アプリ作成の民主化：ソフトウェアの専制は終わる

The tyranny of software is almost over. Since the first computer programmers wrote the first computer programs, we, the users of that software, have been forced to live in the worlds those programs create. The features are the features. The design is the design. Want something else, something better? Learn to code, I guess.

Until now, the people making a given piece of software — mostly well-paid professional developers — have rarely been the same as the ones using it: lawyers, doctors, churches, schools, me. (Where they overlap most directly is with developer tools, which are often the best and most passionately designed software you’ll find anywhere. Wonder why.) Software is built for the masses, designed not to be perfect for anyone but to be passable for everyone. Even when tech companies have tried to build tools to help people tune their software to their own needs, all they’ve been able to offer are hacky go-betweens like IFTTT and Apple Shortcuts. If you’re thinking in if-then statements, then you’ve lost most people.

Then, in the out-of-nowhere way that is common to the recent AI boom, the paradigm changed. In late 2025, an update to Anthropic’s Claude model turned its Claude Code tool from a code generator that was surprising if it worked to one that was surprising when it didn’t. Suddenly, all you needed was $20 a month and a half-formed idea, and an AI model could build you functional software. If you could explain what wasn’t working, Claude Code could probably fix it. Andrej Karpathy, an educator and researcher who was on OpenAI’s founding team, had called this new behavior “vibe coding.” Suddenly the vibes were off the charts.

The rise of AI coding tools like Claude Code — and OpenAI’s Codex, and GitHub Copilot, and Cursor, and Lovable, and Replit, and a thousand others — is already changing the way software developers work. They’re also giving way to an entirely new kind of software: the software we make just for ourselves. Not to raise venture capital for, not to eventually sell to Google. The era of personal software is upon us, and it is changing our relationship with technology forever. It has certainly already changed mine.

AI lets us make apps the way we used to make lists and spreadsheets. Managing the family budget? Do it in a hand-built app with every feature you need and exactly zero you don’t. Can’t make a to-do list app stick? Roll your own. Rather than triangulate a dozen schedules for the next family trip, whip up a custom meal planner (with built-in grocery assigner). Use it forever, use it once, it doesn’t matter. It doesn’t come with a subscription fee or send you marketing emails once a day for the rest of your life. It’s your software. And there’s never been anything like it before.

Robin Sloan, an author and technologist, wrote a blog post in 2020 entitled “An app can be a home-cooked meal.” The post has been shared widely in AI circles over the last couple of years, though Sloan wrote it well before the crop of generative AI tools. In it, he explains why he built a simple messaging app for his family. “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: “I have changed literally nothing in the app, and it’s glorious.”

Sloan is a fairly knowledgeable coder and built that app by hand. These days, though, he’s using AI to make even more home-cooked software. “It’s always weird little things,” Sloan tells me. He runs an olive oil company, and has whipped up ways to pull together product and customer information to automatically generate shipping labels. “It’s just a little Ruby script that pulls data from Shopify and USPS and kind of ties it together and it’s great.” It’s also extremely hacky. “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.” But while Robin’s around, it’s working great.

Personal software has its limits, of course. Your bespoke apps don’t come with a support line or a customer service team. They haven’t been thoroughly tested and make no security guarantees. The notion that large companies will ditch expensive enterprise software for something their marketing department vibe-coded is mostly fiction. So is the idea that we’ll all be running legions of AI agents, filling our phones with bespoke software, and obviating professionally made software altogether. Most of the apps we download are fine, regardless of who or what made them. But we all have those edge cases, the entirely reasonable ways in which we’d love to morph our software to our exact needs — the only problem is everyone else has needs too, and none of them are ours.

We all have those edge cases, the entirely reasonable ways in which we’d love to morph our software to our exact needs

My own edge cases are most present when it comes to productivity tools. Over the years I’ve tried every acronymic get-stuff-done system on the market — GTD, CARE, PARA, BASB, SMART, MIT, ZTD, and more — and dutifully poured my brain’s contents into every app with a checkbox feature. Eventually, I get annoyed with the app’s one tiny missing feature or bizarre design decision, stop using that app, start forgetting things, find another app that does those things better, spend a day porting my whole life into that app, encounter its own missing features and bizarre design decisions, and start the process anew.

Over time, I built the list of features my ideal productivity app requires. Every one of them has been built extremely well by at least one developer, so I know I’m not asking for anything impossible. But there are no apps, not one, exactly zero, that check off the whole list. When I called a bunch of developers to ask why they were missing such obviously crucial features, they all told me the same thing: Everyone has a list of requirements like this. No two users have the same list. And if you build everything for everyone, all you’ll really do is make a mess of your software. “It’s ridiculously easy to build features right now,” says Amir Salihefendic, the CEO of Doist, which makes the popular Todoist app. “But if you just do it naively, you end up with a system that nobody can figure out.” He then described a bunch of features other people have demanded he build, none of which made any sense to me at all.

