Anthropic · CCAR-P
Accuracy, latency, and observability
CCAR-P · Lesson 10 of 18 · 1:11:34
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the Claude Certified Architect – Professional exam, on the exam guide version 1.0, effective July 2026. In this lesson: total latency and time to first token, streaming, the settings that trade accuracy against latency and cost, justifying a configuration with measurements, observability challenges with language models, and monitoring at scale with tracing, OpenTelemetry, masking, sampling, and alerts. Includes three original practice questions.
Preview the topics
- Harborline’s chat assistant takes nine seconds before it shows anything, and the full answers are good. Customers leave before the answer appears. Which change helps most directly?
- Opus, high effort: 96% accuracy, 6 s at p95. Sonnet, medium effort: 94% accuracy, 2 s at p95. The requirement: 93%, within 3 s. Which configuration fits?
- Harborline’s agents handle a million calls a day. A few runs fail without any error, and the logs hold full prompts with health details. Which monitoring strategy fits?