loading

Business Impact

  • Improved Average Handling Time

  • Reduced Manual Effort

  • Improved Caller Experience

Customer Key Facts

  • Location : North America
  • Industry : Healthcare

Problem Context

The customer provides management and support services to healthcare institutions across the United States. Their call center receives over 200k calls each month, covering close to 1 million minutes of conversation between the agents and customers. The agents are the single point of contact between healthcare institutes and customers, and must manually take notes on customers’ personal information, etc. This task is cumbersome and subject to error.

 

Challenges

 

  • High call volumes – 200,000+ calls per month
  • Agents had to take notes manually making the process cumbersome
  • Lack of important/relevant information captured from calls
  • Long query resolution process led to more wait times and customer dissatisfaction
Challenges

Technologies Used

Amazon S3
Amazon DynamoDB
Amazon Fargate
Amazon Transcribe
Amazon Comprehend
Amazon Kinesis Video Streams
Amazon Chime Voice Connector
Amazon API Gateway

Leveraging Machine Learning to Transcribe Calls and Perform Real-Time Call Analytics

The customer wanted to provide a seamless experience for their call center agents and customers. Therefore, they sought to leverage machine learning to transcribe calls and perform real-time call analytics.

Solution

Quantiphi built a solution with real-time speech-to-text and Natural Language Processing (NLP) capabilities that assists call center agents with their day-to-day interactions with clients. The solution was integrated with their existing CISCO telephony systems and transcribes each conversation between customers and the agents. The script transcription is then used to extract entities like name, phone number, location. etc. At the end of the call, post call analysis is performed to summarize the call and understand the sentiments of the speakers. A custom User Interface was also built to display all the information captured, such as caller information, relevant keywords, and call summary.

Result

  • Assists human agents with caller identification, real-time keyword analysis, and custom entity extraction to respond quickly to customer requests
  • Allows agents to focus on more complex aspects of their interactions with callers by reducing the cognitive overhead and enabling higher accuracy

Looking for similar project?

Let's Talk

Get your digital transformation started

Let's Talk