Your Personal Software Revolution

Artificial intelligence is empowering a new generation of "vibe coders" to craft software precisely tailored to their desires, heralding a personal software revolution.

The long-standing dominance of standardized software is nearing its end. For decades, users have been confined to the digital environments created by programmers, accepting predefined features and designs. The only recourse for those seeking something different or superior was to "learn to code," a significant barrier for most.

Historically, the creators of software—predominantly well-compensated professional developers—have rarely been its primary users, a diverse group including lawyers, doctors, religious institutions, schools, and individuals. (The notable exception is developer tools, which often exhibit superior design due to this direct overlap.) Software has typically been mass-produced, aiming for broad acceptability rather than individual perfection. Even attempts by tech giants to offer customization tools, such as IFTTT and Apple Shortcuts, proved to be cumbersome intermediaries, requiring a logic-based approach that alienated many users.

However, a paradigm shift, characteristic of the recent AI boom, occurred unexpectedly. In late 2025, an update to Anthropic’s Claude model transformed its Claude Code tool. Previously, its code generation was surprisingly effective if it worked; now, it was surprising when it *didn't*. Suddenly, for a mere $20 a month and a nascent idea, an AI model could construct functional software. If an issue arose, Claude Code could likely resolve it. Andrej Karpathy, a prominent educator and researcher from OpenAI's founding team, aptly termed this new capability "vibe coding," signifying a dramatic increase in intuitive software creation.

The proliferation of AI coding tools—including Claude Code, OpenAI’s Codex, GitHub Copilot, Cursor, Lovable, Replit, and countless others—is already reshaping professional software development. More significantly, it's fostering an entirely new category of software: applications built exclusively for personal use. These aren't designed to attract venture capital or eventual acquisition by tech giants. The era of personal software has arrived, fundamentally altering our interaction with technology, and for many, it has already proven transformative.

AI now enables us to develop applications with the same ease we once created lists and spreadsheets. Imagine managing your family budget with a custom-built app featuring every necessary function and absolutely no superfluous ones. Struggling to maintain a to-do list? Design your own. Instead of coordinating multiple schedules for a family trip, quickly generate a personalized meal planner complete with a grocery assigner. Whether used indefinitely or just once, these apps come without subscription fees or daily marketing emails. This is truly *your* software, an unprecedented level of personalization.

In 2020, author and technologist Robin Sloan penned a widely circulated blog post titled “An app can be a home-cooked meal.” Though written before the advent of generative AI tools, it resonated deeply within AI circles. Sloan detailed his motivation for building a simple family messaging app: “There will be no sudden redesign, no flood of ads, no pivot to chase a userbase inscrutable to us,” he wrote. “It might go away at some point, but that will be our decision.” Five years later, in late 2025, Sloan updated his post, stating, “I have changed literally nothing in the app, and it’s glorious.”

While Sloan is a proficient coder who initially built that app manually, he now leverages AI to create even more bespoke software. “It’s always weird little things,” Sloan shares. For his olive oil company, he devised a system to consolidate product and customer data for automated shipping label generation. “It’s just a little Ruby script that pulls data from Shopify and USPS and kind of ties it together and it’s great.” He admits it's "extremely hacky," acknowledging, “If I ever get hit by a bus, it’s going to be a problem for my olive oil company, because only Robin knows how to run the software.” Nevertheless, for now, it functions perfectly.

Naturally, personal software has its limitations. Bespoke applications lack dedicated support lines or customer service. They are not rigorously tested and offer no security guarantees. The notion that large corporations will abandon expensive enterprise solutions for something their marketing department "vibe-coded" is largely unrealistic. Similarly, the idea that we will all manage legions of AI agents, filling our devices with custom software and rendering professionally developed software obsolete, is improbable. Most commercially available apps adequately serve their purpose. However, we all encounter specific "edge cases"—perfectly reasonable ways we wish to adapt software to our precise needs—a desire often unmet because everyone else has different, equally specific requirements.

