The text discusses the evolution of AI systems for code reviews, highlighting the transition from traditional static rules to sophisticated systems that offer feedback akin to a senior engineer. The key design choice in developing such AI systems is whether to implement them as agentic AI, which allows for autonomous decision-making, or as pipeline AI, which offers more predictability and control. The distinction impacts processing speed, reliability, and operational costs. Many modern AI systems adopt a hybrid approach, combining pipeline stability with agentic flexibility to create practical solutions. The text emphasizes that the focus should not be on whether to choose agentic or pipeline AI, but rather on providing the model with the right information in the right way. Additionally, the importance of open source in software development is acknowledged, as well as the challenges of code reviews without sufficient context. The ultimate goal is to develop AI that can deliver reliable, context-aware feedback, similar to a highly skilled engineer, in an automated fashion.