Dr. Michael Zimmer, Chief Data Officer, Zurich Insurance, speaks with Marcus Hartmann, Partner at PwC Germany, in a video interview about governance in cloud-based platforms, organizational data processing, in-house talent upscaling, data governance and quality, the need for hub and spoke, and usability of generative AI from a governance standpoint.
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.
Having cloud-based platforms is a game changer because of functionalities, but organizations must govern the cloud platforms with LLMs to reduce risks, says Zimmer. He recalls being a part of old data platforms that could be deployed with a push of the button from development to test to production without changing any code.
According to Zimmer, platforms must have integration patterns and frameworks for management. He states that platforms generally are powerful but lack standards. Therefore, Zimmer affirms investing in structures to make them stable, automate them, and have integration patterns.
From a data platform perspective, having cognitive services that lead to better quality data rules based on AI interests Zimmer. He aims to use LLMs that can speak with data.
Further, Zimmer believes data processing is crucial and that is where the magic lies. Organizations need the right data and its ownership, and a successful company will have its data and provide it in high quality, therefore data platforms are critical, he adds.
However, Zimmer insists that the organization must ensure it has the right skills for people. He recommends organizations invest in the skill of in-house talent and take it to the next level.
As a strategy, rather than going after the next hype, or next solution, one must listen to the data experts as there are patterns that have always been relevant for 20 years, says Zimmer. He assures that the data experts may not know the technology but know what went wrong in the last 20 years.
When asked about handling data governance and quality, Zimmer mentions having group data owners responsible for certain domains at Zurich. In addition, he discusses defining a new policy for data management in Germany that could be well understood by people.
Elaborating further, Zimmer states that the policy has descriptions of what the roles are and what one needs to do. He mentions starting with the finance and fin-ops domain to bring people to the data cataloging part and keep processes running.
Furthermore, Zimmer mentions having meetings where people showed what they did and how they defined data quality rules and measured it. In continuation, he says that the data quality metrics are there in Zurich and the existing warehouse is good.
Also, the organization works on expert judgments that are based on defined rules, but Zimmer believes that taking people along is an essential part of governance.
An organization needs a hub and spoke when it comes to data, he affirms. There must be centralized platforms, IT experts, and business owners who also understand data management. Zimmer states that it boils down to bringing all the roles together and having each role understand the need and importance of the other, and how the problems they face are similar.
For instance, Zimmer stresses that business needs to understand why certain standards exist and what policies they must apply to IT. Similarly, IT must learn that business people are good programmers.
Moving forward, he states that with this mindset shift, everyone can work together, and take ownership of data. He says that data quality in the end is an outcome.
Highlighting data governance, Zimmer shares that as a CDO, he has agreements with IT Security, GDPR, compliance, and audit functions, but he has ownership of the non-operational system.
Additionally, Zimmer holds the responsibility to ensure that the organizational data catalog is used for other topics too. Further, there is an aligned toolset for structural integration, as he believes that consistency is the key.
Commenting on the usability of Generative AI from a governance standpoint, Zimmer talks about building a “group governance” together. He affirms that there is an ethical AI framework being followed, where actions are taken based on how LLMs are used.
Besides this, there are some group guidelines and quality gates, adds Zimmer. He mentions being happy with the structure, whereas sometimes there is a limitation around deploying new tools.
Referring to a chatbot as an AI tool, Zimmer states that he wants to generate benefits out of it, but proper heed must be paid before bringing LLMs into existing solutions. He maintains that it is powerful for certain use cases like input management and summarization but he would not use it for key business areas till there are stable versions where hallucination is not the standard.
In conclusion, Zimmer says that as an industry, the insurance sector is different and highly regulated. Nevertheless, he has numerous cases up his sleeve to generate value out of AI for customers and agents. Zimmer says that the hope lies in the capability of the model and in core processes and not in the fancy hype and chatbots.
CDO Magazine appreciates Dr. Michael Zimmer for sharing his data insights and success stories with our global community.