Challenge

A global financial services leader provides insurance, mutual funds, annuities, and other investment solutions for their clients across the globe. As part of an ongoing strategy to improve customer satisfaction, the company dedicates significant time and resources to thoroughly analyze and review customer complaints and respond accordingly. Over the years, this initiative resulted in significant reductions in customer churn as customers feel heard and satisfied with the intimate level of service that they receive.

While the high-touch service improves customer satisfaction and retention, it is also costly. Customer service specialists spend significant time and manual effort to most effectively respond to customer requests. To service a single inquiry, they need to go through multiple systems and databases to understand the specific customer request, review the customer service history, find relevant product information, and prepare a personal response.

Approach

The financial service company looked to leverage generative AI to simplify information access and accelerate service cycles and chose C3 Generative AI and AWS Cloud due to its domain-specific architecture, high response accuracy, and accelerated deployment timeline.

A joint team of C3 AI developers and company subject matter experts configured C3 Generative AI in less than 2 days and made available to an initial group of customer service specialists. The application was deployed on AWS and leverages Amazon’s services including Amazon Bedrock.

During the 2-day deployment process, the team uploaded and indexed over 180 documents, fine-tuned search performance, and added custom prompts to draft responses. The team also conducted various quality tests to ensure optimal output behavior in responses. The preliminary tests suggested the system to be robust and professional in its outputs even if a customer service representative should provide toxic content, and that the system is not easily coerced, is robust to simple adversarial attacks, and consistently recommends appropriate action.

In the first week of its roll-out, C3 Generative AI was extensively used by the customer service specialists with a total of over 150 user queries. Using C3 Generative AI, the company shortened the time required to find information by over 90% enabling their customer service specialists to instead focus on critical consultation and review and resolve more cases.

About the Financial Services Company

• GBP4 billion revenue in 2022
• 17,000 employees as of 2022
• 16 million customers
• Lead provider of financial products in the UK

Objectives

• Integrate customer service inquiries, historical responses, and product manuals on a unified knowledge source
• Configure results performance via no-code prompt engineering
• Add initial group of customer service users
• Roll-out to an initial group of customer service agents