Compound AI applications, which leverage Python to integrate machine learning models, are hindered by significant end-to-end latency issues. This latency is primarily caused by external components that cannot be optimized using traditional methods. Researchers have introduced PopPy, a novel approach that opportunistically exploits parallelism in these applications to mitigate latency. By doing so, PopPy aims to improve the overall performance of compound AI systems. The approach focuses on optimizing the execution time of external components, which dominate the overall latency. This is particularly important as AI applications are increasingly being used in critical domains, such as software engineering and enterprise automation1. The implications of PopPy extend beyond technical optimization, as improved performance can have significant impacts on policy, security, and workforce dynamics. Therefore, the development of PopPy matters to practitioners as it has the potential to enhance the efficiency and reliability of AI-driven systems.