In the era of personal software, though, you don’t have to build a system that works for everybody. And it is, in fact, ridiculously easy to build features right now. Which is why, over the 2025 holidays, like seemingly everyone with an X account and $20, I got a Claude Code subscription and set out building the app of my dreams. I’d show them!

I gave my app a name — Timetable — and described all the features I needed. It took 20 minutes or so to build a reasonably functional prototype. I then spent several days describing to Claude Code all the things that didn’t work, which mostly meant copying and pasting error codes and typing “what’s the full Terminal command” sixteen thousand times. I know how to code the way I knew Spanish in high school: I can ask about the library and order dinner, but nobody’s confusing me with a native speaker. My interactions with Claude Code amounted to a lot of pointing and gesturing and hoping the tool figured out I wanted soup.

Eventually, I had an app that worked more or less the way I wanted it to. It showed my calendar, my notes, and my tasks all in one place; it looked nice; it was easy to get stuff in and out of. It also, I discovered, only ran locally on my laptop. Thus began several more days of wiring everything up to GitHub, Supabase, Vercel, and other platforms, then an interminable round of complaining to my AI developer bot that no, it’s still not syncing, and why did my Google Calendar connection fail, and I’m so sorry but I have no idea what my GitHub secret code is. All that eventually sorted, I decided to make a native mobile app, since that would feel better, which kicked off several weeks of new errors, new features, and more accounts to sign up for.

Actually writing the code is but one part of creating and maintaining great software, and even the most advanced current tools have their limits. Design is maybe chief among them. Claude Code attacked my app’s design with fervent determination, the way I assume Jony Ive stares at a slab of aluminum and imagines removing all the ports from your laptop, but in this case every background ended up a gradient purple and every icon suggestion closely resembled a hamburger menu. When I pushed the bot to think a little more abstractly about the concept of a day, and to design an icon to fit the combination of journal and planner, it proudly displayed its answer in PNG image form. “I’m so sorry to tell you this,” I typed back, “but that looks like a butthole.” Claude’s next revision was once again three horizontal lines.

“I’ve found that most coding agents suck at writing good interfaces,” says Brian Lovin, a designer and software engineer who works on AI products at Notion and several AI-centric side projects. He has also experienced Claude Code’s love 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 says he’s learned to trust AI completely on the level of adding a tab to the settings panel, but “in the early days, when there’s no scaffolding, I don’t trust it at all.”

Evidently, lots of other people do trust it. These limits don’t appear to have stopped huge swaths of people from becoming app developers. The Information reported that the total number of new apps in Apple’s App Store grew 30 percent in 2025, after nearly a decade of slow decline, and it looks likely to grow even more in 2026. According to Apple’s own count, the App Store had just shy of 2 million apps at the end of 2024 — vibe coders could help double that count by the end of 2026. Meanwhile, GitHub had its fastest year of growth in 2025, and found that 80 percent of new users use the Copilot coding agent within their first week on the platform. Boris Cherny, the creator of Claude Code, tells me he knew the product was going to be a hit when he discovered the sales team was using it — “that’s when I really started to get that this is not just for engineers,” he says. While some of these developers are looking to build The Next Great App, many are also just shipping the thing they built for themselves. Many, many more aren’t shipping anything publicly at all.

In the course of reporting this story, I’ve heard tales — from sources, friends, and readers — of countless different kinds of personal software. (“What are you vibe-coding these days?” is a surprisingly good icebreaker in tech circles right now.) There are infinite variations on the to-do list, because writing a to-do list app is the coding version of learning Deep Purple’s “Smoke on the Water” for beginner guitarists — it’s just the way to learn where your fingers go. I also heard lots of stories about bespoke text editors, habit trackers, trip planners, family information managers, and blatant ripoffs of existing apps built just to avoid subscription prices.

I also heard from Brenden, who made a command line app for ranking fantasy baseball players based on their recent stats and future projections; Nathan, who told me about a script they wrote with Claude Code to introduce the concept of renewable energy to *Transport Tycoon Deluxe*, a game that came out in the 1990s and otherwise only knows about coal; Anthony, who built a tool for optimizing Secret Santa assignments; Tucker, who rigged up a way to mark the location of dog poop in the backyard, for easy finding later; Allan, who built a tracker for their migraines; and Brett, who created a way to track on which of their 102 stairs the mail carrier leaves a package. For most of these apps, and many others I heard of, the total addressable market is exactly one person and the revenue potential is precisely zero dollars. It is personal software in the truest sense — built by, and for, one person’s exact specifications.

Personal software doesn’t have to be built from scratch

My own early attempts at personal software are a semi-permanent record of both agentic AI’s limitations and my own personality flaws. I gave up on Timetable after a while, when I realized I had actually added a bunch of features I didn’t want and the whole thing was getting kind of annoying to use. I built another one, I apparently called it Spring, and I have absolutely no memory of what it even did. Basket was my attempt to build a super-inbox for all the links, notes, tasks, and all the other detritus I collect in my day-to-day life; I built a pretty cool system for texting things into the app and then bailed when my Twilio bill came due. I am apparently just as capable as anyone else of making software that annoys me.

