LHC + WhatsApp API + n8n/AI Integration – Automated Department Transfer via Bot

[joederfmoura]

Hello everyone,

I am developing a customer support flow using WhatsApp API + Live Helper Chat (LHC) + n8n with AI. The goal is to:

Receive messages from WhatsApp in LHC.

Forward the messages from LHC to n8n, where the AI processes them and generates responses.

Deliver the responses back to the user through LHC.

The current challenge is transferring chats to human agents when the AI identifies that it is necessary (for example, “São Paulo”).

My proposal is:

n8n sends a signal indicating the need for transfer, without showing it in the chat history for either the user or the agent.

The bot/LHC analyzes this response and makes automatic decisions, including:

Continuing the conversation via AI

Transferring the user to a specific department

Technical questions:

What is the best way to send this signal from n8n to LHC (API endpoint, webhook, custom field, etc.)?

Is it possible for LHC to perform the department transfer internally based on this signal, without exposing the command in the message history?

Any recommendations for structuring the decision logic in the bot so it interprets n8n responses and automatically decides the chat flow?

If anyone has implemented something similar or can share examples of n8n → LHC integration for automatic transfers, it would be extremely helpful.

Thank you!

[remdex]

Hi,

1. Option
   Take a look here. There is sample how I did that. In this sample all is async and I just use https://api.livehelperchat.com/#/chat/post_restapi_addmsgadmin with status attribute once transfer is required.
   https://doc.livehelperchat.com/docs/bot/n8n-integration#flow-example-with-non-async-flow

2. N8N decides to transfer chat and just calls rest API
   https://api.livehelperchat.com/#/chat/post_restapi_setchatstatus

3. You can return just some specific text as a keyword for LHC and afterwards use just. It adds a little bit overhead as we will be checking what was returned from n8n, but in that case you just need to pass specific text from N8N as a message.

4. Most proper way would be if in Rest API parsing you could differentiate is it tool call AKA function call or regular answer and in that case you could just look how https://doc.livehelperchat.com/docs/bot/chatgpt-responses works :)

[joederfmoura]

Hello @remdex
,
I’m integrating n8n → LHC to automatically transfer chats via bot.
I receive the following JSON from n8n:

{
"msg": "I'll transfer You, Wait",
"comando": "TRANSFER",
"next_dep_id": "2"
}

The problem is that next_dep_id comes as a string, and I need to use it as an integer in the “Change Department” command in LHC via Update Current Chat.

Question: what is the best way to convert next_dep_id to an integer inside the LHC bot so that the command works correctly, without sending extra messages to the user?

I’ve already tried {{parseInt(content_2)}}, but I would like confirmation whether this is the correct method or if there is a native way in LHC to handle numeric strings received from n8n.

[remdex]

Hi,

If you followed chatgpt sample

1. You should be at step where you know that transfer to dep/op has to be done.
2. In that case you can just use transfer to operator trigger response.
3. If you want transfer to specific department you can just have different command from n8n and trigger.
4. If you want transfer explicitly to n8n provided dep I need to check can you use e.g {content_2} as destination dep.

https://doc.livehelperchat.com/docs/bot/bot-update-current-chat#change-department

[joederfmoura]

Hi @remdex, thank you very much for your patience in helping me.

I’m trying to use {content_3} to receive the department ID dynamically.

My Scenario

Department IA → ID 1

Department_São_Paulo → ID 2

Department_Rio_de_Janeiro → ID 3
...

Department_Brasilia → ID 15

The user starts the chat and talks with the AI.
The AI determines which department is the best one for the transfer and returns the following content via REST API CALL response:

{
"msg": "I'll transfer You, Wait",
"comando": "TRANSFER",
"next_dep_id": "2"
}

API Mapping

{content_1} ← "msg"

{content_2} ← "comando"

{content_3} ← "next_dep_id"

Bot Logic

The bot receives the information from the API and displays {content_1} to the user.

If {content_2} == TRANSFER, it triggers "TRANSFER TO DEPARTMENT".

Inside the "TRANSFER TO DEPARTMENT" trigger:

1 - Update Current Chat
Command: Stop and transfer to human

2 - Update Current Chat
Command: Change department
Department ID: {content_3}

My Question

This is the flow I imagined for my scenario.
Since I will have many departments, it would be difficult to maintain static rules for each one, so I thought of passing the department ID dynamically via n8n and handling it in the bot.

Am I doing this correctly? I couldn’t find anything similar in the documentation.

[remdex]

No the best practise. First you define in API parsing custom output named command. So you can reuse it in the future.

• Don't forget priority
• Don't forget conditions I put.

Preferably your api should return

{
"msg": "I'll transfer You, Wait",
"comando": "TRANSFER",
"args" : {"next_dep_id": "2"}
}

Then in chat you store args as chat variable for later reuse just like openai sample. I don't know why you refuse to import rest api and bot sample from there...

Then in rest api you define something like

Then in tool call you store args and search for trigger by command

In trigger you define keyword in your case it would be TRANSFER

E.g

This way you will have 100% dynamic function calling where tool call part will remain the same always. Just import non streaming version to look how it's done

https://doc.livehelperchat.com/docs/bot/chatgpt-responses#rest-api

[remdex]

Please watch and adopt
https://youtu.be/spREDwfeoCE

Bot API
lhc-bot-1.json

Rest API
rest-api-1.json