It all started with a “low and thick” dog. My mother was simply enjoying a city park when this dense, fast-moving canine T-boned her, fracturing her right tibia. But as unfortunate as that was for her, it set in motion an unexpected journey for me: venturing into software development after successfully avoiding coding for two and a half decades.
A fractured tibia brings a cascade of minor frustrations, as anyone who has experienced similar circumstances will tell you. Take, for instance, the hours my father spent wrestling with phone trees, just trying to coordinate my mom’s medical care. While these frustrating calls might seem insignificant in the grand scheme of things, that dog had inadvertently chosen a technologically pivotal moment to wreak havoc.
The Dawn of Vibe Coding
For too long, we’ve been passive consumers of Silicon Valley’s promises, endlessly scrolling app stores, hoping someone else has built exactly what we need. But now, AI offers a democratizing sidekick: vibe coding. The premise is simple yet revolutionary: build your own apps, no matter how niche or trivial, with absolutely zero programming skills.
You simply articulate your frustration, and a sophisticated network of large language models, code generators, and development environments springs into action. This new paradigm means that for the first time in history, even minor annoyances can warrant serious technological solutions.
And “niche and trifling” perfectly describes my target. While others might use vibe coding for resume reviewers or inventory trackers to boost productivity, my focus was different. For years, I’d been professionally and personally fixated on what the policy world calls “sludge.” This is the ever-growing tide of tiny administrative obligations that seems to define modern life and erode our ability to get things done.
Think about the hassle of dealing with insurance, connecting that insurance to your doctor, keeping tabs on airline miles, or navigating your child’s school portal. These aren’t isolated incidents; they’re interconnected mushrooms sprouting from the same frustrating network. The steps required to dispute a charge or unsubscribe from a forgotten streaming service all feel like individual assaults on our precious time.
Battling the Bureaucratic Sludge
This “sludge” represents a calibration issue. Major problems might attract legislative attention or journalistic scrutiny, but these smaller, “too petty to litigate” issues simply become accepted facts of life. The arc of history might bend towards justice, but when you’re fighting a one-dollar bank fee, it often bends directly toward endless hold music.
This is where the fantasy of vibe coding truly captured my imagination. These daily hassles aren’t always accidental byproducts of complexity; sometimes, they’re intentional features. A confusing portal, a dropped call, a process just opaque enough to discourage follow-through – at scale, these function more like policy than bugs.
The app I envisioned would shine a light on this phenomenon, making the cumulative weight of these administrative burdens impossible to ignore. I wanted to create a system that would expose the trembling field of sludge mushrooms for all to see.
My mom, despite her recovering leg, fortunately boasts a Claude Pro subscription. After years of needling her about AI’s broader implications, I pushed those concerns aside one Sunday and drove to her house. Following some tibia-related conversation, I opened her computer and began to “emit vibes.”
- “I’d like to create a communally shared app that gathers and shares information related to how much time and energy we devote to fighting burdensome administrative tasks, bureaucratic sludge, Kafka-esque unsubscribe mazes, byzantine insurance portals, wrongful charges, denied claims, confusing membership plans, and the like.”
I proceeded to describe a dashboard where users could log frustrating incidents, recording the time spent, annoyance level, and what they’d rather have been doing. Every submission would be “dopaminally rewarded” with an inspiring resistance quote and a photo of a cute kitten, puppy, or baby chimp. I’d also train Claude to generate “wider context” paragraphs about systemic sludge patterns and draft complaint letters to relevant regulatory bodies.
From Idea to Interface: My First App
Claude began to “noodle,” and for a moment, I feared my vibes would lead to nothing but an error page. I recalled vague advice from Reddit forums about needing Harvard-level knowledge of dozens of programming languages and cloud platforms. Was vibe coding just an elaborate “stone soup” hoax?
