WhatsApp support agent for a WooCommerce shop

What it is

This n8n workflow answers your shop’s customers on WhatsApp. A message reaches your Twilio number, and an AI agent replies through Twilio a moment later. The agent can see what is in stock in WooCommerce, read an order by its number and log a support ticket in Google Sheets. It remembers the last 50 exchanges of each customer’s chat. The file had eight nodes. Mindket rebuilt one and added a ninth, because the OpenAI service it used has closed.

Who it suits

Small WooCommerce shops whose customers would rather message than email, and that want the routine questions answered at any hour. It fits a team that keeps support in a sheet. Skip it if you need to approve each reply first. The agent answers on its own.

You need some n8n, a Twilio account with a WhatsApp sender and a WooCommerce store with API keys. Its author puts setup at about 16 minutes, with the accounts ready.

Where to use it

A published n8n that Twilio can reach. Customers write to your shop’s WhatsApp number, and the replies come back in the same chat.

• n8n Cloud: import the file and publish it, since the trigger only listens while it is published
• Your own n8n server: it must be reachable from the internet over HTTPS
• WhatsApp: through Twilio’s Sandbox while you test, then an approved WhatsApp sender
• Google Sheets: the tickets land in one tab for your team to work through

Setup

Get it saves a zip with workflow.json and Mindket’s setup notes in your downloads folder. Make the ticket sheet and the WooCommerce keys first.

1. In n8n, create a workflow, open its menu, choose Import from File and pick workflow.json.
2. Add a Twilio credential with your Account SID and Auth Token, and select it in Twilio Trigger and Send an SMS/MMS/WhatsApp message.
3. Add an OpenAI credential and select it in OpenAI Chat Model.
4. Add a WooCommerce credential with your store’s address, consumer key and secret, and select it in Get Inventory and Get Orders.
5. Make the sheet, add a Google Sheets OAuth2 credential, select it in Create Tickets and replace YOUR_SHEET_ID with the sheet’s ID.
6. In Message a model, replace YOUR_SHOP_NAME in the system message and add your own rules: delivery times, returns, opening hours.
7. Save and press Publish. The trigger then subscribes to your Twilio account’s incoming messages.
8. Send your WhatsApp number a question about a product and read the reply.

The original’s Message a model called an assistant kept in its author’s OpenAI account. OpenAI’s deprecations page, read on 2 November 2026, gives 26 August 2026 as the shutdown date of the Assistants API. So the node is now an AI Agent with the same name, message, memory and tools, a new OpenAI Chat Model and a short system message of Mindket’s.

How to use it

Leave it published. Each message to your number starts a run. The agent then decides which of its three tools, if any, the question needs, and whether the answer can come from the chat so far. A stock question reads every in-stock product from WooCommerce. An order number fetches that order. A complaint it cannot settle becomes a row in the sheet, numbered by the moment it was made. Then the reply goes back through Twilio.

Example

A customer writes to your WhatsApp number.

Hi, my order 1042 still hasn't arrived after a week. Can someone check?

The agent calls Get Orders with 1042, reads the order’s status and, if it is still processing, says so in plain words. If the customer wants a person, it asks for a name and an email, calls Create Tickets and sends back the issue number. The row looks like this:

For newcomers and experts

If you are new. n8n draws a workflow as boxes joined by lines, and each box, a node, does one job. An agent is a node that decides which of its tools to call. Its tools, memory and model hang under it on thin lines. Test with the Twilio Sandbox and your own phone, then open the run under Executions to see which tool it called.

For experts. The agent’s prompt is the customer’s text plus the current time. Its memory is keyed by the sender’s number without the whatsapp: prefix.

Good to know

The agent replies without review, and one tool can read any order.

• Billing: Twilio bills each WhatsApp message in and out, and OpenAI bills the agent’s calls, more of them when it uses a tool
• Order privacy: Get Orders returns any order whose number a customer types. Tell the agent to compare the order’s billing phone with the sender’s number before it shares anything
• No review: replies go out as the agent writes them, so test the rules in the system message with your own phone first
• Memory: Simple Memory lives inside n8n, so a restart clears every chat’s history
• WhatsApp rules: the Sandbox only reaches phones that joined it. A live number needs a WhatsApp sender approved through Twilio
• Model date: OpenAI’s deprecations page lists GPT-4.1 mini (gpt-4.1-mini-2025-04-14) for removal from the API on 14 January 2027, with gpt-6-luna as its replacement
• Your data: customers’ messages, order details and your product list go to OpenAI. Tickets keep names, emails and phone numbers in your sheet
• Credentials: the JSON holds none, and nothing from its author’s sheet or OpenAI account