This article explores the intersection of artificial intelligence (AI) and reconstruc-tive qualitative social research, examining how large language models (LLMs) can aid in interpreting empirical data, such as data from interviews or group discussions. The proposed method of Reconstructive Social Research Prompting (RSRP) aims to utilize specific prompts to encourage language models, such as GPT-4, Claude 3.5, or Mistral, to produce interpretations that align with the standards of reconstructive social research. In Chapter 2, the paper outlines the characteristic features of reconstructive social research, emphasizing its distinction from other qualitative methods and the significance of fundamental and subject-specific theories, methodologies, and methods. Chapter 3 introduces the development and testing of RSRP prompts, specifically social science prompts that guide a language model in performing methodical and methodologically sound interpretations. Subsequently, the concept of a modular prompt structure is present-ed, and the iterative process between researchers and language models to refine interpretations, termed Distributed Interpretation, is exemplified. In Chapter 4, the concept is elaborated further with an example from a current research project. This example illustrates the practical application of RSRP in analyzing group discussions and demonstrates how language models can generate meaningful interpretations. Finally, in Chapter 5, the article highlights the transformative potential of RSRP for qualitative research as a whole, arguing that the efficiency and depth of qualitative data analysis can be significantly enhanced. It suggests that through collaborative and iterative processes, LLMs are evolving into active partners in qualitative research, challenging traditional notions of intelligence, authorship, and interpretation.