I would like to create a chatbot that sets appointments with customers via WhatsApp. For testing purposes, I would like to implement this with the N8N chat.
Important requirements:
The chatbot should open the chat and inform the lead with a standardised message that they can book an appointment.
After 24 hours, the chatbot should follow up again. Also with standardised messages.
If the lead asks questions in between, the chatbot should answer them.
The chatbot should always have the entire chat in its memory.
It is therefore a chatbot that sends messages proactively according to defined rules, but is also able to react.
So far, I have only developed reactive chatbots that respond. How do I implement this hybrid form?
I asked ChatGPT about this and ChatGPT advised me to create two AI agents. One for the proactive message after 24 hours and one for the reactive behaviours. Somehow, however, the whole thing doesn’t really work. The AI agent above sends out a message, but it is not transferred to the chat window, probably because the chat is no longer active after a certain time (the time that is bridged in the ‘Wait’ node), although I have entered the session ID in the simple memory node. I also don’t know how the workflow recognises which AI agent should respond to the messages. Somehow I’m still a bit stuck.
You can use something similar to what we have created for a client.
This first one is normal website chatbot which you can expand as per your requirement, by adding google calender and many other things.
Second to get all the data regarding the chat memory with the client, you can use following system.
Chat Memory System
And for the last followup setup, you can just use the simple Google sheet or Airtable related automation to followup with specific clients.
Note: As we have develop a custom code base website chatbot, inside code we have add some line of code to send chat memory on webhook once client end the chat with us.
If you want those code or any farther help, please feel free to reach out.
Happy to help.
(If this solve your problem, please mark it as solution, because this will help others too.)
If I understand you correctly, you are proposing three systems:
A reactive agent that answers questions (workflow 1]
A chat memory (workflow 2)
A proactive follow-up agent (workflow 3)
Now my question:
The first workflow should access the chat memory – just like the follow-up agent. How can I use the Airtable database in workflow 2 maintained in the chat memory system as a shared memory for both?
In your example, the workflow is started via webhook triggered from mail and a website chatbot, so that it can also access older, long-term stored chats. This allows you to respond to chats from long ago. However, if I start the whole thing via the n8n chat trigger, a session ID is generated which, as far as I know, is only temporary and only valid within this workflow.
My questions about this:
Can I access this session ID with another workflow and thus continue the same chat? If not, what workaround would you recommend?
