I Created an App to Revive My Dying Yard

After stepping away from my computer for just five minutes, I returned to find two things waiting for me: a working app in a preview window and a bug report. The message read, “~ Channel is unrecoverably broken and will be disposed!” That sounded ominous! But right below it was a button to fix the issue. It was pretty wild that I had just instructed a computer to create an entire app with a single prompt, yet it still needed me to click a button to resolve a bug. I clicked it anyway, and in 233 seconds, the app reported success, using terms like “blockages” and “race conditions.” I didn’t grasp any of it, but the thrill was undeniable.

This was my second or third attempt at vibe-coding an app, depending on whether you count one that never made it past the preview stage. That unfinished project was a web app designed solely to check if a local upscale grocery store was hosting its annual Peach-o-Rama event. So far, no peaches. Regardless of how you count it, my current project was far more ambitious: an app to help me tame my unruly yard.

When my husband and I moved into our house eight years ago, we didn’t think much about yard maintenance. Sure, you mow the lawn and trim the hedges, but don’t the shrubs and trees pretty much take care of themselves? We neglected the yard until the weeds took over. The flower beds around the house quickly became overrun with weeds of epic proportions. Clearly, there was more to this “yard” business than we had anticipated.

We managed to win a few battles against the weeds but eventually lost the war and called in a landscaper. His one-time visit allowed us to leave the yard mostly on autopilot for a few years. It worked for a while, but then the weeds began to creep back, and the shrubs started showing signs of distress. As spring approached this year, I resolved to figure out what was going on with our yard.

I had a rough idea of where to start, but I wanted some assistance and a way to organize the tasks that needed to be done. Why not create an app for that?

I crafted a detailed prompt, essentially a list of requests: Help me manage a long list of yard care tasks; make recommendations; consider the weather; use image recognition to diagnose plant issues. I entered it all into an AI development platform, aiming to create an Android app I could load on my phone and take outside—where the plants live. I figured it would take about an hour, leaving me the rest of the day to document my yard’s condition and follow the app’s guidance.

My timing was a bit off. Sure, I had a working app in a preview window within minutes. It was logically organized, with sections for different plant zones and an AI “plant doctor” where I could upload images from my phone. But it had a significant color scheme issue.

For some reason, the app defaulted to dark mode, featuring dark purple and brick red accents. The text was nearly unreadable, and honestly, it was ugly. I suggested a white background with light green, pink, and blue accents, reminding it to prioritize readability. It returned with a more visually appealing design and an enthusiastic greeting at the top of the home screen: “Welcome Back, Gardeneer!” I actually liked the adventurous twist of “gardeneer,” so I kept it.

I retained the basic structure that the AI had created but made a few adjustments, like integrating live weather data instead of the odd climate presets it had suggested. Apparently, the AI thought I could just select a “profile” to match the day’s weather, which would adjust its watering recommendations. It seemed like a strange choice when live weather data is easily accessible via API, and it wouldn’t be the last time I had to remind the AI about the difference between the real world and a theoretical one. Regardless, I uploaded it to my phone and started using it as quickly as I could, too excited about launching my first app to worry about further iterations.

However, I soon realized I had overlooked some critical features. I couldn’t edit tasks once created or schedule them for specific days. I could create profiles for individual plants and group them by zone, but I couldn’t link them to specific tasks or do much else with them. There were separate tabs for one-off and recurring tasks, but every chore I added seemed to ignore this sorting and ended up on the recurring tab.

This led to a lot of tedious back-and-forth. I requested updates, waited for the AI to implement them, deleted the old version of the app from my phone, and replaced it with the new one. I’d notice other issues, like a date picker that didn’t actually allow date selection, and then I’d have to return to the chatbot. Instead of just managing an unruly yard, I now had an unruly app to deal with, too. There’s a lesson in there somewhere, I’m sure.

On the bright side, the AI plant doctor was effective right from the start. It was essentially a “Hey AI, diagnose this plant” button, and I uploaded a picture of a struggling rhododendron. After a minute, it provided a detailed report on the plant’s health (critically bad!), potential contributing factors, and actionable items I could add to my planner with a tap. That was exactly the kind of help I needed for my yard.

