The Myth Costs You More Than the Truth
The "no irreplaceable people" platitude exists to make org charts feel stable. Luu's evidence says otherwise: specific engineers, researchers, and decision-makers show up again and again at the hinge points of consequential outcomes. In an agent-heavy company, this effect concentrates rather than dilutes. Agents can own reversible, well-specified work, but the humans who remain are setting goals, making tradeoffs, and owning accountability for anything irreversible. Pretending those humans are fungible is not humility — it is a operational miscalculation.
Judgment Is the Scarce Input Now
When agents handle the broad, measurable base of execution, the narrow consequential tip of the pyramid belongs entirely to human judgment — the calls you would want to defend in a room with a customer. Luu's framing suggests that the individuals who carry genuine taste, context, and the ability to read ambiguous situations are not overhead; they are the system's most load-bearing component. Finding and keeping one such person compounds in ways that adding more agents cannot replicate.
Design Your Org to Surface Them
Luu's argument implies that most organizations are structured to obscure individual contribution rather than locate it. For a founder, the translation is concrete: build evaluation and feedback loops that make individual impact legible, not just aggregate output. Match your approval layers to stakes, so high-judgment people spend their time on high-stakes decisions rather than reviewing typo fixes. An org that cannot see where one person's judgment is doing the real work cannot protect that asset, retain it, or replace it thoughtfully when it leaves.
