Go back

From 3-5 minutes to 30 seconds when searching for information in customer service.

Markel’s goal was to modernize their internal support team processes with Gen AI, aiming to enhance customer experience while also ensuring the safe and secure deployment of AI.

From 3-5 minutes to 30 seconds when searching for information in customer service.

Markel Insurance SE

Markel Insurance SE, based in Germany, specializes in commercial liability and professional risks, offering products like professional liability insurance and cyber attack insurance for freelancers and small to medium enterprises.

The Problem - solving increasing customer service challenges

Markel’s 1st Level Support team faces common challenges on a daily basis and were keen to utilize AI to overcome the challenges.
Spending an average of 3-minutes per request using CTRL+F - Like many insurers, Markel handles frequent customer inquiries, often requiring manual search through detailed documents, which can take up to 3 minutes for each customer response.
Frequently required the support from 2nd Level Support and Senior Underwriting - Due to the complexity of communicating relevant information, customer inquiries are frequently escalated to specialists, leading to additional delays in responding to customers.
Unsure how to safely adopt AI connected to company data that scales with business growth - Eager to embrace cutting-edge technology yet cautious about the potential risks of linking sensitive data with large language models like ChatGPT, Markel needed specialized guidance.

The Solution - instant knowledge access with GenAI

Now Markel's support team can ask a Gen AI assistant to search for the information and generate the answer.

AI-assistant for 1st Level Support

An advanced GenAI assistant that retrieves factual information and generates accurate, intent-aware responses, helping support members to get answers for questions within seconds instead of minutes of manual research.

Markel’s data made AI ready

Our data engine, refinery, securely stores, processes, and refines Markel's detailed support documents. This enables the GenAI assistant, using Microsoft Azure ChatGPT, to securely retrieve accurate answers. Markel's support documents, typically complex PDFs, average 65 pages in length.

Leveraged advanced retrieval

We enhanced GPT's capabilities by integrating a custom query-intent strategy with our advanced Retrieval-Augmented Generation (RAG) platform, cognition, thereby improving response quality and relevance.

The Results - going from minutes to seconds in researching answers

  1. Accurate, fact-based AI responses that Markel trust
  2. Assisted internal support teams & quicker customer responses
  3. Company knowledge made instantly accessible

"Kern AI has been an invaluable partner for adopting LLMs."

Our journey together started seamlessly with our first case, and collaborating with the Kern team has been a breeze. We look forward to taking on more use cases together and demonstrating the remarkable potential of LLM adoption in the insurance industry

Christian Parschik
Head of Digital Underwriting

Awards & Recognition

We're excited to share that our successful collaboration with Markel Insurance SE has led to two prestigious awards: the Collaboration Champion Award from InsurTech Hub Munich and the ITC Europe Digital Insurance Agenda Diamond Award.
This honor is a testament to our commitment to excellence and innovation in the insurance industry.
This achievement highlights our team's expertise in rapidly transforming an initial prototype into a fully operational, value-adding system. This quick and efficient development played a crucial role in the project's success, greatly improving Markel's workflow optimization and service quality in an impressively short time.