My yard is dying, so I made an app for that

The author experimented with Google’s Gemini model, giving it a detailed prompt to build a gardening app. Gemini returned a preview window of the app and a bug message, which the author fixed by clicking a button; after 233 seconds the model reported completion. The episode illustrates that the model can produce functional code from natural‑language instructions, though it still relies on user intervention for error handling.

Summary written by Kernelia from the original article by The Verge — AI. The story and its rights belong to its author.