AI-powered Zendesk support responses with RAG, OpenAI, and Supabase knowledge base
Needs changesNeeds changes. This template uses 1 thing(s) that n8n 3.0 removes: AI Agent v1 (old agent modes). 1 breaking item(s) found. Static check, last run 2026-10-19.
What to replace before upgrading
| Removed in 3.0 | Replace with |
|---|---|
| AI Agent v1 (old agent modes) RAG AI Agent (v1.6) | Latest AI Agent node (Tools Agent). For SQL Agent, use a Postgres or MySQL tool with a recent AI Agent. |
Nodes used (9)
- Webhook
n8n-nodes-base.webhook - Edit Fields (Set)
n8n-nodes-base.set - Supabase Vector Store
@n8n/n8n-nodes-langchain.vectorStoreSupabase - Postgres Chat Memory
@n8n/n8n-nodes-langchain.memoryPostgresChat - Embeddings OpenAI
@n8n/n8n-nodes-langchain.embeddingsOpenAi - Vector Store Question Answer Tool
@n8n/n8n-nodes-langchain.toolVectorStore - AI Agent
@n8n/n8n-nodes-langchain.agent - If
n8n-nodes-base.if - HTTP Request
n8n-nodes-base.httpRequest
What it does
⚡ How it works This workflow automates first responses to new Zendesk tickets with the help of AI and your internal knowledge base. Webhook trigger fires whenever a new ticket is created in Zendesk. Ticket details (subject, description, requester info) are extracted. Knowledge base retrieval – the workflow searches a Supabase vector store (with OpenAI embeddings) for the most relevant KB articles. AI assistant (RAG agent) drafts a professional reply using the retrieved KB and conversation memory stored in Postgres. Decision logic: If no relevant KB info is found (or if it’s a sensitive query like KYC, refunds, or account deletion), the workflow sends a fallback response and tags the ticket for human review. Otherwise, it posts the AI-generated reply and tags the ticket with aireply. Logging & context memory ensure future ticket updates are aware of past interactions. ------ 🔧 Set up steps This workflow takes about 15–30 minutes to set up. Connect credentials for Zendesk, OpenAI, Supabase, and Postgres. Prepare your knowledge base: store support content in Supabase (documents table) and embed it using the provided Embeddings node. Set up Postgres memory table (zendesktickethistories) to store conversation history. Update your Zendesk domain in the HTTP Request nodes (<YOURZENDESKDOMAIN>). Deploy the webhook URL in Zendesk triggers so new tickets flow into n8n. Test by creating a sample ticket and verifying: AI replies appear in Zendesk Correct tags (aireply or humanrequested) are applied Logs are written to Postgres