I built and shipped a multilingual party-game collection with AI.

Three releases, three very different results

Wacha! is my third released iOS app. A previous paid app generated $37.18 in roughly two months. Another app took about three months to build and earned only ¥3 from ads. Those two apps do not support multiple languages, so I am not linking them here as recommendations for international readers. They matter to this story because they taught me that finishing a product, monetizing it, and helping people discover it are three separate challenges.

Wacha! reached 12 downloads on its first day. The number is modest, but after the earlier results it felt like a useful early signal. Instead of hiding the small numbers, I use them to decide what to improve next.

AI was part of the entire workflow

I used AI for more than code: reading an expanding Swift codebase, implementing game rules, generating visual assets, translating the interface, preparing App Store review notes, and creating outreach and website content.

I have also used Claude Code and Cursor professionally. For this personal workflow, Codex has been the best fit for me—not as a universal benchmark, but because it can connect long natural-language requirements, source analysis, image generation, and visual feedback from screenshots in one continuous context.

Developing away from the desk

I sometimes leave my Mac running at home and work remotely while I am away. I send concrete tasks to Codex for implementation, review preparation, localization, and promotion. AI keeps production moving, while product decisions, testing, and responsibility for the release remain mine.

The next step

I am continuing to add games, improve multiplayer behavior, and make it easier for groups to start playing immediately.