Build an AI Customer Service Bot with n8n + Claude + Airtable

First, you’ll need a running n8n instance. The fastest way is n8n Cloud, which gives you a managed instance with zero DevOps. If you prefer self-hosting, Docker is the recommended approach.

For this tutorial, we’ll use n8n Cloud since it handles SSL, updates, and uptime automatically. Create your account and navigate to the workflow editor.

# Self-hosted alternative (Docker)
docker run -it --rm \
  --name n8n \
  -p 5678:5678 \
  -v n8n_data:/home/node/.n8n \
  n8nio/n8n

Every n8n workflow starts with a trigger. For our chatbot, we’ll use a Webhook node that listens for incoming messages. This webhook URL is what your frontend chat widget will POST to.

In your workflow editor, click the + button and search for “Webhook.” Set the HTTP method to POST and note the production URL — you’ll need this later.

Now the core of the bot: connecting Claude. n8n’s native Anthropic node supports Claude Opus, Sonnet, and Haiku. Add it after the Webhook node and configure your system prompt.

The system prompt is critical — it defines your bot’s personality, knowledge boundaries, and escalation behavior. Here’s a production-tested template:

You are a helpful customer service assistant for
[Company Name]. Your role is to:

1. Answer questions about our products and services
2. Help troubleshoot common issues
3. Escalate complex issues to human agents

Rules:
- Never make up information you don't have
- If unsure, say "Let me connect you with a
  human agent who can help with that."
- Keep responses concise (under 3 sentences)
- Be warm but professional