We all possess those unique "edge cases," the perfectly valid instances where we desire to mold software to our exact specifications.

My own "edge cases" frequently arise with productivity tools. Over the years, I've experimented with every acronym-driven "get-stuff-done" system available—GTD, CARE, PARA, BASB, SMART, MIT, ZTD, and more—diligently transferring my thoughts into countless apps featuring a checkbox. Inevitably, I become frustrated by a minor missing feature or an odd design choice, abandon the app, start forgetting tasks, then seek a better alternative. This leads to a day spent migrating my entire life into the new app, only to discover its own set of deficiencies and peculiar design decisions, restarting the cycle anew.

Eventually, I compiled a comprehensive list of features essential for my ideal productivity app. Knowing each feature has been expertly implemented by at least one developer, I was confident my demands were not unrealistic. Yet, not a single existing app, not one, met all these criteria. When I questioned developers about these seemingly crucial omissions, their responses were consistent: every user has a unique list of requirements, and attempting to build everything for everyone inevitably results in a chaotic, unusable product. Amir Salihefendic, CEO of Doist (makers of the popular Todoist app), explains, “It’s ridiculously easy to build features right now. But if you just do it naively, you end up with a system that nobody can figure out.” He then cited numerous features other users had requested, none of which resonated with me.

In the burgeoning era of personal software, however, the imperative to create a universally appealing system vanishes. Indeed, building features is now remarkably straightforward. This realization led me, during the 2025 holidays—like countless others with an X account and $20—to acquire a Claude Code subscription and embark on building my dream app. I was determined to prove it possible!

I named my application "Timetable" and meticulously described all the features I required. Within about 20 minutes, a reasonably functional prototype was complete. I then dedicated several days to articulating to Claude Code everything that wasn't working, largely involving the tedious process of copying and pasting error codes and repeatedly typing “what’s the full Terminal command.” My coding proficiency is akin to my high school Spanish: sufficient to ask for the library or order dinner, but far from native fluency. My interactions with Claude Code often felt like a lot of pointing, gesturing, and hoping the tool understood I wanted soup.

Ultimately, I developed an app that largely functioned as desired. It consolidated my calendar, notes, and tasks, offered an appealing interface, and facilitated easy data entry and retrieval. However, I soon discovered it only operated locally on my laptop. This necessitated several more days of integrating it with GitHub, Supabase, Vercel, and other platforms, followed by an endless series of complaints to my AI developer bot about syncing failures, Google Calendar connection issues, and my complete ignorance of my GitHub secret code. Once these hurdles were cleared, I decided to create a native mobile app for a better user experience, which initiated weeks of new errors, feature additions, and more platform sign-ups.

The act of writing code represents only one facet of creating and maintaining robust software, and even the most advanced current tools possess limitations. Design, perhaps, is chief among them. Claude Code approached my app’s design with fervent determination, much like how one might imagine Jony Ive contemplating an aluminum slab to envision the removal of all laptop ports. In my case, however, every background became a gradient purple, and every icon suggestion closely resembled a hamburger menu. When I challenged the bot to conceptualize the idea of a "day" more abstractly and design an icon combining journal and planner, it proudly presented a PNG image. “I’m so sorry to tell you this,” I typed back, “but that looks like a butthole.” Claude’s subsequent revision was, once again, three horizontal lines.

“I’ve found that most coding agents suck at writing good interfaces,” states Brian Lovin, a designer and software engineer who contributes to AI products at Notion and various AI-centric side projects. He, too, has experienced Claude Code’s predilection for purple gradients. “I don’t know how to get it to not do that, except just annoyingly prompting it more and more and more.” Lovin trusts AI implicitly for minor tasks, like adding a tab to a settings panel, but admits, “in the early days, when there’s no scaffolding, I don’t trust it at all.”

