Gemini Built a Yard-Care App From One Prompt
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Gemini (via Google AI Studio) built a yard-care Android app from a single text prompt, producing a functional preview in roughly five minutes and auto-fixing a flagged bug in 233 seconds.
- The AI "plant doctor" image-recognition feature correctly diagnosed the author's rhododendron as "critically bad" and identified that suffocating landscape fabric and sun-baked river rock were cooking the root system.
- Gemini defaulted to dark mode with dark purple and brick red accent colors that made text illegible, and initially offered preset "weather profiles" rather than pulling live weather data via API.
- The author spent a full sunny afternoon iterating on prompts — deleting and reinstalling the app on their phone to fix broken features like a date picker that didn't work — instead of doing actual yard work.
- The author concluded the app was overbuilt for a small urban-suburban yard and could probably have been replaced by a Gemini chat and a Google Keep to-do list, adding the project is unlikely to reach the Play Store.
- A prior vibe-coding experiment was a "Peach-o-Rama" tracker for a local grocery chain's annual event, which Gemini tried to implement as code that would only pretend to check the store's website and social channels.
Why it matters: This first-person test shows that single-prompt AI app generation now produces functional software in minutes — but defaults like dark purple text on a dark background and fake 'weather profiles' reveal how little Gemini models the physical world, forcing hours of iteration to fix what a human designer would have caught immediately.
