Measuring Serverless Apps in Amazon Bedrock

Learn to implement comprehensive observability strategies for serverless generative AI applications built on Amazon Bedrock in this conference talk from Conf42 Observability 2025. Explore the fundamental challenges of monitoring probabilistic AI systems compared to traditional deterministic applications, understanding why conventional monitoring approaches fall short when dealing with generative AI models. Discover the essential building blocks required for creating robust generative AI applications, including proper architecture patterns and design considerations specific to serverless environments. Master key objectives for AI application monitoring, focusing on performance metrics, cost optimization, and reliability measures that matter most in production AI systems. Examine practical implementation techniques for integrating observability tools and frameworks into your Amazon Bedrock applications, with detailed coverage of logging, tracing, and monitoring strategies. Follow along with a hands-on demonstration showing real-world implementation of observability patterns, including code examples and configuration details for monitoring AI model performance, token usage, and response quality. Gain insights into measuring and optimizing the unique characteristics of generative AI workloads, such as latency variability, output quality assessment, and resource consumption patterns that differ significantly from traditional web applications.