Intact Financial Corporation is the leading provider of property and casualty insurance in Canada, a leading provider of specialty insurance globally, and a leader in commercial lines in the United Kingdom and Ireland. Intact faced a challenge in managing its extensive network of customer service call centers and needed a viable solution in 6 months and a long-term solution in 1 year. With up to 20,000 calls per day, the manual audit process was inefficient and struggled to keep up with growing call traffic and rising customer service expectations. QA agents had to manually select calls to audit, which was not a scalable solution. To address this, Intact turned to artificial intelligence and speech-to-text technology to gain valuable insights from calls and improve customer service. The company developed an automated solution called Call Quality (CQ) using artificial intelligence services from amazon Web Services (AWS). Implementing CQ enabled Intact to handle 1,500% more calls (15x more calls per auditor), reduce agent handle time by 10%, and generate valuable insights into agent behavior, leading to better customer service.
amazon Transcribe is a fully managed automatic speech recognition (ASR) service that helps developers add speech-to-text capabilities to applications. Use deep learning to convert audio to text quickly and accurately. In this post, we demonstrate how the CQ solution used amazon Transcribe and other AWS services to improve critical KPIs with ai-powered contact center call analytics and auditing.
This allowed Intact to accurately transcribe customer calls, train custom language models, simplify the call auditing process, and extract valuable customer insights more efficiently.
Solution Overview
Intact aimed to develop a cost-effective and efficient call analytics platform for its contact centers by using speech-to-text and machine learning technologies. The goal was to refine customer service scripts, provide training opportunities for agents, and improve call handling processes. By doing so, Intact hoped to improve agent efficiency, identify business opportunities, and analyze customer satisfaction, potential product issues, and training gaps. The following figure shows the architecture of the solution, which is described in the following sections.
Intact selected amazon Transcribe as its speech-to-text ai solution for its accuracy in handling both English and Canadian French. This was a key factor in Intact's decision, as the company was looking for a versatile platform capable of adapting to its diverse business needs. amazon Transcribe offers deep learning capabilities, which can handle a wide range of speech and acoustic features, and its scalability to process from a few hundred to more than tens of thousands of calls daily also played a critical role. Additionally, Intact was impressed that amazon Transcribe could adapt to various post-call analytics use cases across their organization.
Call processing and model service.
Intact has on-premises contact centers and cloud contact centers, so they created a call acquisition process to receive calls from both sources. The architecture incorporates a fully automated workflow, powered by amazon EventBridge, that triggers an AWS Step Functions workflow when an audio file is uploaded to a designated amazon Simple Storage Service (amazon S3) bucket. This serverless processing process is based on amazon Transcribe, which processes call recordings and converts them from speech to text. Notifications of processed transcripts are sent to an amazon Simple Queue Service (amazon SQS) queue, which helps decouple the architecture and resume the Step Functions state machine workflow. AWS Lambda is used in this architecture as a transcription processor to store the processed transcripts in an amazon OpenSearch Service table.
The call processing workflow uses custom machine learning (ML) models created by Intact running on amazon Fargate and amazon Elastic Compute Cloud (amazon EC2). The transcripts in OpenSearch are then further enriched with these custom machine learning models to perform component identification and provide valuable insights, such as named entity recognition, speaker role identification, sentiment analysis, and phrasing. of personally identifiable information (PII). Regular improvements to new and existing models added valuable information that can be extracted, such as call reason, script compliance, call outcome, and sentiment analysis across various business departments, from complaints to personal lines. amazon DynamoDB is used in this architecture to control queue limits. Call transcripts are then compressed from WAV to MP3 format to optimize storage costs in amazon S3.
Machine Learning Operations (MLOps)
Intact also created an automated MLOps pipeline that uses Step Functions, Lambda, and amazon S3. This channel provides self-service capabilities for data scientists to track machine learning experiments and push new models into an S3 bucket. It offers flexibility for data scientists to perform parallel deployments and capacity planning, allowing them to seamlessly switch between models for both production and experimentation purposes. Additionally, the application offers backend dashboards tailored to MLOps functionalities, ensuring seamless monitoring and optimization of machine learning models.
Interface and API
The CQ app offers a robust search interface purpose-built for call quality agents, equipping them with powerful auditing capabilities for call analysis. The backend of the app is powered by amazon OpenSearch Service for search functionality. The app also uses amazon Cognito to provide single sign-on for secure access. Lastly, Lambda functions are used for orchestration to retrieve dynamic content from OpenSearch.
The app offers customized trend dashboards to deliver actionable business insights, helping identify key areas where agents allocate their time. Using data from sources such as amazon S3 and Snowflake, Intact creates comprehensive business intelligence dashboards that display key performance metrics such as silent periods and call handling time. This capability allows call quality agents to delve deeper into call components, facilitating training opportunities for specific agents.
Call Quality Trends Dashboard
The following figure is an example of the Call Quality Trends Dashboard, showing the information available to agents. This includes the ability to filter based on multiple criteria, including dates and languages, average handle time by components and unit managers, and talk time versus silent time.
Results
The implementation of the new system has led to a significant increase in efficiency and productivity. There has been a 1,500% increase in audit speed and a 1,500% increase in the number of calls reviewed. Additionally, by building MLOps on AWS alongside the CQ solution, the team has reduced the delivery of new ML models to provide analytics from days to mere hours, making auditors 65% more efficient. This has also resulted in a 10% reduction in agents' time per call and a 10% reduction in average wait time as they receive targeted training to improve their conversations with customers. This efficiency has allowed for more effective use of auditors' time to design coaching strategies, improve scripts, and train agents.
Additionally, the solution has provided intangible benefits such as extremely high availability with no major downtime since 2020 and high cost predictability. The solution's modular design has also led to robust deployments, significantly reducing time for new releases to less than an hour. This has also contributed to a near-zero failure rate during deployment.
Conclusion
In conclusion, Intact Financial Corporation's implementation of CQ, powered by AWS ai services, has revolutionized its approach to customer service. This case study serves as a testament to the transformative power of artificial intelligence and speech-to-text technology to improve the efficiency and effectiveness of customer service. The design and capabilities of the solution position Intact well to use generative ai in future transcription projects. As next steps, Intact plans to continue using this technology by processing calls using amazon Transcribe streaming for real-time transcription and implementing a virtual agent to provide human agents with relevant information and recommended responses.
Intact Financial Corporation's journey is an example of how ai adoption can lead to significant improvements in service delivery and customer satisfaction. For customers looking to quickly get started on their call analytics journey, explore amazon Transcribe Call Analytics for live call analytics and agent support and post-call analytics.
About the authors
Etienne Brouillard is the Principal AWS ai Architect at Intact Financial Corporation, Canada's largest property and casualty insurance provider.
amy dani is a Senior Technical Program Manager at AWS focusing on ai/ML services. Throughout his career, he has focused on delivering transformative software development projects for the federal government and large enterprises in industries as diverse as advertising, entertainment, and finance. Ami has experience driving business growth, implementing innovative training programs, and successfully managing complex, high-impact projects.
Prabir Sekhri is a Senior Solutions Architect at AWS in the Enterprise Financial Services industry. During his career he has focused on digital transformation projects within large companies in industries as diverse as finance, multimedia, telecommunications as well as the energy and gas sectors. His experience includes DevOps, security, and enterprise storage solution design and architecture. Apart from technology, Prabir has always been passionate about playing music. He leads a jazz ensemble in Montreal as a pianist, composer and arranger.
