Learn and transfer actuarial functions from source to target systems  Millions of life insurance contracts around the world run on outdated IT systems that make them cumbersome and expensive to manage. This is being met by challenging economic and regulatory conditions and rapidly advancing digitalization. A comprehensive modernization of the most important core systems includes the transfer of millions of policies and the reimplementation of the associated actuarial functions. A so-called migration from one or more source systems to a target system is required. Such a migration for a medium-sized portfolio today takes 2-3 years, costs 20-30 million euros and ties up valuable resources. This presentation shows the current possibilities to automate this process, at least partially, with the help of AI. The focus is on replacing manual activities with machine learning. It is an industrial approach that automates the creation of essential actuarial functions of the targeted policy administration system and brings them into operational use. We use Deep Neural Networks, Symbolic Regression, Neural Trees and XAI and combine this with domain knowledge to learn and explain the actuarial functions with sufficient quality and AutoML to optimize and automate the process.