The insurance industry is undergoing a transformative shift, driven by advancements in data analysis via artificial intelligence (AI). Among other areas, this trend involves creation of new products, and specifically ones involving climate risks (see Section 2). Hyper-personalized underwriting and pricing enabled by AI, makes it possible to tailor such policies to individual risk profiles.
And indeed, in our work with customers, we were involved in an initiative to create a new insurance product for commercial properties located in hazardous areas. These areas, previously deemed too risky to insure, due to their susceptibility to extreme weather events, are now becoming viable thanks to advanced AI-driven solutions.
The new insurance product targets commercial properties, particularly warehouses and manufacturing plants, located in northern European areas prone to natural disasters such as severe cold, floods, snowstorms, and hailstorms. In the past, our customers avoided insuring such assets. However, they recently concluded that AI-driven precise risk assessments provided by specialist vendors could enable the introduction of an attractive and profitable new insurance product tailored to such assets.
The key to this breakthrough lies in the combined utilization of multiple risk models based on machine learning. Yet, due to lack of the needed AI and data in-house, the project involves working with external vendors providing these risk models. Each vendor specializes in a certain geographic area or country: Norway, Sweden, and Finland, and a specific weather condition: wintertime severe cold and springtime floods. Each analyzes a comprehensive array of environmental, physical, and financial characteristics of assets, enabling precise underwriting and pricing. This information includes openly available satellite photos and historical weather figures that the vendor already has, as well as highly sensitive financial and proprietary data about the asset that should be provided to the vendor by the insurance company.
However, due to the confidentiality of the information involved, providing such sensitive data to the vendor, if at all possible, requires a long and expensive process of filtering out masking, or synthesizing sensitive elements. The data coming out of this process is often unusable due to its inaccuracy and incompleteness, rendering the whole process a waste of time and money. The only practical way to test and use the vendors’ models was then to install them on-prem. Again, an expensive task that may take years to complete.
This is exactly where Multyx comes into action. With Multyx’s sensitive data collaboration platform the insurance company does the following three steps quickly and efficiently.
First, the insurance company enabled its sensitive dataset for analysis in Multyx with no changes to the content needed by the AI analysis. Then, once approved, five different vendors boarded Multyx, and the insurer tested their models with the dataset to determine which ones to use in the project. Second, after three vendors were selected, the insurer enabled model training with the data. The trained models remained only inside Multyx for usage only by the insurer. Third, the three models were put into production to support underwriting and pricing of the new product.
Notably, across these three phases, none of the insurer sensitive data was shared with any of the vendors avoiding a need to approve such sharing or to alter the data before sharing it. Post training, the now sensitive AI models are also not installed on the insurer premises on the one hand, and are not given back to the vendor, on the other hand. These qualities mean quick time to market for the new insurance product, at a significant lower cost, due to skipping data alteration, on-premises installations, and the tedious legal approval process involved in sharing sensitive data and/or the trained model.
To summarize, launching this new insurance product line provides several advantages for the insurance company:
By leveraging Multyx the following could be achieved, which made the project practical:
In conclusion, this specific usage of Multyx by the insurance company exemplifies how Multyx facilitates taking advantage of externally-provided cutting-edge AI and machine learning to drive the development of new insurance products. With Multyx, you can confidently leverage the power of data and AI partnerships to enhance your insurance offerings easily and safely.
To learn more about facilitating the creation of new insurance products via secure data collaboration for leveraging AI and advanced analysis, and to start evaluating your options, contact us at Multyx.