Between Hope and Implementation – Generative AI Promises Efficiency

From chatbots to claims processing: the use of generative AI promises efficiency and new customer experiences. But the road to this is paved with cultural hurdles, data protection issues and a lack of strategic anchoring .

During an expert lunch led by Dr. Marcel Thom, Insurance Lead Partner, Deloitte, representatives from ERV, Amazon Web Services and Deloitte discussed the status quo of GenAI in the Swiss insurance industry. Walter Wattinger, CEO of European Travel Insurance ERV, stated: “Swiss insurers have definitely caught up in the last two years.” His company relies on AI in claims processing, while others have introduced chatbots, for example. Switzerland is proceeding more cautiously than Anglo-Saxon markets when it comes to data protection and quality requirements. “That’s not a disadvantage,” says Wattinger, “but now we need to speed up implementation.”

Christian Richter from AWS also confirmed: “Data protection is rightly very high here. But we are now seeing major investments in local AI infrastructures.” However, there is still a lack of productive implementation in the core processes. “Many use cases get stuck in customer interaction, whereas the greatest potential lies in claims, underwriting and sales.”

Strategic or tactical? Two speeds

Madan Sathe, strategy consultant at Deloitte, sees a dichotomy in the Swiss market: around five insurers are thinking strategically, with a clear AI vision, responsibility and a KPI-based approach. The others have individual pilot projects, but are acting reactively. “AI is a transformation project, not a sprint. If you don’t embed it in the corporate strategy, you’ll get stuck in the experimental phase.”

International insurers are further along: “They are building generalized AI factories, collecting hundreds of use cases and extracting common functionalities from them. Switzerland is only just getting started.”

Claims processing use case: real-time meets efficiency

The example of ERV shows how concrete benefits arise. “We automate the classification and checking of claims, including direct payment,” explains Wattinger. Initial results: massively shorter processing times and an extraction quality of over 90 percent. According to Wattinger, it is important that GenAI helps employees: “We want to automate repetitive tasks so that people can concentrate on complex cases.”

The biggest hurdles: Data, culture, trust

Data quality remains a key issue. “No AI without clean data,” said Sathe. Christian Richter added: “We are seeing a trend towards hybrid solutions with cloud and in-house models. Open source approaches and Swiss LLMs such as SwissGPT are becoming increasingly important.”

Culturally, the change is just as challenging: “Many older employees don’t dare to ask if they don’t understand something. You have to pick them up in a targeted way,” explained Binci Heeb, Editor-in-Chief of thebrokernews, who offers training courses for older employees herself. Richter confirmed: “The war for talent is becoming the war for senior talent. The experience of older employees is actually enhanced by GenAI.”

What counts now: Speed, talent, trust

Wattinger summed it up succinctly at the end: “We now need speed, talent and trust.” Practical implementation, employee training and an open dialog about benefits, risks and limitations are needed. After all, AI can only develop its full potential if it is seen as an integrated part of the value chain.

The Swiss insurance industry is on the cusp of digital transformation. GenAI provides the tools. Now we need the courage to implement it.

Binci Heeb

Read also: Regulation under pressure (1st article from HZ Insurance Forum 2025)