AI Customer Support n8n Workflow: WhatsApp + Web Chat, RAG

Turn WhatsApp and Web Chat into instant AI customer support with RAG—powered by n8n

This AI Customer Support n8n workflow automatically replies to customers on WhatsApp and an n8n Web Chat, using Ollama for answers, Qdrant for RAG knowledge retrieval, and Redis to deduplicate messages and maintain conversation context.

What this workflow does

• Listens for new messages from either the n8n Web Chat webhook or the WhatsApp Business Cloud webhook.
• For WhatsApp, filters to real incoming text messages and uses Redis to deduplicate by message ID.
• Normalizes both channels into a shared format, applies your business configuration, and checks Redis to determine whether the conversation was seen before.
• If the message is just a greeting, sends a short tailored welcome for new vs returning conversations, then marks the conversation as seen in Redis.
• Otherwise, uses an Ollama-powered AI agent with Redis chat memory, and optionally Qdrant retrieval to answer using your knowledge base.
• When appropriate, it can read Products/Properties and append Leads/Tickets in Google Sheets.
• Responds back to the original channel (WhatsApp send or Web Chat response) and marks the conversation as seen.

Use cases

• SaaS support that answers FAQs via WhatsApp + Web Chat with RAG-backed accuracy.
• Lead capture: convert inbound questions into Leads/Tickets stored in Google Sheets.
• Ecommerce/property-style inquiries: look up Products/Properties and respond immediately.

Technical details

• Nodes/stack: if, set, redis, whats app, sticky note, n8n-nodes-langchainagent (AI agent)
• Integrations: Ollama (response generation), Redis (dedup + chat memory), Qdrant (RAG retrieval), Google Sheets (product/property lookups and lead/ticket creation)