Compound AI represents an advanced stage of artificial intelligence development, characterized by the integration and interaction of multiple AI models and systems to perform complex tasks. Analogous to the human brain, which comprises specialized centers like the cortexes and lobes working together, Compound AI systems are formed from a rich ecosystem of AI/ML components, including custom and off-the-shelf models, vector databases, retrieval systems, and coordination algorithms. Such systems are deeply interconnected, e.g., allowing outputs from one model to become inputs for another, creating adaptive and self-optimizing networks of data processing pipelines. In this talk, we take a closer look at Compound AI systems, particularly focusing on the opportunities and challenges related to building such novel systems.