Every team building generative AI hits the same fork eventually: keep pushing one model to do everything, or start splitting the work across specialized components. Neither answer is obviously right, and the teams getting it wrong are finding out the expensive way. This session brings together practitioners who have landed on different sides of that decision, covering when compound architectures actually improve reliability and when they just add complexity and cost without a real payoff. Expect disagreement on where the line sits. What this session will cover: - When splitting a system into specialized components improves reliability, and when it does not - How teams are deciding between one capable model and several coordinated ones - The hidden costs of compound systems that do not show up until production - Real tradeoffs teams have made, including ones they would reverse If you are mid-argument internally about whether to split your system into specialized pieces, this panel gives you both sides of that fight from people who have actually shipped it.