Customer service
Basic

WhatsApp customer service with an AI agent

An agent that answers the repetitive part of WhatsApp, keeps the context of the conversation, and hands over to a person when the question leaves its scope.

Time to implement

2 to 4 hours

Estimated cost

15 to 30 EUR / month

Tools

4 in the stack

The stack

Built with tools you already run

n8n
OpenAI (GPT-4)
Evolution API
Google Sheets

The problem

The customer asks, hours pass, the sale goes elsewhere.

In restaurants and clinics the first reply usually decides the enquiry, because whoever is asking is contacting more than one place at the same time. The team, meanwhile, is serving the counter, dealing with suppliers and putting out fires, and WhatsApp comes second.

So messages arrive at night, at the weekend or during lunch, and by the time somebody answers it is four hours later. The enquiry has gone cold, or it has already been answered by someone else.

The flow

How it works, step by step

01

The customer sends a WhatsApp message

Evolution API captures it through a webhook and passes it to n8n.

02

n8n looks up the context

It checks in Google Sheets whether this is a new or a returning customer.

03

The AI answers with that context

The model receives the history plus the business information and replies.

04

The reply goes back to WhatsApp

The customer has an answer in well under a minute.

05

If it is complex, it hands over

The message is flagged and the team is notified with a link to the conversation.

Walkthrough

The build in detail

01

The inbound webhook

Evolution API is configured to call the n8n webhook whenever a message arrives. The payload already carries the number, the text and the timestamp.

02

Customer context

A Google Sheets node looks the number up in your own records. If it is there, you bring in past orders and preferences; if it is not, you create the row.

03

The prompt

The OpenAI node gets a system prompt with the identity of the business (what it sells, opening hours, what it must never answer), then the conversation history, then the current message. The scope is defined here, not later.

04

Reply and record

The reply goes back through Evolution API and is written to the history, so the next conversation starts with context instead of from zero.

05

Handover that keeps the thread

One rule: if the customer asks for a person, or the model detects that the question is outside its scope, the team gets a notification with a direct link to the conversation. The customer does not repeat anything.

What to expect

What this case delivers

Answers within seconds, at any hour
Covers weekends, holidays and the night
Frees the team from the questions that repeat
Volume grows without hiring for it

Blueprint and video in preparation

We are preparing the .json file you can import into n8n, and the video walkthrough. Want this case running in your company already? Book a call.

Book a call

Want this case in your company?

The production version includes integration with the systems you already run, monitoring, a fallback path and support. It is usually working within a week.

Book a free consultation

No commitment · Free of charge