By Will Mitchell · September 9, 2026 ·
A member asked whether a smarter model makes compound engineering redundant. The short answer is no, but the honest answer is more useful: it is an opinionated workflow, it is wrong for research, and its value scales with how many people touch the codebase.
