Dr. Michael Zimmer, Chief Data Officer, Zurich Insurance, speaks with Marcus Hartmann, Partner at PwC Germany, in a video interview about value delivery in data and analytics with AI, his secret sauce for success, the need for value delivery, strategy and roadmap, and understanding the potential of generative AI.
Zurich Insurance is a German multi-line insurer serving people and businesses in more than 200 countries and territories. In addition to providing insurance protection, Zurich also offers prevention services that promote well-being and enhance climate resilience.
Zimmer calls himself a colorful data spot on a gray workday. Speaking on value delivery with regards to data and analytics, he says that AI is at a point where it can identify fraud and recoveries, based on the data. Unlike in the past, data can now be stored to deliver measurable value. He states that organizations need to bring AI to the businesses, show value, and grow together.
Further, Zimmer notes that insurance companies have huge amounts of unstructured data that were handled in the past by suppliers. He adds that the risk in that was the lack of labeled information in between.
The secret sauce to Zimmer’s success was starting to deliver quickly with AI. He asserts that it is important to generate value by storing raw labeled data in a structured way. This saves the company millions of dollars; it generates trust, and people are willing to change their processes as they see the impact on the company.
Most importantly, Zimmer says that the unstructured data is now stored in a modern AI platform aligned with GDPR, and this data opens up immense possibilities. Sharing one instance, he discusses how the team is working on claims values of cars by leveraging the stored training data.
Further, being a cost center, delivery is a must. In the insurance scenario, growth comes with the financing, and with 13 years of counseling experience generating benefit is in his DNA, says Zimmer.
Moving forward, he states that data is and needs to be a part of the business. In his role, Zimmer has worked with the business by helping centralize, bringing AI platforms together, raising the data lake use cases from 5 to 65, and increasing users from 10 to 170.
Moreover, he affirms rolling out the data governance concept and says that the lake and structures are now strategic. He urges people to keep in mind that delivery needs the business, the IT, and also some other stakeholders.
Also, the role of the CDO is to bring those worlds together and bridge them while showing the direction. According to Zimmer, value delivery is important as it creates trust.
When asked about the kinds of value created, Zimmer shares a situation where for cases of document extractions, other companies would charge 25 cents per page to 1 euro 50 cents, but his company did it for 4 cents.
Speaking on strategy and roadmap, Zimmer says that it is all about data strategy following the business strategy and ensuring that the strategies are aligned. He then discusses end-to-end digitization which is about AI services, automation, and bringing the systems together.
Furthermore, Zimmer discusses redesigning his existing strategy. To meet business goals, he has identified critical value drivers and relevant tasks, and on the other hand, there are AI topics that could be used as bread-and-butter business, he adds.
Shedding light on architecture, Zimmer believes that architecture cannot be changed every other day, and shares that the organizational data warehouse is 20 years old and still working. He asserts that the organization will not switch to a new technology just to have something new.
Commenting on the status of the organizational data lake, Zimmer says that he has started to build a new approach to lake house with silver data and gold data. He further says that the lake house is not completely based on vendors, but it boils down to combining the right things.
According to Zimmer, the data warehouse team must also become experts in the data lake. With about 80% of the functionality at hand, the right skill set is necessary, he adds. Zimmer stresses that with the business strategy, people, technology, and roadmaps must fit in. He also asserts bringing the compliance and business heads to the table for the data talk.
When asked about mechanisms to respond to requirements around Generative AI, Zimmer states that it is just another model that needs maintenance. Data leaders must realize its potential and where to use it, he says.
Thereafter, Zimmer admits that there are existing AI models that are doing summarizations and identifying fraud, and he is willing to run a model parallelly to assess benefits. Looking at the pricing model, he says that running everything in GPT will incur massive expenses.
Therefore, it is about ensuring that the right expectations are set. Commenting further on the hype of GPT, he notes that people are unaware of the legal limitations and governance aspects as they continue to use the public version of OpenAI with business data.
Concluding, he states that it is critical to understand what is possible and what is not with generative AI.
CDO Magazine appreciates Dr. Michael Zimmer for sharing his data insights and success stories with our global community.