Automated service composition is still an open field of research. A major challenge poses the connection of semantic and syntactic information. Recent advancements in AI such as large language models brought about new directions of study. In this work we explore the combination of two new types of AI models, namely Large Reasoning Models (LRMs) and Large Action Models (LAMs). LRMs excel in reasoning and logical thinking and may bridge the gap between natural language requirements and the syntactical service descriptions. LAMs are trained to interact with the environment and execute plans, therefore they are suitable for composition execution. We developed a system architecture and prototype combining the two model types to solve automated service composition tasks. We extended an existing benchmark with additional domains to increase diversity. Using the prototype we conducted experiments with different LRMs to evaluate the performance of the new approach. The results revealed performance differences between the used models and domains. Overall the prototype was able to solve given tasks but further research is required to improve consistency and efficiency. This work is a first step in researching the combined capabilities of LRMs and LAMs.