AWS re:Invent 2025 - Serverless full-stack in action: AI-driven developer experience (DEV309)

Leveraging AI-Driven Developer Experience for Serverless Applications

Overview

The presentation showcased how developers can leverage AI-powered tools, specifically the Kiro CLI, to streamline the development of serverless applications. The focus was on building a full-stack image generation application, including both the backend and frontend components.

Key Highlights

Serverless Development Lifecycle

• The "inner loop" of rapid local development and iteration
• The "outer loop" of CI/CD, deployment, testing, and production monitoring
• Importance of fast feedback cycles and collaboration across the development lifecycle

Kiro CLI: AI-Powered Developer Assistant

• Kiro CLI is a CLI tool that integrates with large language models to assist developers
• Provides access to a variety of MCP (Model Capability Provider) servers, each with specialized capabilities
• Enables developers to leverage AI for tasks like creating serverless projects, updating infrastructure, and testing deployments

Building the Serverless Application

1. Backend Development:

   • Used Kiro CLI to create a new SAM (Serverless Application Model) project with Node.js, an HTTP API, and two Lambda functions
   • Leveraged MCP servers to automatically generate the necessary infrastructure, including an S3 bucket for image storage
   • Integrated the Titan image generation model from Bedrock to handle image creation within the Lambda functions

2. Frontend Development:

   • Created a React-based frontend application using Vite
   • Integrated the frontend with the backend API to allow users to generate and view images

3. Observability and Monitoring:

   • Instrumented the Lambda functions with AWS Lambda Power Tools for structured logging, tracing, and custom metrics
   • Enabled AWS CloudWatch Application Signals to provide advanced application-level monitoring and observability

Steering Files: Controlling AI Behavior

• Kiro CLI supports the use of "steering files" to define rules and guidelines for the AI assistant
• Allows developers to control aspects like commit messages, code style, testing standards, and security best practices
• Ensures the AI-generated code and infrastructure align with the project's specific requirements and conventions

Seamless Local-to-Cloud Workflow

• Demonstrated the ability to open a deployed Lambda function directly in Visual Studio Code using the AWS Toolkit
• Enabled rapid iteration and deployment by leveraging features like AWS SAM Sync for incremental updates

Key Takeaways

1. AI-powered tools like Kiro CLI can significantly streamline the development of serverless applications, reducing the need for manual coding and infrastructure management.
2. Leveraging specialized MCP servers allows developers to tap into domain-specific expertise (e.g., serverless, documentation) to enhance the AI's capabilities.
3. Steering files provide a powerful mechanism to control the behavior of the AI assistant, ensuring the generated code and infrastructure align with project-specific requirements and best practices.
4. The integration between Kiro CLI and cloud-based tools (e.g., AWS Lambda, CloudWatch) enables a seamless local-to-cloud development workflow, facilitating rapid iteration and deployment.
5. Comprehensive observability and monitoring, enabled by tools like AWS Lambda Power Tools and CloudWatch Application Signals, are crucial for ensuring the reliability and performance of serverless applications.

Business Impact

• Accelerates the development of serverless applications by automating repetitive tasks and infrastructure management
• Improves developer productivity and reduces the risk of human error through the use of AI-powered assistance
• Ensures the generated code and infrastructure adhere to best practices and project-specific requirements
• Enhances the overall reliability and observability of serverless applications, enabling better monitoring and troubleshooting

Real-World Examples

• The presented image generation application showcased how Kiro CLI and the integrated MCP servers can be used to build a full-stack serverless solution, including both backend and frontend components.
• The ability to open a deployed Lambda function directly in the IDE and leverage features like SAM Sync demonstrates the seamless local-to-cloud workflow enabled by these tools.
• The incorporation of observability and monitoring capabilities, such as structured logging, tracing, and CloudWatch Application Signals, highlights the importance of these features for ensuring the reliability and performance of serverless applications in production.