DooDooLamb News
My yard is dying, so I made an app for that
Brief published June 14, 2026 ยท Original source published June 12, 2026
Original reporting by Allison Johnson at theverge.com.
Automated brief. Verify important details at the original source.
Local Man Discovers AI Can Code Better Than Him, Immediately Questions Life Choices
A software engineer at The Verge has just published what might be the most accidentally honest piece of tech journalism in 2024: a meditation on watching Google's Gemini AI build a better app than he could, faster than he could, while his yard slowly dies from neglect.
Allison Johnson's journey began with the most relatable problem in suburban America: a yard that looked like it was auditioning for a post-apocalyptic movie set. Instead of doing the reasonable thing (buying a sprinkler timer, hiring a landscaper, or accepting his fate as someone whose lawn would never make the HOA newsletter), Johnson decided to solve this with code. Because of course he did. This is 2024, and we solve everything with apps now, even the basic human inability to remember to water plants.
The twist? Johnson didn't actually write the app. He just described his yard-related existential crisis to Gemini in what he calls "a lengthy prompt," wandered away for five minutes to presumably contemplate his wilting grass, and returned to find a fully functional application waiting for him like a digital genie had granted his most boring wish.
The AI had built him a yard work to-do list app, complete with what Johnson describes as surprisingly thoughtful features. Not just a basic checklist, mind you, but something that actually understood the nuanced hell of suburban lawn maintenance. The kind of app that knows the difference between "water the roses" and "water the roses but not too much because they're still recovering from that fungal thing and also check if the sprinkler system is actually hitting them or just creating an expensive puddle on the driveway."
But here's where the story gets beautifully absurd. Mid-creation, Gemini threw an error message that read "Channel is unrecoverably broken and will be disposed!" which sounds like either a critical system failure or the AI having an existential breakdown about being asked to care about someone's landscaping problems. Johnson admits this "sounded bad," which is the kind of masterful understatement you'd expect from someone watching an AI panic while building them free software.
The error message was apparently just Gemini's way of clearing its throat before continuing to code. Like a programmer muttering "this is all garbage" before proceeding to write the most elegant solution anyone's seen all week. The AI shrugged off its own crisis, kept building, and delivered exactly what Johnson needed. This is either remarkably advanced error handling or proof that AIs are already learning to be passive-aggressive about their working conditions.
What Johnson created here, probably without meaning to, is a perfect encapsulation of where we are with AI coding tools in 2024. Not quite the "AI will replace all programmers" dystopia that keeps senior engineers awake at night, but definitely past the "AI can barely write a hello world function" stage that let us all sleep peacefully in 2022. We're in this weird middle ground where AI can build you a genuinely useful app in the time it takes to make a sandwich, but it might also have what appears to be a small nervous breakdown while doing it.
The technical term for what Johnson experienced is "vibe-coding," though he probably made that up. It's coding based purely on vibes and natural language descriptions rather than actual technical specifications. You describe the vibe you want ("I need to stop killing my yard through neglect"), and the AI figures out the technical implementation. It's like having a junior developer who's really good at translating your rambling product requirements into actual features, except the junior developer is a large language model trained on the entire internet and has no concept of why anyone would care about grass.
This raises uncomfortable questions for everyone who's ever spent three hours debugging a CSS layout or arguing with a package manager. If an AI can build a functional app from a casual description of someone's yard troubles, what exactly are we all doing here? Johnson seems to have stumbled into the same existential territory that many developers are quietly exploring: the strange experience of watching a machine solve problems you've been paid to solve, faster and with less complaining.
The really unsettling part isn't that the AI wrote code. It's that the AI apparently understood the problem better than most human developers would have. Anyone who's ever tried to explain their actual software needs to a programmer knows the frustration of describing a simple workflow and getting back questions about database schema and authentication protocols. Gemini just said "cool, dying yard, got it" and built something that worked.
Of course, we don't know how well this app actually functions in practice. Johnson's piece reads more like a mood board than a product review. Will the AI-generated yard app remind him to water his roses without flooding his foundation? Will it account for seasonal variations in plant care? Will it eventually gain sentience and judge him for letting his lawn get to this point in the first place? These are the real questions, and Johnson seems more interested in the philosophical implications of AI coding than the practical ones.
But maybe that's the point. We're all just building apps to solve problems we created by being human in the digital age. Johnson couldn't keep his yard alive, so he made an app about it. The app probably won't save his grass, but it might make him feel like he's doing something productive about the problem. Which is, when you think about it, exactly what most software does: give us the illusion of control over chaos we're too overwhelmed to actually manage.
The real winner here is Google, who just got a glowing testimonial disguised as yard care content. Nothing says "our AI is ready for enterprise adoption" quite like a tech journalist accidentally proving that Gemini can build better software than most development teams, even when it's having what appears to be a minor existential crisis.
Johnson's yard is probably still dying, but at least now he has an AI-generated app to feel guilty about not using.