Agentic evaluation remains massively under-resourced as a field.

→ View original post on X — @julien_c