DEV Community

Cover image for Integrating AWS Connect with Amazon Lex to understand voice input Part1
Muhammed Ashraf
Muhammed Ashraf

Posted on

Integrating AWS Connect with Amazon Lex to understand voice input Part1

In this series of articles, we are going to discuss the below:

  • How to create a flow on Amazon Connect.
  • How to integrate it with Amazon Lex.
  • Sending Contact Trace Records to S3 using Kinesis Firehose.

HLD

Image description

Overview

Amazon Connect is service used to help you build your own contact center it’s a managed service by AWS which allow you to build your flows to handle customers’ requests & issues without the need of taking care of the infrastructure, we have used Amazon Connect in the setup to provide a contact flow to handle a customer with two categories (VIP & Golden) with two different queues for each category of customers, the flow is handling internet new requests & internet issues.

Amazon Connect is integrated with Amazon Lex which is a service allows you to create chat bots to handle customers requests and issues with bot which will be used later in the contact flow to recognize customer requests.

Amazon Lex is also a managed service used to build conversational bots which can be integrated by other services such as Lambda function,

Amazon Lex will be invoked in the contact flow using Invoke Lex block and the Lex chat bot created to recognize two types of requests, new service requests & internet down problems.

Design Components

Image description

Flow

Image description

Implementation Steps

Through AWS Console you will have to create an Amazon Connect Instance

I have created the instance with the below configuration for simplicity

Image description

Note: The access URL should be unique

Remember the Administrator credentials since it will be used later

Image description

Set your telephony settings in order to allow instance can recieve and initiate calls

Image description

Then we will configure our storage for call related information (recording & transcripts)

Image description

Then we are going to connect with the administrator credentials to our instance through the Access URL

Image description

You will have the below Dashboard

Image description

We will create a very simple flow in this article, then we may explore creation of more flow in future

Go to flows on the left as below

Image description

Then create flow and name it as sample flow

Image description

On the left we have call flow blocks, they are the main components of our IVR flow, we will go into more detail regarding these blocks later.

Image description

The type of this flow is inbound flow, which means it will receive incoming calls from customers

We added two blocks related to the logging,

Set Logging Behavior: When logging is enabled, data for each block in your flow is sent to Amazon CloudWatch Logs.

Set recording and analytics behavior: This Specifies recording behavior and configure Contact Lens conversational analytics

Image description

then we will add these two blocks,

Image description

The play prompt will be configured to welcome the customer

Image description

The other block is to check hours of operation, This block helps you to define the working hours of your queue

Image description

We will then use the below three blocks

Image description

Get customer input block: will receive an input from the user through the dial pad and based on the value it will be routed to the related queue

Set working queue block is defining current working queue to transfer the call

we will go into details later for creating queues, but I have selected the already created Basic Queue

Image description

the final block is the actual transfer to queue, here where the customer connected to the actual agent, after the customer request is fulfilled, the call should be terminated

After adding the terminate block, now you are ready to claim a number for your flow, but before moving forward you will have to save and publish your flow

for error in each flow block you will have to play prompt that "we are not able to process your request now" for better user experience

Image description

now go the dashboard then phone numbers in channel to claim a number

Image description

the difference between DID and toll free in a brief that the DID can be dialed from anywhere around the world, while the toll free can be dialed only from the country that you are claiming the number for, more details here.

channel will be voice, and you can choose any country just for testing purposes I have selected DID & UK

Image description

then select the flow that you are going to link this number with

Image description

Now you are ready to go and test your flow, I have used skype to dial and test this flow, in part two we are going to go into more details regarding voice recognition with Amazon Lex

Top comments (0)