View German version here.

INTRODUCTION

Due to the rising costs in claims management and the heavy workload of product managers handling multiple projects, our client – a well-established insurer in the German market – looked for ways to cut down on manual administrative work. One key area for improvement was the process of consolidating insurance terms and conditions (T&C), which evolve over time. Doing this manually is time-consuming, as it requires reading, analyzing and comparing content from documents that can be over 50 pages long. 

The main goal of our project was to significantly reduce the time required for T&C consolidation by using generative artificial intelligence (GenAI), while also providing a clear and accurate overview of the differences in T&C being consolidated.

WHY NOW?

AI and GenAI are being widely adopted across industries, particularly since the ChatGPT launch in December 2022. The insurance sector is using AI in various ways to get a sense of its capabilities, which raises the topic of efficiency gains through AI. GenAI, in particular, can rapidly process large amounts of text, offering great potential to significantly ease the workload during T&C comparisons in insurance firms. 

WHAT WE DID

Using GenAI, we developed a solution to reduce manual effort and efficiently compare about 20 T&C documents, each around 50 pages long and with varying structures developed over 20 years:

• Having identified the relevant criteria for comparison, we converted PDF-based T&Cs into a structured machine-readable format with the help of GenAI. 
  
• Using multi-layered prompts, we identified and analyzed each section with GenAI to assess critical coverage differences of the T&C documents being compared.
  
• As part of an iterative process, we reviewed the AI-generated evaluations and adjusted based on results, repeating until we achieved the desired quality. 
  
• Overall, the project took about four weeks, with the biggest efforts being the conversion and restructuring of PDF files and the refinement of GenAI prompts. A well-structured input greatly speeds up and improves the quality of GenAI's output.
  
• Essential for project success was the close collaboration between technical and subject-matter experts, e.g. ensuring that the prompts received the necessary contextual information while remaining machine-processable.

WHAT WE DELIVERED

We provided our client with an easy-to-follow, chapter-by-chapter summary of the results for all comparison criteria specified. Each section of the reference T&C documents was matched with the corresponding section of other T&C documents. The differences in content between these sections were rated (using ‘traffic light’ logic) according to their impact on the insurance customer.

Using this methodology, product managers do not need to go through countless pages of analysis. Instead, they can quickly use a GenAI-generated overview to focus on the sections marked ‘yellow’ or ‘red’, enabling them to find solutions faster and decide how the cover conditions should be consolidated.

The GenAI solution offers several benefits:

• Time savings: Reduces the need for reading and manual analysis.
  
• High precision: Ensures accurate AI-generated results without human fatigue.
  
• Consistent evaluation: Uses a structured ‘traffic light’ logic to assess differences.
  
• Detailed insights: Provides a strong foundation for making consolidation decisions and more.

The same logic can be applied to compare conditions across other insurance product lines. Additionally, this technical approach can be adapted for other use cases, such as aligning contracts with internal policies or assessing the integration of regulatory requirements within documents.

This project also showcases how the collaboration between Capco’s client-facing consultants and internal GenAI capability can help significantly automate processes and deliver greater and more competitive value to our clients, using a GenAI-infused approach.