Wanted builds app review automation system in one week using internal tools

Details

Automated the app review management task, which had been neglected for three years, in just one week
Immediately built the server, DB, UI, and AI agent by leveraging existing internal infrastructure and AI tools such as Backyard, Ennoia, Claude Design, and Montage
Maintained store reply quality by implementing a workflow where AI drafts replies and humans approve them using Slack emojis
Configured a vibe-coding-based development environment by integrating code and documentation into a single repository and automatically syncing it with Confluence

Tech Stack and Implementation

Deployed web, API, and DB in a k8s environment instantly using natural language via Backyard
Configured three AI agents for reply writing and Jira issue creation by leveraging MCP on the Ennoia platform
Implemented standardized UI without designers by connecting Wanted's design system, Montage, to Claude Design

Performed review collection and reply posting by directly calling the App Store Connect API and Google Play Developer API
Implemented Slack notifications and remaining time display features for batch failures, considering the 7-day limit of the Google Play API
Extracted and managed version information from App Store reviews by matching RSS feed IDs