Distributed tracing plus GPU profiling shows where load creates queueing, memory pressure, or blocked tool calls. Average response time alone hides the bottleneck. From an NVIDIA systems-engineering lens, Option C aligns with the way agentic services should be decomposed and measured. The selected option specifically C states “Profile each major system stage using distributed tracing, analyze GPU utilization with NVIDIA performance tools, and map queuing delays against varying workload patterns.”, which matches the operational requirement rather than a superficial wording match. The practical pattern is trajectory-level evaluation, distributed tracing, task-completion metrics, latency breakdowns, and regression gates. The NVIDIA implementation angle is not cosmetic here: NeMo Evaluator and agentic metrics focus on trajectories and goal completion, not only the fluency of the last response. The distractors fail because manual spot checks are useful but cannot replace regression tests across query classes, temporal drift, and tool failure modes. This is exactly where NVIDIA’s stack is strongest: separating acceleration, orchestration, policy, and observability.