Case Study

How AlgoLib works with a UK InsurTech company use AI to design new insurance products

Recently, AlgoLib LTD produced a set of quantitative solution services to a UK InsureTech company. The company wants to create and launch a new InsureTech product by applying machine learning algorithms to a huge set of market data; they hope that, by employing new technologies, they can underwrite more suitable contracts to their clients.

The client approached us with a description of the product desired and corresponding specifications of input data and output expectations. After a quick one-month preparation step, AlgoLib’s quantitative solutions team finished the prototypes within just two months by creating algorithms from AndromedaTM ESG and Recurrent Neural Networks. C++ powered applications with documentations are delivered at this stage. After several exchanges of tech reports and ideas between teams on both sides, the client agreed on the final version of the prototype during the fourth month of the project. AlgoLib delivered the R&D results with full source codes in Matlab and C++. A tech report and a paper fully describing the algorithms have also been delivered, with the agreed intellectual property transferred.

At the second stage of deployment, the client contracted AlgoLib to create an online application that realises all the functionalities that AlgoLib had developed in the R&D stage. The web application uses AlgoLib’s cloud computing units to provide real-time insurance quotes to the client and analyses insurance portfolio risks using data visualisation and machine learning algorithms. The project took four months to finish and the final deliverables are a fully functional business intelligence platform tailored for this insurance company to sell, manage, monitor and analyse all newly underwritten products.

The client is satisfied with the product that we have created and is interested to work with AlgoLib on upgrades to the platform soon when more data becomes available from sales. Considerations to let AlgoLib provide long-term support for the software have been under discussion.