Communication

WhatsApp Business AI automation and integrations

The short answer

WhatsApp Business is automated through the WhatsApp Business Platform Cloud API, hosted by Meta, with built-in nodes in n8n. Businesses can reply freely within 24 hours of a customer's last message; outside that window, only pre-approved template messages can be sent, and those are billed per message. AI customer service agents are allowed, but Meta's terms bar general-purpose AI assistants as the main product. Good uses are support, booking, order updates and reminders.

What you can automate with WhatsApp Business

AI customer support agent

Answer common questions on WhatsApp from your knowledge base and order data, and hand off to a person with a summary when needed.

Read the build guide
Appointment reminders and rescheduling

Send utility template reminders, understand replies such as "can we do 4pm", and update the booking or route to staff.

Read the build guide
Order and delivery updates

Send shipping and delivery templates triggered by your order system, and answer follow-up questions in the open window.

Read the build guide
Reservations and guest messaging

Take booking requests, confirm availability, and send pre-arrival and check-in details for hotels and restaurants.

Read the build guide
Lead follow-up

Reply to inbound leads from click-to-WhatsApp ads or your site within seconds, qualify them and book a call.

Read the build guide
Orders by chat

Let repeat B2B customers place or repeat orders in WhatsApp, with the agent confirming items and totals before submitting.

Read the build guide

What makes WhatsApp Business easy or hard to automate

The Cloud API is a clean HTTP API. You send messages with a POST and receive inbound messages and delivery statuses on a webhook. Text, images, documents, location, buttons and list messages are all supported, so an agent can offer choices instead of asking people to type.

The rules are what make WhatsApp different from SMS or email:

  • The 24-hour window. A customer’s message opens a 24-hour customer service window. Inside it, any reply is allowed. Outside it, only approved templates.
  • Templates. Templates are written ahead of time, categorized as marketing, utility or authentication, and reviewed by Meta. Meta can recategorize a template, which changes its price.
  • Per-message pricing. Since July 2025, Meta bills delivered template messages rather than conversations. Free-form messages inside an open window are free, and utility templates sent inside the window are free too. Check Meta’s current rate card for your markets.
  • Quality and limits. Blocks and reports lower your quality rating, which can hold back messaging limit increases.

The AI layer

WhatsApp is the channel. The AI makes the conversation useful:

  • Answering inside the window. An agent can handle questions about orders, hours, bookings and policies while the window is open, using your data through tool calls.
  • Understanding replies to templates. Customers answer reminders with free text. A model turns “running late” or “need Friday instead” into an action.
  • Handing off cleanly. When a person takes over, the model writes a summary so the customer does not repeat themselves.

Keep the agent focused on your business. That is better for customers and keeps you inside Meta’s policy on AI use.

Limits to plan around

  • Window timing. Track the last inbound message per customer. If the window has closed, the workflow must switch to a template.
  • Template approval. Plan templates early and write them generically enough to reuse. Rejections and recategorizations take time.
  • Portfolio-wide limits. All numbers in one business portfolio share a messaging limit, so a marketing push can starve operational messages.
  • Webhook volume. Every message produces several status webhooks (sent, delivered, read). Filter them early.

How Lumeor typically connects it

We usually connect the Cloud API directly to n8n with the WhatsApp nodes for routine messaging, and use a dedicated agent service when conversations need retrieval, tool calls and memory. Our n8n automations suite case study covers messaging workflows on WhatsApp and Telegram with GPT. For support use cases, start with the AI customer support agent guide. If you also need SMS in the US, see our Twilio page.

Frequently asked questions

What is the WhatsApp 24-hour customer service window?

When a customer messages your business, a 24-hour window opens. Inside it you can send free-form replies, which are not charged. After it closes, you can only start the conversation again with a pre-approved template message, which is billed per delivered message according to its category and the recipient's country.

Can we run an AI chatbot on WhatsApp?

Yes, for your own business purposes. Meta's WhatsApp Business Solution terms, in effect since January 15, 2026, bar providers whose main product is a general-purpose AI assistant from using the platform. A business using AI to answer its own customers about orders, bookings or support is still allowed.

Does n8n work with WhatsApp?

Yes. n8n has a built-in WhatsApp Business Cloud node for sending messages and templates, uploading and downloading media, and a Send and Wait action, plus a WhatsApp Trigger for incoming messages and status updates. You still need your own Meta app, WhatsApp Business Account and phone number.

How many people can we message?

Messaging limits are set per business portfolio and count unique users you contact outside an open customer service window in a rolling 24 hours. New portfolios start low, and limits rise automatically with verification and good message quality. Replies inside the window do not count.

Do customers need to opt in?

Yes. WhatsApp requires that you have the customer's permission to message them on WhatsApp before sending templates. Keep a record of how and when each person opted in, and make it easy to opt out.

Next steps