Intact Financial improves critical KPIs with AI-powered call auditing and analytics | AWS Events

Introduction

This tutorial outlines how Intact Financial enhanced its call auditing and analytics processes using AI, specifically through Amazon Transcribe and serverless architecture. By implementing these technologies, Intact significantly improved its key performance indicators, including a 15x increase in call auditing efficiency and a 10% reduction in agent handle time. This guide will walk you through the steps to create a similar solution for your organization.

Step 1: Understand Your Call Auditing Needs

• Assess current call auditing processes and identify inefficiencies.
• Determine key performance indicators (KPIs) that need improvement, such as:
  • Call auditing frequency
  • Auditor efficiency
  • Agent handle time
• Engage with stakeholders to gather insights about challenges faced in call handling and auditing.

Step 2: Explore AI Technologies

• Research AI-powered solutions that can streamline call analysis.
• Focus on Amazon Transcribe for speech-to-text capabilities, which converts audio recordings into text.
• Consider serverless architecture for scalability and cost-effectiveness:
  • Look into AWS Lambda for running code in response to events without managing servers.
  • Explore other AWS services that can complement your solution, such as Amazon S3 for storage.

Step 3: Design Your Analytics Solution

• Draft an architecture diagram that outlines how components will interact:
  • Use Amazon Transcribe to transcribe audio calls.
  • Store transcripts in Amazon S3.
  • Use AWS Lambda to trigger analytics processes on new transcripts.
• Define workflows for how data will move through the system and be analyzed.

Step 4: Implement the Solution

• Set up Amazon Transcribe:

  • Create a transcription job using the AWS Management Console or AWS SDKs.

• Configure AWS Lambda:

  • Write a Lambda function to analyze transcripts and extract valuable insights.
  • Use the following code snippet as a starting point:

  import json
  
  def lambda_handler(event, context):
      # Retrieve the transcription result
      transcription = event['Records'][0]['Sns']['Message']
      # Perform analysis on the transcription
      results = analyze_transcription(transcription)
      return {
          'statusCode': 200,
          'body': json.dumps(results)
      }
  

• Store the analysis results in a database or dashboard for easy access and review.

Step 5: Monitor and Optimize Performance

• Regularly review the effectiveness of the call auditing process.
• Track improvements in KPIs:
  • Monitor call auditing frequency and auditor efficiency.
  • Analyze reductions in agent handle time.
• Adjust the solution based on feedback from agents and auditors to continuously improve performance.

Conclusion

By following these steps, organizations can leverage AI technologies to enhance call auditing and analytics. With tools like Amazon Transcribe and serverless architecture, significant improvements in efficiency and performance can be achieved. Consider starting with a pilot project to test the solution and gather insights before a full-scale implementation. Keep iterating on the process to ensure ongoing improvement and adaptation to changing needs.