Situation
Dutch healthcare insurance provider CZ’s customer service department was facing a burgeoning volume of calls. On average, the company handles around 6,000 customer conversations per day, with a significant increase during the end-of-year period. The majority of the calls tend to be driven by straightforward questions that can easily be resolved through self-service, such as ‘how many physiotherapy sessions am I entitled to’ and ‘how many have I already used’? Or ‘will my policy cover my next consultant’s appointment’?
CZ is always looking for ways to improve its customer experience, so it wanted to explore options to help decrease waiting times and smooth customer interactions. It also wanted to help improve the employee experience by decreasing the volume of calls its employees had to deal with and freeing them up to focus on more urgent, complex customer queries.
To help find a solution, CZ partnered with Valcon due to our expertise in AI and natural language processing (NLP) solutions. In addition to our technical expertise in NLP, CZ was keen to work with a partner that has a proven track record in change management. Effective change management is essential to secure adoption, embed new processes and achieve sustainable success with AI projects.
Approach
With the rapid evolution of GenAI, Valcon used its knowledge and proficiency in AI to ensure CZ could benefit from the most recent advancements. The solution followed a robust AI governance approach, prioritising safety, compliance and data privacy, and was structured to meet regulatory and organisational requirements. This approach resulted in the creation of a GenAI competence centre at CZ, which would help the insurer meet its customer service objectives. The approach had three main steps:
- Business analysis: to identify areas where GenAI could deliver the most business value, Valcon started by making a careful analysis of use cases in the customer service department. We found the majority of calls were about the same topics: reimbursements (19%), claims (18%) and payments (15%). So we focused on developing GenAI assistants that could help answer these ‘high impact topics’.
- Solution development: for each high-impact topic, we developed a GenAI assistant to classify and respond to a wide range of questions by using a large language model (LLM). The LLM can also summarise policy conditions and healthcare usage in an accessible language (Dutch B1). To measure the accuracy of the assistant’s answers, we collected a test set of questions and answers annotated by customer service employees. With natural language processing techniques, we continuously monitor the performance of the assistant based on the test set, which helps us to prevent any hallucinations (faulty responses from GenAI) being sent to customers.
- Deployment in mobile app: we gradually deployed the GenAI assistants on CZ’s mobile app, starting at small scale with a minimum viable product (MVP), which consisted of classifying customer questions according to the correct policy terms, and ending with the successful live launch of all GenAI assistants. The assistants enabled customers to access information quickly and efficiently via their mobile app, empowering them to find answers independently and reducing their reliance on customer service.
Results
The introduction of the GenAI assistants has brought several key benefits to CZ’s customer service department, including:
- High customer satisfaction: 96% of customers who receive a response from the GenAI assistant are satisfied and do not escalate their queries to a live chat representative.
- Reduced call volume: there has been a significant decrease in calls related to reimbursements, claims and payments, allowing customer service agents to focus on more complex issues. The reduction in call volume led to shorter waiting times and decreased operational costs.
- Improved job satisfaction: with lower-level queries being dealt with by the Gen-AI assistants, employees have experienced lower stress levels, which has contributed to a more positive work environment.
- Increased self-service adoption: over 1 million queries have been handled by the GenAI assistants, empowering customers to find answers independently.
Looking ahead, the Valcon and CZ team are evolving the GenAI assistants into an agentic AI framework. This will enable the assistants to combine information across topics to deliver precise, comprehensive answers to customer questions.
Would you like to speak to someone at Valcon about how we can transform your customer service with GenAI? Get in touch with: [email protected] and [email protected]