My worry lasted mere seconds. Claude emerged from its processing, declaring my concept “fantastic” and “genuinely useful.” A couple of clarifying questions later, I was staring at a real interface. The “Log Incident” and “Dashboard” tabs weren’t fully functional yet, and we still needed to implement data storage and context generation, but the bare bones of an online application had materialized.
I spent the next hour ironing out kinks, with Claude guiding me through fixes. I understood very little of the underlying code, merely following (and issuing) commands. Yet, steady progress was made, with Claude’s confident, reassuring, and clear assistance always a “whimper away.” It felt like building an elaborate Lego creation: you don’t understand what each individual piece does, but by following the directions precisely, the final product emerges.
I drove home invigorated, a feeling I hadn’t experienced since a brief arc-welding phase in my twenties. “I can’t believe I, a regular person, can make this!” All the websites and apps I navigated daily had always been mysterious, like pyramids built by an unfathomable priesthood. Now, I was a pyramid builder.
I wasn’t alone in this newfound power. Someone in Florida had built Stratus, a guitar pedal that generates effects from plain English descriptions. Another person created a Plywood Cutting Visualizer, and MIXCARD transforms Spotify playlists into physical postcards. The barrier between a creative idea and its practical realization had seemingly dissolved for anyone with a passing itch and an afternoon to spare.
However, this accessibility presented its own challenges. What happens when everyone builds their own app? The environmental, political, and economic concerns about AI came roaring back, accompanied by a new worry. Before the pandemic, I hosted “Admin Night” gatherings where friends would tackle personal sludge collectively. I noticed many modern administrative tasks were the residue of yesterday’s “tech solutions” – systems that promised efficiency but then broke, expired, or started demanding monthly fees.
Would AI be any different, or was I simply contributing to the creation of more sludge? A few days later, I revisited my mom. My app was progressing, but the final stretch felt as long as the first 90 percent. I needed accounts at GitHub to store my code, Supabase for user records, and Netlify to serve the app – each step straightforward, yet each an opportunity for error.
I inadvertently exposed my API key in the public GitHub repository, but Claude swiftly caught it and moved it to a secure location. Then came security considerations. My Instagram algorithm, having detected my new hobby, suggested a prompt for app builders on how to compel Claude to run a security audit. Sure enough, we discovered a vulnerability where user-submitted text was inserted unsanitized into the page’s HTML, allowing malicious code execution in visitors’ browsers. It was an easy fix.
The tasks were menial but manageable: ferrying credentials between services, clicking “Deploy,” watching failures, pasting errors back to Claude, and repeating. It was more assembly than engineering, like putting together an Ikea daybed. After three more visits to my mom’s house, by the time her tibia was strong enough for crutches, my app was ready for its trial run.
I invited my dad to the computer, interviewed him about that frustrating phone tree, and typed his answer into the app. He found the process cathartic. He recounted how, when calling to make a doctor’s appointment, he always had to listen through irrelevant options, including information for sending a fax, before getting to the appointment-booking function. He estimated he had spent three hours navigating such systems, rating the annoyance a three out of ten, and would have rather been gardening.
His submission was rewarded with an Ursula K. Le Guin quote and a picture of a dachshund napping. From there, my auto-context generator sprung to life. It explained that automated phone systems often prioritize options based on call volume data or administrative convenience, not user intent, which is why fax options might appear before appointment booking. It also noted that simple tweaks could fix misleading claims like “menu options have changed recently,” if companies considered patient friction an actual problem worth solving.
My father, already impressed that I could operate a computer, was particularly struck by this conjuring of perspective. Next, Claude and I developed an automated content moderation process. Now, before user submissions appear on the public dashboard, they pass through rigorous filters. Two days later, I sent the URL to my Admin Night community. We exist because the sludge problem is communal, and any real solution must be communal too. Soon, they were submitting complaints about Audible’s credit policy, Hulu double-billing, and the log-in gymnastics required to buy a daughter’s prom ticket. One member, Danielle, aptly described the whole thing as a “grievance dragnet.”
Source: Wired – AI
