AI Chatbot Call Center: Demo Call Center (Production-Ready, Part 2)

Description

Workflow Name: ☎️ Demo Call Center

Template was created in n8n v1.90.2

Skill Level: High

Categories: n8n, Chatbot

Stacks

Execute Sub-workflow Trigger node
Chat Trigger node
Redis node
Postgres node
AI Agent node
If node, Switch node, Code node, Edit Fields (Set)

Prerequisite

Execute Sub-workflow Trigger: Telegram Call In Workflow (or your own node)
Sub-workflow: Taxi Service (or your own node)
Sub-workflow: Taxi Booking Worker (or your own node)
Sub-workflow: Demo Call Back (or your own node)

Production Features

Scaling Design* for n8n *Queue mode** in production environment
Optional Rate Limit design to prevent overused
Optional Long Terms Memory design
Multi-Service** design
Testing Flow** with or without dependance on other workflow.
Error Management**

What this workflow does?

This is a n8n Demo Call Center Workflow demo. It is the main entry node for a Multiple Services Chatbot. It will receive message from the Call In Workflow, and decide which service should be use, or which service provider should be process the selected result.

How it works

The Flow Trigger node will wait for the message from the Call In Workflow or other Sub-workflow.
When message is received, it will first check for the Rate Limit.
If ok, load the Session Data from Cache.
Next, check the current Session for the channel_no (default is chat).
if channel_no is chat, continue to the AI Agent for chit-chat.
if channel_no is taxi or others, pass to the Service Input (i.e. Taxi Service) or Service Worker (i.e. Taxi Booking Worker) to handle it directly.
The AI Agent should decide which service (i.e. taxi) will be used at some point and update the channel_no in Session, and pass to the Service Node (i.e. Taxi Service) at once.
In case of any error, reply the error in Call Back.

Set up instructions

Pull and Set up the required SQL from our Github repository.
Create you Redis credentials, refer to n8n integration documentation for more information.
Select your Credentials in Rate Limit, Session, Provider and New Session.
Create you Postgres credentials, refer to n8n integration documentation for more information.
Select your Credentials in Postgres Chat Memory, Load User Memory and Save User Memory.
Modify the AI Agent prompt to fit your need

How to adjust it to your needs

In Session, we have a timestamp fields which is created at the Input node. The usage of this is combined to use with the session id to create a unique session, since some media, such as Telegram, do not have a unique session along with the chat.
You can use any AI Model for the AI Agent node
Learn we use the prompt for the Load/Save User Memory on demand.
Include is our prompt for the taxi service. You can add more service similar to this.