CASE STUDY 01 / 08 · CUSTOMER EXPERIENCE
AI Customer Support Agent
The WhatsApp AI CRM Assistant receives messages through Wassenger, checks and updates customer records in Google Sheets, uses an AI agent with conversation memory and FAQ context to respond, then logs both sides of the conversation.
01 / THE PROBLEM
The challenge.
Businesses receive repetitive customer questions and can lose potential customers when responses are slow.
02 / THE SOLUTION
The connected answer.
The WhatsApp AI CRM Assistant receives messages through Wassenger, checks and updates customer records in Google Sheets, uses an AI agent with conversation memory and FAQ context to respond, then logs both sides of the conversation.
03 / HOW IT WORKS
Inside the logic.
A customer sends a WhatsApp message, which reaches n8n through a webhook from Wassenger, the service connecting the workflow to WhatsApp.
The workflow pulls the customer’s details out of the message and checks them against a customer database kept in Google Sheets. If the customer is already known, their record is updated. If not, a new customer record is created.
It then retrieves the earlier conversation history, so the assistant knows the context of what has already been discussed.
An AI agent — using an OpenAI chat model, conversation memory and an FAQ knowledge base kept in a Google Sheet — writes the reply.
Both the incoming and outgoing messages are logged to Google Sheets, and the reply is sent back to the customer on WhatsApp.
- 01
WhatsApp message
Wassenger webhook receives the incoming message.
- 02
Extract customer info
Identify the customer from the message.
- 03
Search customer database
Look up the customer in Google Sheets.
- 04
Existing customer?
Yes
Update the customer record.
No
Create a customer record.
- 05
Retrieve conversation history
Bring earlier messages into context.
- 06
AI agent generates reply
OpenAI chat model + conversation memory + FAQ knowledge base sheet.
- 07
Log inbound message
Save the customer’s message.
- 08
Log outbound message
Save the generated response.
- 09
Send reply
Deliver through Wassenger to WhatsApp.
BUILD GUIDE
How to build it.
A practical guide to recreating this workflow in n8n.
- 01
Connect WhatsApp through Wassenger
Create a Wassenger account, link a WhatsApp number, and add an n8n Webhook node. Paste the webhook URL into Wassenger so every incoming message is sent to n8n.
- 02
Extract the customer details
Add a Set (Edit Fields) node that pulls the phone number, name and message text out of the webhook payload into clean fields.
- 03
Prepare the Google Sheets database
Create a spreadsheet with a Customers tab, a Messages log tab and an FAQ tab. Connect your Google account in n8n credentials.
- 04
Look up the customer
Use a Google Sheets "Get rows" node filtered by phone number, then an IF node to check whether a row was found.
- 05
Update or create the record
On the "yes" branch, update the existing row (e.g. last contact date). On the "no" branch, append a new customer row.
- 06
Load conversation history
Read earlier messages for this phone number from the Messages tab so the assistant has context.
- 07
Build the AI agent
Add an AI Agent node with an OpenAI Chat Model, a memory sub-node keyed by phone number, and a Google Sheets tool pointing to the FAQ tab. Write a system prompt telling it to answer only from the FAQ and be friendly and brief.
- 08
Log both messages
Append the incoming message and the AI reply as two rows in the Messages tab, with timestamp and direction.
- 09
Send the reply
Use an HTTP Request node to call the Wassenger send-message API with the phone number and the generated reply.
- 10
Test and activate
Message the number from a test phone, check each node’s output in n8n, adjust the prompt, then switch the workflow to Active.
04 / TECHNOLOGY USED
Connected tools.
05 / WORKFLOW SCREENSHOTS
Behind the build.
Screenshot: WhatsApp AI CRM Assistant workflow in n8n
Select the image to enlarge.
06 / RESULT
What it demonstrates.
A WhatsApp assistant that answers from a business’s own FAQ, remembers earlier conversations, and keeps a customer record and full message log without manual work.
WHAT THIS PROJECT SHOWS
UP NEXT
Job & Resume Tracker