Building Agentic AI Applications with Amazon Kiro
Contact us to book this course
On-Site, Virtual
1 day
This practical, hands-on course is designed to help developers build production-ready AI applications using Amazon Kiro's spec-driven development approach. Throughout this intensive one-day course, students will learn to leverage Kiro's advanced features including vibe coding, specification-driven workflows, and automated hooks to transform natural language requirements into maintainable, enterprise-grade applications deployed on AWS.
As a case study, students will convert a basic ui into an intelligent shopping application that uses generative AI to summarize product reviews and provide personalized product recommendations based on shopping habits and browsing history. Working with Amazon Bedrock, Bedrock AgentCore, and AWS deployment services like Amplify, Lambda, and DynamoDB, students will experience the complete journey from rapid prototyping to production deployment with CI/CD automation.
By the end of this course, students will be able to:
-
Install and configure Amazon Kiro IDE and CLI for AI-driven development
-
Differentiate between Amazon Kiro and Amazon Q Developer and understand when to use each tool
-
Apply vibe coding techniques for rapid prototyping and exploratory development
-
Implement spec-driven development workflows to transform natural language requirements into production code
-
Generate and utilize comprehensive specifications including requirements documents, design artifacts, and implementation tasks
-
Integrate Amazon Bedrock and Bedrock AgentCore to build AI-powered features for review summarization and product recommendations
-
Configure SageMaker for machine learning model integration in application workflows
-
Create and deploy automated hooks for code quality, security scanning, and documentation maintenance
-
Deploy full-stack web applications to AWS using Amplify, Lambda, and DynamoDB
-
Implement basic CI/CD pipelines using GitHub Actions for automated deployment to AWS
-
Apply AWS security best practices for AI-powered applications including IAM roles, encryption, and credential management
-
Navigate the learning path from Kiro fundamentals to advanced AWS AI/ML development
Prerequisites
-
Basic understanding of Python or object-oriented programming concepts
-
Familiarity with command-line interfaces (terminal/shell)
-
Basic understanding of web application architecture (frontend/backend concepts)
-
Git version control basics (clone, commit, push, pull)
Course outline
-
- Introduction to Amazon Kiro
-
-
- What is Kiro? AI-native IDE overview
- Code-OSS foundation and VS Code compatibility
- Kiro vs. Amazon Q Developer: when to use each
- Kiro vs. other AI IDEs (Cursor, Windsurf)
- Claude Sonnet 4.5/3.7 models powering Kiro
- Kiro Architecture and Setup
- Kiro IDE interface overview
- Kiro CLI capabilities and use cases
- Builder ID and authentication options
- Open VSX plugin compatibility
- Pricing tiers and free preview limits
- Spec-Driven Development Philosophy
- The problem with "vibe coding" alone
- Spec-driven development workflow overview
- Requirements → Design → Tasks → Implementation
- When specs provide value vs. quick prototyping
- Documentation and maintainability benefits
- Course Application Overview
- AI-powered shopping application features
- Review summarization with Bedrock
- Personalized recommendations with AgentCore
- AWS deployment architecture preview
- Starter template walkthrough
-
- Deep Dive: Spec-Driven Development Workflow
- Three-phase spec workflow detailed
- Phase 1: Requirements generation (user stories, EARS notation)
- Phase 2: Design specification (diagrams, interfaces, schemas)
- Phase 3: Task generation and implementation
- Keeping specs synchronized with code changes
- Amazon Bedrock Fundamentals
- Bedrock overview and foundation models
- Text summarization use cases
- API integration patterns
- Authentication and IAM roles
- Cost considerations and optimization
- Bedrock SDK for Python/JavaScript
- Building the Review Summarization Feature
- Feature requirements overview
- User stories for review summarization
- Design considerations (API endpoints, data flow)
- Bedrock integration architecture
- Testing strategies for AI features
- Error handling and edge cases
Lab 2: Implement Review Summarization with Specs
- Kiro Hooks Architecture
- What are hooks? Event-driven automation
- Hook triggers: file save, create, delete, manual
- Use cases: testing, security, documentation, code standards
- Creating hooks with natural language prompts
- System prompt optimization
- Repository folder monitoring
- Team-wide hook enforcement via Git
- Security Scanning with Hooks
- Credential leak detection patterns
- Security best practices automation
- Pre-commit security validation
- Integration with AWS security tools
- Handling false positives
- Security hook examples
- Amazon Bedrock AgentCore
- AgentCore overview and capabilities
- Building recommendation systems
- Context management for personalization
- User behavior tracking and analysis
- Real-time vs. batch recommendations
- Integration patterns with shopping applications
- AWS Deployment Architecture (12 minutes)
- AWS Amplify: hosting and frontend deployment
- AWS Lambda: serverless backend functions
- DynamoDB: NoSQL database for application data
- API Gateway integration
- Architecture diagram and data flow
- Connecting AI services (Bedrock, SageMaker)
- CI/CD with GitHub Actions (10 minutes)
- CI/CD pipeline fundamentals
- GitHub Actions workflow structure
- Automated testing in pipeline
- Environment-based deployments (dev, staging, prod)
- Secrets management in GitHub
- Rollback strategies
- AWS Security Best Practices (8 minutes)
- IAM roles and least privilege principle
- Credential management (never hardcode!)
- Encryption at rest (DynamoDB, S3)
- Encryption in transit (HTTPS, TLS)
- VPC and network security considerations
- CloudWatch logging and monitoring
- Cost management and billing alerts
Lab 4: Deploy to AWS with CI/CD (1 hour 20 minutes)