What saved my efforts was the realization that personal software doesn’t have to be built from scratch. Knowledgeable developers might be newly capable home cooks, but the rest of us are more like customers at Chipotle. We don’t make the food, we don’t even really assemble it, but we get to decide what goes where and how it’s served to us. For most of us, the future of software is not building our own Excel from scratch, it’s using the models to build spreadsheets wildly more capable than we could create ourselves. It’s building the Chrome extension for your favorite app that is really only missing a Chrome extension. It’s tweaking the way things look to suit your exact taste and needs.

Going forward, a professional developer’s job might be largely to build infrastructure. “The minute you need multiple devices to stay in sync with a database, with some level of security … you’re talking developer primitives,” says Maggie Appleton, a designer and digital anthropologist currently working at GitHub Next. Appleton has been tracking the rise of personal software for years, and coined the term “barefoot developers” for the people who step up to learn the skills required to help their communities in ways no Big Tech firm would. She is a big believer in more people building software, if slightly less convinced than I am that *everyone *should do so. And she thinks we need “some sort of effort of open-source, really good primitives that you can plug and play together.” Some basic security systems; a few design best practices; a sturdy login system; payment support. Then, let anyone and everyone build on top.

Notion is maybe the best current (closed-source) example of what this might look like. Notion initially took off as a mainstream take on the low- and no-code movement — the app itself offered a bunch of building blocks, like images and tables and to-do lists, that you could organize however you wanted. Users loved the customization, to the point that sharing wildly over-designed Notion pages became something of a productivity nerd pastime. More recently, though, Notion has made the process even simpler: Just tell the built-in AI assistant what you want to accomplish, and it’ll build the page and system for you. “And because Notion provides so many underlying building blocks,” Ivan Zhao, Notion’s CEO, told me last year, “the AI only needs to write macros. The AI does not need to write software from scratch.”

This new approach comes with risks, though. The big one: Your ideas might be bad. A number of developers told me they’ve found their users to be incredibly good at noticing what they don’t like, and all over the place when it comes to offering solutions. And when your bespoke AI “solution” makes things worse, or breaks the rest of the software in some way, who’s to blame? How is customer support supposed to help people when literally everyone is using the app differently? Developers are struggling to build these open-ended platforms without letting users break or ruin them by accident.

Others I spoke to echoed this sentiment, and said that if you’re a developer with lots of users in this new era of infinite customization, it might be more important than ever to build software that just works out of the box. It’s still true that most people use default settings, and there’s no reason to believe every user is going to eagerly tweak every app’s every pixel to their exacting specifications. “I think there’s a responsibility to ensure we provide a coherent user interface,” says Balint Orosz, the CEO of the note-taking app Craft, “so if you like the core product, it feels like home.” The goal is not to ask people to reimagine everything every single time they open the app, he says, but to let the user say “I want this bigger” and have it get bigger in a way that makes sense.

Much of the current vogue in AI technology is geared toward making this kind of software adaptation both easy and universal. The Model Context Protocol gives developers an easy way to expose their data to AI agents. Lots of developers are also building command line interfaces to better integrate with coding tools. More and more apps are integrating directly with Claude and ChatGPT, too. Sure, if you want, you can use all this cool new tech to ask your chatbot about your email. That’s the boring answer. The better answer is to build yourself a brand-new way to email.

In this new world, the most important thing you’ll need is taste. Not objectively good taste, necessarily, so much as a keen sense of your own. You need to be like Rick Rubin, the famous music producer, who once told 60 Minutes that what made him successful was not any particular technical ability, but “the confidence I have in my taste, and my ability to express what I feel.” Rubin practices that art with A-list celebrities; you need to be able to do it with AI. Otherwise, you’ll land in what Lovin calls “doom loops,” telling your chatbot only what you don’t like and counting on the model to be the creative one. That way lies madness — and bad software.

I have no opinions about databases, it turns out, but I do care about typefaces and background colors. And so, the first actually useful bit of software I managed to vibe-code is just a way to smash a bunch of existing apps into a single screen. I keep all my bookmarks in Raindrop, which I find ugly to look at; all my tasks in Todoist, which I forget to check; all my notes in Obsidian, where they remain forever unorganized and forgotten; and all my events in Google Calendar, which I use religiously and without which I might never successfully leave my house. I failed over and over to build an app to replace those, but building a nicer way to look at them all took four API keys and an afternoon. And, yeah, a lot of “why doesn’t that button do anything” and “what does this error code mean” and “let’s try a color other than purple.” I kept telling Claude Code to make me an app that looked like a paper planner, and it pretty much delivered.

My app will never be in the App Store, and I probably couldn’t explain how it works in a way that would make any sense. That’s the beauty of the era of personal software: I don’t have to. You don’t have to try and figure it out, either. We are no longer required to use the software we are prescribed, or accept something that works fine for everyone and perfectly for no one. The best apps will be the ones that actually help you improve upon them. If you know what you need and what you like, you can make things work exactly the way you want. No learning to code required.

18 Comments

Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.

• David Pierce

-

-

-

-

-