How to Build an AI Customer Service Bot with n8n and Claude

In this guide you'll build a fully automated customer service workflow using n8n's AI agent nodes and Claude. The result: a bot that answers FAQs from your knowledge base, escalates complex issues to a human queue, and creates support tickets automatically.

Prerequisites: n8n installed (Community Edition), Claude API key from console.anthropic.com, Supabase account for the vector knowledge base, 30-60 minutes.

Step 1 — Set up your Supabase knowledge base. Create a new project, enable the pgvector extension, and create a documents table with a vector(1536) column. Upload your FAQ documents and embed them using Supabase's built-in embedding API.

Step 2 — Build the n8n workflow. Start with a Webhook trigger. Add a Supabase Vector Store node to perform similarity search against your FAQ embeddings. Connect to a Claude AI Agent node with a system prompt instructing it to use the retrieved context and only answer from the provided facts.

Step 3 — Add routing logic. Add an n8n If node to check response confidence. Route high-confidence answers directly to the reply node. Route uncertain responses to a Slack notification node for human review.

Step 4 — Add a ticket creation node (Linear, Jira, or HubSpot) for unresolved issues. Test with 20 real support queries, then deploy.