AWS re:Invent 2025 - Improve DevEx for GenAI Deployments: Dynatrace + Amazon Bedrock (SEC207)

AWS re:Invent 2025 - Improve DevEx for GenAI Deployments: Dynatrace + Amazon Bedrock (SEC207)

Introduction

• This session explores how Dynatrace and Amazon Bedrock can help improve the developer experience (DevEx) for deploying and managing generative AI (GenAI) applications.
• The presentation covers the key challenges faced by developers when working with GenAI models and how the Dynatrace and Amazon Bedrock integration can address these challenges.

Challenges in Deploying and Managing GenAI Applications

• Complexity of GenAI models: GenAI models, such as large language models, can be highly complex and resource-intensive, making them difficult to deploy and manage.
• Unpredictable performance: GenAI models can exhibit unpredictable performance, with fluctuations in latency, throughput, and accuracy, which can impact the user experience.
• Lack of visibility and observability: Developers often lack visibility into the runtime behavior of GenAI models, making it challenging to identify and resolve issues.
• Difficulty in scaling and optimizing: Scaling and optimizing GenAI applications can be a complex task, requiring specialized expertise and tooling.

Dynatrace and Amazon Bedrock Integration

• Dynatrace is a leading observability and application performance monitoring (APM) platform.
• Amazon Bedrock is a managed service that provides access to a variety of GenAI models, including large language models, from leading providers.
• The integration between Dynatrace and Amazon Bedrock provides the following benefits:

Improved Observability and Visibility

• Dynatrace automatically instruments and monitors GenAI models deployed on Amazon Bedrock, providing detailed insights into their performance and behavior.
• Developers can access real-time metrics, such as latency, throughput, and accuracy, to understand the runtime performance of their GenAI models.
• Dynatrace's AI-powered root cause analysis helps identify and resolve issues quickly, reducing the time to resolution.

Simplified Deployment and Scaling

• The integration allows developers to easily deploy and scale GenAI models on Amazon Bedrock, leveraging Dynatrace's automated deployment and scaling capabilities.
• Dynatrace's intelligent scaling algorithms ensure that GenAI models are provisioned with the appropriate resources to meet the changing demand, optimizing cost and performance.

Enhanced Debugging and Troubleshooting

• Dynatrace's advanced tracing and debugging capabilities provide developers with a comprehensive view of the entire GenAI application stack, from the model inputs to the output.
• Developers can quickly identify and resolve issues related to model performance, data quality, or integration with other components of the application.

Real-world Use Cases and Examples

• The presentation showcases several real-world use cases where the Dynatrace and Amazon Bedrock integration has helped organizations improve the DevEx for their GenAI deployments:
  • A financial services company using GenAI for automated customer service and document processing, leveraging Dynatrace to ensure reliable and scalable performance.
  • A healthcare organization using GenAI for medical diagnosis and treatment recommendations, utilizing Dynatrace to monitor model accuracy and identify potential biases.
  • A retail company using GenAI for personalized product recommendations, employing Dynatrace to optimize the user experience and track the impact of model updates.

Key Takeaways

• The Dynatrace and Amazon Bedrock integration provides a comprehensive solution for improving the DevEx of GenAI deployments.
• By addressing the key challenges of complexity, unpredictability, and lack of visibility, the integration helps developers deploy and manage GenAI applications more effectively.
• The enhanced observability, simplified deployment, and advanced debugging capabilities enable organizations to unlock the full potential of GenAI technology and deliver better user experiences.