • Product recommendations
  • Category browsing

This ensures customers always receive accurate and up-to-date information.

3. ✅ RAG-Powered Knowledge Base

The workflow uses Qdrant Vector Store and OpenAI embeddings to create a semantic knowledge base from company documents stored in Google Drive.

Advantages:

  • AI can answer questions about:

    • shipping policies
    • returns and refunds
    • FAQs
    • store information
    • sizing guides
    • promotions

  • Context-aware responses

  • Better accuracy than traditional keyword search

  • Easy document updates through Google Drive synchronization

4. ✅ AI Guardrails and Safer Conversations

The workflow includes dedicated AI guardrails before the AI agent processes customer messages.

Benefits:

  • Safer interactions
  • Reduced hallucinations
  • Better brand protection
  • More controlled AI behavior
  • Prevention of off-topic conversations

5. ✅ Personalized Shopping Experience

The AI agent acts like a virtual shopping assistant.

Capabilities:

  • Product recommendations
  • Outfit suggestions
  • Personalized guidance
  • Product comparisons
  • Contextual conversations using memory

This creates a premium customer experience similar to an in-store assistant.

6. ✅ Conversation Memory

The workflow includes a buffer memory system that maintains context during conversations.

Advantages:

  • More natural conversations
  • Better understanding of customer preferences
  • Improved multi-step interactions
  • Personalized responses across the session

7. ✅ Human Escalation System

When the AI cannot fully resolve a request, the workflow automatically escalates the conversation to human support via Gmail.

Benefits:

  • Seamless customer handoff
  • Better customer satisfaction
  • Reduced frustration
  • Support continuity
  • Full conversation transcript included

8. ✅ Scalable and Modular Architecture

The workflow is built using modular n8n nodes and reusable AI tools.

Advantages:

  • Easy to customize
  • Easy to scale
  • Simple maintenance
  • Flexible integrations
  • Suitable for multiple e-commerce industries

Ideal Use Cases

This workflow is ideal for:

  • Fashion e-commerce stores
  • Luxury boutiques
  • WooCommerce websites
  • Online retail businesses
  • Customer support automation
  • AI shopping assistants

How it works

This workflow implements an AI-powered customer support chatbot for an e-commerce store (Fashionart). It integrates with webiste chatbot (chat trigger), Qdrant (vector store), Google Drive (knowledge base), WooCommerce (product data), and Gmail (human escalation).

  1. User message received – The workflow starts when a customer sends a message via the chat trigger.
  2. Guardrails applied – The input is first passed through a Guardrails node to enforce safety and content policies.
  3. AI Agent processes the request – An E-Commerce Customer Support AI Agent (LangChain agent) decides which tool to call based on the user’s intent:
    • rag_search – Retrieves company policies, FAQs, and static knowledge from Qdrant (populated from Google Drive documents).
    • get_product – Fetches live product details from WooCommerce.
    • get_many_products – Searches/multiple products from WooCommerce.
    • get_human_support – Escalates to staff by sending a transcript + customer phone/email via Gmail.
    • Calculator – Optional tool for simple calculations.
  4. Memory context – A Window Buffer Memory node keeps conversation history (last 10 messages) for contextual replies.
  5. Response returned – The agent’s output is sent back to the customer via the chat node.

The workflow also includes a manual ingestion branch (Test workflow → Google Drive → Qdrant) to vectorize documents into the knowledge base.

Set up steps

  1. Create Qdrant collection

    • Ensure Qdrant is running at http://qdrant_jush:6333
    • Run the Create collection HTTP request node (or manually create a collection named fashionart with vector size 1536, Cosine distance).

  2. Configure Google Drive

    • Set the folder ID in the Get folder node (contains all product/policy documents).
    • Ensure the Download Files node can access the drive (authenticate with OAuth2).

  3. Set up OpenAI embeddings

    • Add your OpenAI API key in the Embeddings OpenAI and Embeddings OpenAI2 nodes.
    • Same for OpenAI Chat Model1 (or switch to the existing Google Gemini Chat Model with proper credentials).

  4. Configure WooCommerce tools

    • Add WooCommerce credentials (API key, secret, store URL) in get_product and get_many_products nodes.

  5. Set up Gmail for human escalation

    • Authenticate the get_human_support Gmail node.
    • Update the recipient email ([email protected] to the actual support email).

  6. Adjust agent system prompt

    • Review the system prompt inside the E-Commerce Customer Support AI Agent node – it contains the full assistant behavior, tone, and tool usage logic.
    • Change any store-specific details (name, address, examples).

  7. Activate chat trigger

    • Deploy the workflow.
    • Use the webhook URL from the When chat message received node to connect to javascript chatbot or any chat frontend.

  8. Populate the vector store

    • Click Test workflow (manual trigger) to run the ingestion branch – this will read documents from Google Drive, split them, embed them, and store them in Qdrant.

  9. Set workflow to active – Change "active": false to true in the JSON or via n8n UI.

👉 Subscribe to my new YouTube channel. Here I’ll share videos and Shorts with practical tutorials and FREE templates for n8n.

image

Need help customizing?

Contact me for consulting and support or add me on Linkedin.