Hi everyone
I’m running a WhatsApp automation with ManyChat + n8n. Everything works fine when the client sends a message — it goes to n8n, automation replies, no problem.
But here’s the issue:
When I, as the human agent, reply manually to the client in the chat from my Phone Whatsapp, ManyChat still sends the client’s last message again through the webhook to n8n.
-
This makes the automation think the client just sent that message again.
-
Result → Automation replies to the old message → creating a loop and confusion.
What I’ve noticed:
-
The webhook payload shows the timestamp of the client’s last message (not my agent reply).
-
So ManyChat is basically re-sending the last subscriber input whenever I send something.
Question: How can I prevent ManyChat from re-triggering the webhook when the message source is me (agent) instead of the subscriber (client)?
-
Is there a way to filter only subscriber messages in the Flow or webhook trigger?
-
Or a setting to pause automation when an agent is chatting manually?
Any workaround, best practice, or advice would be hugely appreciated
Thanks in advance!
— Shoaib
FIX I TRIED.
My n8n automation (so no reply goes out when I intervene), but the problem is that ManyChat itself still sends the last AI reply back into the chat whenever I (the agent) type something manually.
So even though n8n doesn’t respond, the client still sees the previous automation message repeated.
It looks like the issue is on ManyChat’s side — it’s pushing the last subscriber/AI message into the thread whenever I reply, instead of ignoring it.
Do you know if there’s any way to stop ManyChat from doing that? Or a setting to prevent “last automation message” from being resent when an agent replies
Thanks a lot for the suggestion I already tried filtering inside n8n so that only messages where
from=subscriber
get processed — and I also tested returning empty responses. That part works fine.
The problem is that even when n8n doesn’t send anything, ManyChat still pushes the last automation/AI reply back into the chat as soon as I (the agent) send a manual message.
So the client ends up seeing the old automation response again, which creates confusion.
It seems like ManyChat itself is re-inserting the last bot reply into the conversation whenever an agent intervenes.