QuestionQ73

VerifiedOperational Efficiency and Optimization for GenAI Applications

A company runs multiple Amazon Bedrock-based generative AI (GenAI) applications across several business units for customer service, content generation, and document analysis. Some applications have unpredictable token-consumption patterns. The company needs a comprehensive observability solution that delivers real-time visibility into token-usage patterns across multiple models. The observability solution must support custom dashboards for multiple stakeholder groups and offer alerting for token consumption across all foundational models used by the company’s applications.

Which combination of solutions will meet these requirements with the LEAST operational overhead?

Choose two
Explanation

Amazon Bedrock emits InputTokenCount and OutputTokenCount CloudWatch metrics for each ModelId, so CloudWatch dashboards can centrally visualize token consumption across foundation models and CloudWatch alarms can alert when token thresholds are crossed. Amazon Bedrock invocation logs include input and output token counts, model ID, caller identity, and optional request metadata; CloudWatch Logs Insights can aggregate those records by application and place the resulting queries in dashboards. Together, these native CloudWatch capabilities meet monitoring, stakeholder-dashboard, and alerting needs without operating a separate streaming, storage, or search pipeline.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!