“Tracing is significantly more impactful for debugging multi-step AI agents than for single-call LLM applications because the context in agents is dynamic and unpredictable over many steps.”