The Pattern
AWS Amplify does what it promises for frontend deployment. Push to Git, get automatic builds, instant CloudFront distribution across 600+ edge locations. The auto-branch detection works as advertised - every push that compiles triggers deployment without manual intervention.
The interesting decision: deliberately bypassing Amplify's integrated backend services.
Why Teams Skip Amplify's Backend Integration
Amplify Gen 2's TypeScript-first approach abstracts AWS infrastructure complexity. For teams comfortable trading control for convenience, that's the point. But enterprise architectures often require granular control that sits outside Amplify's abstraction layer.
The trade-offs surface quickly:
Access control: API Gateway provides request-level validation and error handling that Amplify's integrated approach doesn't expose without escape hatches to CloudFormation or CDK.
Language constraints: Amplify's backend definition requires TypeScript. Teams with established Python or Go Lambda pipelines face a rewrite-or-integrate decision.
Authorization patterns: Protecting Bedrock model access and S3 operations through managed authorizers requires stepping outside Amplify's unified model.
The Terraform Gap
The current AWS Terraform provider doesn't support Amplify monitoring or alerting configuration. Teams must configure these in console, breaking infrastructure-as-code workflows. This friction point echoes feedback from development teams struggling to translate console configurations into declarative code.
The pattern: manual deployment first, then reverse-engineer Terraform. Not ideal, but common when provider coverage lags platform features.
What This Means in Practice
Amplify serves two distinct use cases:
- Pure frontend teams wanting zero infrastructure overhead get exactly what they need
- Platform teams requiring custom Lambda architectures use Amplify for deployment only, managing backend services separately
The code-first developer experience is real. So is the constraint when your architecture doesn't fit the abstraction.
Worth noting: The Cognito integration via environment variables (VITE_COGNITO_DOMAIN, VITE_COGNITO_CLIENT_ID) suggests authentication setup still requires manual User Pool configuration - another point where Amplify's automation stops and manual AWS management begins.
The platform delivers on deployment speed. Whether that justifies the backend integration constraints depends on your control requirements. History suggests teams choose control over convenience when the stakes are high.