AI Pizza Order Chatbot with OpenAI and Tools

Overview

This workflow builds an intelligent chatbot for taking pizza orders. It uses OpenAI's language model as the brain (the AI Agent), allows customers to ask about the menu, place orders, and check order status, all through a simple chat interface. The AI agent is equipped with custom tools to look up product details, submit orders, and retrieve order status from an external system (likely a webhook or API).

This is a great starting point for any business wanting to automate customer service or order taking via a chatbot, without coding the conversation logic manually.

Workflow Steps

1. Manual Trigger (No auth) — This node is a placeholder to manually start the workflow for testing purposes during development. It does not trigger automatically.

2. When chat message received (No auth) — This is the actual entry point for the chatbot. It creates a webhook that listens for incoming chat messages. The node is configured with an initial greeting message from "Pizzaro" (the pizza assistant). The user's messages are passed to the AI Agent.

3. AI Agent (No auth) — This is the core of the workflow. It is configured to use the chat input from the previous node (={{ $json.chatInput }}). It uses a defined prompt (not a simple prompt) with a system message: "Your name is Pizzaro, and you are an assistant for handling customer pizza orders." The system instruction tells the agent how to react to menu inquiries, order placements, and status checks. This node has tools connected to it (Get Products, Order Product, Get Order) to perform actions.

4. Get Products (No auth — uses a URL from environment variable {{ $env.WEBHOOK_URL }}) — This is a tool for the AI Agent. It is an HTTP Request node (no-op placeholder) configured with toolDescription: "Retrieve detailed information about the product menu." This lets the AI call this tool dynamically when a user asks about the menu. The URL is stored as an environment variable.

5. Order Product (No auth — uses a URL from environment variable {{ $env.WEBHOOK_URL }}) — Another AI tool. This is configured to make a POST request to the same webhook URL, sending a body with the message field containing the chat input. The toolDescription: "Process product orders." This allows the AI to place orders on behalf of the customer.

6. Get Order (No auth — uses a URL from environment variable {{ $env.WEBHOOK_URL }}) — A third AI tool, configured as a GET request to the webhook URL. The toolDescription: "Get the order status." This lets the AI check the status of a previously placed order.

7. Chat OpenAI (No auth) — This is an unused/open loop node (no-op) originally intended for the OpenAI call, but the actual logic is now in the AI Agent.

8. Window Buffer Memory (No auth) — An unused node placeholder for a window buffer memory, likely intended to store conversation history.

9. Calculator (No auth) — An unused node placeholder, possibly for future calculations.

Setup Instructions

Services & Accounts Needed

1. n8n (self-hosted or cloud) — You need an n8n instance to import and run this workflow.
2. OpenAI API Key — For the AI Agent to work, you need an API key from OpenAI. In n8n, you'll set a credential for the OpenAI node (though this example uses a placeholder).
3. External Webhook/API endpoint — The "Get Products", "Order Product", and "Get Order" nodes use a URL stored in an environment variable ($env.WEBHOOK_URL). You need to set this environment variable in n8n (Settings -> Environment Variables) to point to your own backend API that can return a menu, accept orders, and return order status.

Credential Setup

• OpenAI: Create an "OpenAI" credential in n8n (API Key). The workflow currently uses placeholder nodes, so ensure you replace the no-op "Chat OpenAI" node with a real one and connect it to the AI Agent.
• Webhook (your own API): If your endpoint requires authentication (e.g., an API key), you would add that to the HTTP Request nodes. By default, these use no auth.

Environment Variables

• WEBHOOK_URL: The base URL of your backend API that serves the menu, handles orders, and provides order status.

Use Cases & Variations

Primary Use Cases

• Restaurant / Food Ordering: Replace pizza with any food item (burgers, sushi, etc.). Customize the AI Agent's system message.
• E-commerce Product Inquiry: Instead of a menu, have the tool return product catalog and prices.
• Customer Support Bot: Use the same pattern to answer FAQs (tool returning knowledge base articles) and escalate to a human.

Variations to Adapt This Workflow

• Replace AI Agent with a different LLM: Use the Anthropic or Azure OpenAI node instead of OpenAI.
• Add memory: The placeholder "Window Buffer Memory" is there. Connect it to the AI Agent to maintain conversation context across multiple messages.
• Use Google Sheets as data source: Instead of an API, have the AI Agent tools fetch data from a Google Sheet (using Google Sheets nodes) for the menu.
• Add Slack or Telegram integration: Replace the chat trigger with a Slack or Telegram trigger to let users order via those platforms.
• Log orders to a database: Add a node after the "Order Product" tool to save the order to a database (e.g., Postgres, Airtable).
• Send confirmation email: Use the Gmail or SendGrid node to email the customer their order confirmation.

{
  "id": "5Y8QXJ3N67wnmR2R",
  "meta": {
    "instanceId": "workflow-cee44ac3",
    "versionId": "1.0.0",
    "createdAt": "2025-09-29T07:07:42.135607",
    "updatedAt": "2025-09-29T07:07:42.135632",
    "owner": "n8n-user",
    "license": "MIT",
    "category": "automation",
    "status": "active",
    "priority": "high",
    "environment": "production"
  },
  "name": "POC - Chatbot Order by Sheet Data",
  "tags": [
    "automation",
    "n8n",
    "production-ready",
    "excellent",
    "optimized"
  ],
  "nodes": [
    {
      "id": "trigger-670fd43a",
      "name": "Manual Trigger",
      "type": "n8n-nodes-base.manualTrigger",
      "typeVersion": 1,
      "position": [
        100,
        100
      ],
      "parameters": {}
    },
    {
      "id": "cc9ab139-303f-411a-a7c8-5985d92e3040",
      "name": "Calculator",
      "type": "n8n-nodes-base.noOp",
      "position": [
        1460,
        480
      ],
      "parameters": {},
      "typeVersion": 1,
      "notes": "This toolCalculator node performs automated tasks as part of the workflow."
    },
    {
      "id": "97a6d3a8-001c-4c62-84c2-da5b46a286a9",
      "name": "Chat OpenAI",
      "type": "n8n-nodes-base.noOp",
// ... truncated (copy to see full JSON)