Do We Really Need More Platforms?

(The following story was originally written in Chinese and translated into English with the help of AI. Some parts are true, some are fake—just take it with a laugh.)

An AI-Powered Teaching Platform

My college senior project was to build a platform—an AI-powered teaching platform. That sounds too vague, so let me narrow it down: imagine Coursera plus an LLM teaching assistant, where each course gets its own fine-tuned TA LLM. But honestly, do we really need a platform? The problem we were trying to solve with this platform was the pain point of distance learning: when learning online, there’s no one to ask. That’s also the problem I ran into when I took a Coursera course, so it was really my personal pain point. I took an Objective-C app development course, but seriously, it’s 2023 (when I was a sophomore)—who is still making native iOS apps with Objective-C? Everyone uses Swift. Not only was the Xcode version in the course half of the latest Xcode version at the time (yes, that far behind), but the UI and many features were completely different.

I’ve been talking too much about my own pain. I even emailed the department chair back then and later became a Coursera TA—I’ll save that story for next time, it’s a fun experience. Back to the platform itself. The goal was to give students someone to ask when taking classes remotely, right? In 2026, LLMs are already so powerful that we might not even need a specialized LLM for each course, but I still think the idea was pretty novel at the time.

Why do I have this doubt? Because whether it’s a startup competition or a school project, as long as you’re building software, it seems impossible to escape the fate of becoming a platform. A platform feels like the destination for every idea, the last mile to scaling up or turning something into a product. If the problem I’m solving is “learning a foreign language while watching videos,” I might just want to make a YouTube Chrome extension. But if I want to make language learning big and keep everyone spending time on my server, then it seems I have to expand that extension into a platform. From a simple “enhanced YouTube subtitles,” it becomes: I can track your study hours; now that I have study hours, I should give you more resources; based on the language you’re learning and the videos you like, I should recommend content; so I need to design an algorithm; and that algorithm has to run on my server because it’s proprietary. Boom, I start spending time designing my platform, frontend, backend, database, referral links to get people to subscribe, and even partnering with not-so-famous YouTubers (because they’re the ones I can afford).

But back to my original intention: I just want to learn Korean well enough to understand what Chodan is talking about. I don’t want some platform.

Today’s Coding Agents (any LLM-based software that can call tools and write code) are so powerful that I should be able to build one myself, right? So I should skip the step of searching for existing products and just develop a Chrome extension based on my own needs. But I want to share it, so I open-source it. People start filing GitHub issues with feature suggestions, and I implement them with Codex. Users email me saying, “Can you add xxx?” and I agree, so I implement that with Codex too. The browser extension gets bloated and laggy when enabled, and more and more buttons appear on the page. Then something in my brain misfired: with so many features, I should build a full website to present them properly. So I start designing frontend, backend, database, referral links to get people to subscribe, and even partnering with even-less-famous YouTubers…

What Sindre Sorhus Inspired in Me

A legend among legends. For an introduction, check out Chaping’s video:

https://youtu.be/SGKvQHF3rOY?si=G3OfAHo2YVuFJMOc

The apps he designs are tiny, but each one solves a real problem. Even though we already have enough apps and platforms on our phones, Sindre’s apps are so small that I think they’re worth downloading—clean, elegant, and don’t take up space. Simple but not Easy.

So I started wondering: what is the finish line for these startup competitions and hackathons? Is it another app where finding the logout button is as hard as climbing to the sky, or a platform so complex I can’t be bothered to learn it? I don’t want either—more like a grave than a finish line. I don’t want to be the CEO of Google or Meta; I don’t want to host too much stuff. I just want to elegantly solve a pain point. Or is it impossible to make money without building a platform? Or am I just overthinking, and who cares whether you solve it elegantly—just make it so complex users can’t understand it, because users are idiots anyway, and once they pay, I’ll make it simple.

Damn, that’s super pathetic, okay? And in some ways, pretty unethical.

So what can I do? Nothing? No, we should make one-time paid software mainstream, let the subscription model die completely, celebrate the system where you can fully own something, and praise the craftsman spirit—starting with you and me.