AI Engineer YouTube · May 18, 2026

Build Agents That Run for Hours (Without Losing the Plot) — Ash Prabaker & Andrew Wilson, Anthropic

Build Agents That Run for Hours (Without Losing the Plot) — Ash Prabaker & Andrew Wilson, Anthropic video thumbnail
Why it matters

Why self-evaluation is a trap and adversarial evaluator agents work better; why context compaction doesn't cure coherence drift but structured handoffs do; how to decompose work into testable sprint contracts; how to grade subjective output with rubrics an LLM can actually apply; and how to read traces as your primary

My takeaway: Why self-evaluation is a trap and adversarial evaluator agents work better; why context compaction doesn't cure coherence drift but structured handoffs do; how to decompose work into testable sprint contracts; how to grade subjective output with rubrics an LLM can actually apply; and how to read traces as your primary