Despite these limitations, it's clear that many individuals *do* trust these tools. These constraints haven't prevented vast numbers of people from becoming app developers. The Information reported a 30 percent increase in new apps in Apple’s App Store in 2025, reversing a nearly decade-long decline, with further growth anticipated in 2026. Apple’s data shows the App Store nearing 2 million apps by the end of 2024; "vibe coders" could potentially double that figure by the end of 2026. Concurrently, GitHub experienced its fastest growth year in 2025, finding that 80 percent of new users adopted the Copilot coding agent within their first week. Boris Cherny, creator of Claude Code, recognized its potential when he observed his sales team using it—"that’s when I really started to get that this is not just for engineers,” he recalls. While some developers aim to build the next groundbreaking app, many are simply releasing tools they created for themselves, and a significant number are not publicly releasing anything at all.

During the research for this story, I gathered numerous accounts—from sources, friends, and readers—detailing diverse forms of personal software. (“What are you vibe-coding these days?” has become an unexpectedly effective icebreaker in tech circles.) To-do list apps, for instance, appear in infinite variations, serving as the coding equivalent of learning Deep Purple’s “Smoke on the Water” for aspiring guitarists—a fundamental exercise. I also heard many stories about custom text editors, habit trackers, trip planners, family information managers, and straightforward replicas of existing apps designed solely to circumvent subscription fees.

I also heard from Brenden, who developed a command-line application for ranking fantasy baseball players based on recent stats and future projections; Nathan, who described a script written with Claude Code to introduce renewable energy concepts to *Transport Tycoon Deluxe*, a 1990s game originally limited to coal; Anthony, who built a tool to optimize Secret Santa assignments; Tucker, who devised a method to mark the location of dog poop in the backyard for easy retrieval; Allan, who created a migraine tracker; and Brett, who developed a system to record on which of their 102 stairs the mail carrier leaves a package. For most of these and many other apps I encountered, the total addressable market is precisely one person, and the revenue potential is exactly zero dollars. This is personal software in its purest form—crafted by, and for, one individual's specific requirements.

Crucially, personal software does not always necessitate building from scratch.

My own initial forays into personal software serve as a semi-permanent record of both the limitations of agentic AI and my own behavioral quirks. I eventually abandoned "Timetable" upon realizing I had inadvertently incorporated unwanted features, making the entire experience cumbersome. I built another, apparently named "Spring," of which I have absolutely no recollection of its function. "Basket" was my attempt at a super-inbox for all the links, notes, tasks, and miscellaneous digital detritus I accumulate daily; I developed a rather clever system for texting items into the app, only to abandon it when my Twilio bill arrived. It seems I am just as capable as anyone else of creating software that ultimately annoys me.

My efforts were salvaged by the realization that personal software need not be built from the ground up. While skilled developers might now be adept "home cooks," the rest of us are more like customers at Chipotle. We don't prepare the food, nor do we truly assemble it, but we dictate the ingredients and how they are served. For most, the future of software isn't about constructing our own Excel from scratch; it's about utilizing AI models to create spreadsheets far more capable than we could design ourselves. It's about developing that missing Chrome extension for a favorite app or fine-tuning existing interfaces to perfectly match one's aesthetic and functional needs.

Moving forward, the primary role of professional developers might shift towards building foundational infrastructure. “The minute you need multiple devices to stay in sync with a database, with some level of security … you’re talking developer primitives,” explains Maggie Appleton, a designer and digital anthropologist at GitHub Next. Appleton, who has tracked the rise of personal software for years and coined the term “barefoot developers” for those who learn coding to serve their communities in ways Big Tech wouldn't, strongly advocates for more people creating software, though she is slightly less convinced than I am that *everyone* should. She believes there's a need for “some sort of effort of open-source, really good primitives that you can plug and play together”—basic security systems, design best practices, robust login systems, and payment support—allowing anyone to build upon them.

Notion stands as perhaps the best current (albeit closed-source) example of this vision. Notion initially gained traction as a mainstream interpretation of the low- and no-code movement, with the app itself offering a host of...