How to use Claude to rebuild a vacation planning app
The author reports rebuilding the app, adding mobile clients with feature parity, and reducing monthly costs from about $20 to $0. They also describe remaining code quality issues.
news.ycombinator.com
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Rebuild [vacation organizer] using its existing [domain model], [API Gateway and Lambda backend], [DynamoDB schema] and [Svelte frontend]. Then implement [Flutter mobile clients] against the same APIs, including [photo sync requirements].
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- kaydub
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The author reports rebuilding the app, adding mobile clients with feature parity, and reducing monthly costs from about $20 to $0. They also describe remaining code quality issues.
- Human involvement
- The author assessed code quality and recently found Flutter code smells; the post does not detail a formal review process.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Added LLM vacation-plan generation to an existing organizer.
- Rebuilt the application with a serverless backend and Svelte frontend.
- Built Android and iOS Flutter apps using the same backend.
Kaydub originally built a vacation organizer for use with their wife before adopting LLMs. The initial application provided basic create, read, update and delete functionality. They later integrated an LLM to generate vacation plans, which they describe as working okay. After becoming aware of Claude Code, they rebuilt the application rather than simply translating the existing implementation.
The earlier system used Spring Boot on AWS ECS with an application load balancer and CloudFront, plus supporting services including Aurora RDS. The replacement uses API Gateway, Lambda and DynamoDB with a Svelte frontend. The author says the domain model was essentially the only part retained and reports monthly costs dropping from about $20 to $0.
They also built Flutter applications for Android and iOS that share the web application's backend and have feature parity. Phone photos can queue for background upload, with Wi-Fi syncing by default and automatic image processing through an S3 event Lambda. Forwarded reservation emails are parsed using an LLM; ambiguous messages enter a travel inbox for manual routing.
The author reports frontend, backend APIs and mobile applications were entirely LLM-built, but identifies remaining code quality work, especially in Flutter.
Who it's for: developers rebuilding a personal organizer across web and mobile clients.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · kaydub
Frequently asked questions
What architecture replaced the original app?
API Gateway, Lambda and DynamoDB replaced the Spring Boot and ECS setup, with a Svelte frontend.
How does the app handle uncertain reservation emails?
When the LLM cannot identify the email or its vacation, the message enters a travel inbox so the user can choose the vacation and object type.
More details
Remaining steps
- Added photo syncing features to the Flutter application.
About this account
The author attributes the frontend, backend APIs and Flutter apps entirely to LLM-generated code.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=48464402. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: kaydub on Hacker News
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · AWS API Gateway · AWS Lambda · DynamoDB · Svelte · Flutter
Vacation app · Serverless · Flutter
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.