This working paper provides a detailed overview on implementation strategies, critical evaluation, and best practices to use generative AI for research and analysis in historical studies. It distinguishes between two key applications of AI in historical research: as a supplementary tool and as a methodological component. The paper further emphasizes a responsible, purposeful, and demand-oriented approach with generative AI, which is guided by the principle of minimal computing. It outlines various application scenarios and provides clear frameworks for evaluating when and how to implement AI technology, addressing both individual and institutional responsibilities in order to maintain scientific integrity and adherence to data security and privacy. The aim is to maximize research impact and to maintain space for creative exploration and breakthrough innovation, while balancing performance and reliability needs with ethical considerations and sustainability.