The landscaper’s set-it-and-forget-it solution had involved covering the flower beds with landscape fabric and river rock. He claimed this would solve the weed problem for a long time, and the existing plants would be fine. Plus, he offered a discount for cash payments. Done deal.

Now, years later, something was clearly amiss. The leaves on a bush near our front door had turned yellow, and flies buzzed around it constantly. The rose bushes grew lanky, and blooms were sparse.

The AI quickly pointed to the landscaper-recommended fabric and rocks as the culprits. It explained that they were suffocating the root system, which was also drying out because the landscape fabric had likely become clogged with dirt over the years. Additionally, the sun-baked rocks were essentially cooking the roots from above on hot days. No wonder our yard looked terrible; it was a miracle anything was still alive.

By that point, it was too late in the day to start Operation: Rhododendron Rescue. After all the back-and-forth with my app, I had wasted a full afternoon of nice weather typing prompts into a chat window. Every time I hit enter and sent the AI on a new coding mission, I was probably consuming the equivalent of a microwave dinner’s worth of electricity at a data center somewhere. The irony wasn’t lost on me.

Even though my app still wasn’t perfect, I set aside my feature requests the next day and decided to act on Dr. AI’s urgent recommendations for the rhododendron. I spent a sweaty afternoon with a podcast in my ears, raking back the river rock, cutting back the landscape fabric, and pruning some of the shrub’s twiggy bits. After that, I turned my attention to another rock bed, this one overrun with weeds growing on top of the fabric. Pro tip: Don’t use landscape fabric in your yard.

It was exhausting work under the blazing sun, and as I got closer to the thorny Himalayan blackberry vines invading the yard, the expletives started flying. But here’s the secret about yard work that I didn’t know eight years ago: it’s incredibly satisfying. That feeling when you get your tool under a big weed and pull the whole thing out, roots and all? Or when you get your shovel under a blackberry bush and rip it out of the ground, sending it back to hell? There’s nothing like it. Weeding is tough, but it’s also addictive. Once I get started, I can easily convince myself to stay out for another 20 minutes when I should really pack it up.

Eventually, I called it a day, opened my app, and crossed off a couple of the yard tasks I had completed. After spending several hours literally in the weeds, I had a new list of feature requests in my mind. I wanted ongoing assistance from the AI as I worked on reviving my plants, not just a one-time diagnosis. And while the idea of organizing my yard by zones appeals to my Type A nature, I’m not sure it’s particularly useful for me. I’m managing a small urban-suburban backyard, not Central Park. Could this app have just been a chat with the AI and a to-do list in a notes app? Probably.

I don’t think my “Gardeneering” app will ever make it to the app store, but the process has been quite enlightening. It’s hard to convey how surreal it is to watch a computer transform your text prompt into a functional piece of software—kind of like sharing a dream. But you really need to have a crystal-clear vision of the problem you want your app to solve. I could have saved myself a lot of back-and-forth if I had done a bit more groundwork to clarify my needs before firing off prompts.

My vibe-coding adventure has also highlighted something I knew logically but didn’t fully grasp: AI has no concept of the real world. It didn’t hesitate to put black text on a dark purple background because legibility isn’t a concern for a computer. It tried to interest me in generalized rather than real-time weather information, because what even is real-time weather to a machine? Even when I was working on my “Is It Peach-o-Rama Yet?” app, the AI attempted to pass off a version that would pretend to check the grocery store’s website and social media, but would really just cross-reference the date with the fact that Peach-o-Rama usually starts in mid-July. I had to insist that it actually mattered whether Peach-o-Rama was happening.

I haven’t given up on the yard app just yet, but the right version is likely much simpler than the one I started with. As for the advice I received from the AI, it seems to be spot-on. It’s only been a few days since I pulled the rocks and fabric back from the rhododendron, but I can already see new leaves sprouting on one of the branches. Maybe there’s still some life left in my yard after all.