Use the Amazon Connect integration to connect PolyAI voice agents to your AWS ecosystem. Utilize Amazon Connect contact flows for dynamic call routing and DynamoDB and Lambda for real-time data processing.

​
Capabilities

​
Call handling and analytics

PolyAI voice agents manage inbound calls routed via Amazon Connect. Agents capture customer intent, execute knowledge-based responses, and reduce the load on live agents. Call performance, customer satisfaction, and agent efficiency are tracked using Amazon Connect analytics, integrated with PolyAI’s conversation data.

​
Live agent handoff

Calls requiring human assistance are routed to Amazon Connect agents with full contextual data for continuity and personalized support.

​
Secure data access

PolyAI securely retrieves and updates customer data using AWS STS and Amazon DynamoDB.

​
How the integration works

  1. Inbound call handling: Amazon Connect selects a number from a pool of PolyAI-provisioned phone numbers for incoming calls.
  2. Routing to PolyAI: Calls are transferred to the PolyAI conversational assistant, which retrieves contact attributes from DynamoDB using an AWS IAM role with appropriate permissions.
  3. Guided conversational flows: The PolyAI assistant greets the caller and executes customized conversational flows, such as identity verification or information collection.
  4. Handoff or resolution: If the query is resolved, the call is terminated. Unresolved queries are routed back to Amazon Connect for queue-based agent handling.
  5. Data management: Data from the interaction (e.g., transcripts, query status) is sent to Amazon Connect and stored in S3 buckets using the same AWS IAM role.

​
Setup guide

​
Prerequisites

  • Amazon Connect instance: Ensure an active Amazon Connect instance with administrative access.
  • PolyAI project: Set up a PolyAI assistant aligned with your Amazon Connect workflows.
  • AWS services:
    • DynamoDB: For securely storing call attributes.
    • Lambda: For processing event data.
    • STS: For secure role-based data access.
  • Integration credentials:
    • IAM role: Create an IAM role with policies for DynamoDB and Lambda access, allowing PolyAI to interact with AWS resources.

​
1: Configure Amazon Connect

  1. Log in to the AWS Management Console and open Amazon Connect.
  2. Use the Flow Designer to:
    • Create or edit a contact flow.
    • Add:
      • A PolyAI handoff node to route calls to the voice assistant.
      • A Lambda function node to retrieve and process call data.
  3. Save and publish the contact flow.

​
2: Set up DynamoDB for call attributes

  1. Create a DynamoDB table:
    • Define a primary key, such as CallID or CustomerID.
    • Add attributes like customer_name or reservation_number.
  2. Link the table to the Lambda function for real-time updates.

​
3: Configure AWS Lambda function

  1. Create a Lambda function in the AWS Management Console:
    • Use a provided or custom script to process call events.
  2. Assign an IAM role with:
    • Permissions for DynamoDB and STS.
  3. Deploy the function and link it to the Amazon Connect contact flow.

Example Lambda function template:

import boto3

def lambda_handler(event, context):
    dynamodb = boto3.resource('dynamodb')
    table = dynamodb.Table('YourDynamoDBTable')

    # Extract call attributes
    call_id = event['Details']['ContactData']['ContactId']
    caller_number = event['Details']['ContactData']['CustomerEndpoint']['Address']

    # Update or retrieve data
    response = table.update_item(
        Key={'CallID': call_id},
        UpdateExpression="SET caller_number =:val",
        ExpressionAttributeValues={':val': caller_number}
    )

    return {
        'statusCode': 200,
        'body': 'Call attributes updated successfully'
    }

​
4: Integrate PolyAI assistant

  1. Provide PolyAI with:
  2. Set up the integration using AWS CloudFormation:
    • PolyAI will share an AWS CloudFormation template using Amazon S3.
    • Create a service role in your AWS account with the following policy, with <YOUR_ACCOUNT_ID> replaced by your actual 12-digit AWS Account ID:
{
   "Version": "2012-10-17",
   "Statement": [
       {
           "Sid": "VisualEditor0",
           "Effect": "Allow",
           "Action": [
               "cloudformation:*"
           ],
           "Resource": "*"
       },
       {
           "Sid": "VisualEditor1",
           "Effect": "Allow",
           "Action": "s3:*",
           "Resource": "arn:aws:s3:::amazon-connect-integration-sample-code/<YOUR_ACCOUNT_ID>/*"
       },
       {
           "Sid": "VisualEditor2",
           "Effect": "Allow",
           "Action": "lambda:*",
           "Resource": "arn:aws:lambda:*:<YOUR_ACCOUNT_ID>:function:PolyAI*"
       },
       {
           "Sid": "VisualEditor3",
           "Effect": "Allow",
           "Action": "iam:*",
           "Resource": [
               "arn:aws:iam::<YOUR_ACCOUNT_ID>:policy/PolyAI*",
               "arn:aws:iam::<YOUR_ACCOUNT_ID>:role/PolyAI*"
           ]
       },
       {
           "Sid": "VisualEditor4",
           "Effect": "Allow",
           "Action": "dynamodb:*",
           "Resource": "arn:aws:dynamodb:*:<YOUR_ACCOUNT_ID>:table/PolyAI*"
       }
   ]
}
  1. Deploy the PolyAI CloudFormation stack, which will automatically create:
    • A DynamoDB table.
    • A Lambda function.
    • The required IAM roles.
  2. Update your Amazon Connect flow to route calls to PolyAI voice agents.

​
5: Test and deploy

  1. Test integration:
    • Verify everything works, like:
      • Try routing a call to various PolyAI assistants.
      • Make dummy data retrieval and status updates to test DynamoDB.
      • Simulate a live-agent handoff processes to make sure there are no problems.
  2. Monitor analytics:
    • Use the Amazon Connect analytics dashboard to track call volume, routing efficiency, and agent handoffs.

For support, contact PolyAI at support@polyai.com

OverviewFive9