I Created an AI App Because My Garden Was Dying

Five minutes after entering the prompt, the screen displayed a preview window of a working app and one bug report message. "The ～ channel has become unrecoverable and will be discarded"——it sounded ominous, but right below it was a button labeled "Fix Bug."

It was a peculiar experience to have AI generate an entire app from a single lengthy prompt, yet require just a button click to fix bugs. Still, when I clicked the button, Gemini reported the fix was complete in merely 233 seconds. Technical jargon like "blocking" and "race condition" filled the explanation, but honestly, I didn't understand much. Nevertheless, the experience was thrilling.

This was my second or third attempt using Google's AI coding tool, "Gemini." The catalyst was a very personal problem——my home garden's grass was beginning to die. I wanted a personal app that could manage the plant's condition and track watering schedules and soil status. However, I am neither an engineer nor a developer. Traditionally, creating such an app from scratch would have been unthinkable.

Yet when I conveyed detailed requirements to Gemini in natural language, a fully functional application was generated. UI layout, data management features, notification mechanisms——most of the features I envisioned took shape from text instructions alone. Without writing a single line of code.

Of course, it wasn't perfect. Bugs occurred, and some parts didn't work as intended. But watching AI detect the bug itself, propose fixes, and execute them——witnessing this entire sequence makes me feel that the very definition of "software development" is changing.

Now, with AI, anyone without specialized knowledge can create tools to solve their own life problems. While the garden management app is a small example, its potential is far from insignificant. The wall of "you can't build it if you can't code" is steadily becoming lower.