Agentic engineering

AI agents do the implementation; the human steers with high-level vision, problem statements, and taste. dhh hates the term ("marketing slop") and argues it should eventually just be called programming — in English, "the one programming language more beautiful than Ruby."

Key practices: under-specify and iterate (agile insight applied to prompting), differential evaluation (pick between options rather than specify), cross-model review (one frontier model implements, another reviews), parallel threads to stay in flow.

Counterpoint workflow: matt-pocock's AI Engineer workshop front-loads alignment (grilling-session) and then removes the human from implementation entirely (human-in-loop-vs-afk via Kanban DAG + Ralph loops), where DHH stays conversational throughout. Both agree the human's residual job is taste and architecture stewardship.

Sources: DHH on Lex #501, Matt Pocock workshop