翻訳待ち：You don't have to make money with AI. You could just be happier

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

You don't have to make money with AI. You could just be happier. Posted on:August 22, 2026 | at 12:00 AM You don’t have to make money with AI. You could just be happier. Back in July I upgraded Claude from Pro to Max to get some real work done. Subscription rate limits are a special kind of annoying. The 5-hour window and the weekly cap are fine for casual use. Pro, plus a couple of comparable subscriptions from other vendors, rationed sensibly, gets you through most side work. But when you’re trying to finish something serious on a deadline, getting cut off mid-task is the one thing you can’t afford. The work got done early. The requirements didn’t balloon, which, thank God, is rarer than it should be these days. That left me with two weeks of subscription and about $100 already paid for. What do I do with it? My first instinct was to not let it go to waste and get the ROI positive. Every social feed right now is full of people who made tens of thousands of dollars a month with AI (most of them, I’m fairly sure, are warming you up to buy a course). The cost was sunk, I’ve been a developer for a while, and I’m decent at both code and product. Why not try? So for a week and change I brainstormed, prototyped, and built. The demos all looked fine. But every time I measured the gap between a nice demo and something a stranger would trust and pay for, the remaining work was enormous. And honestly, the whole exercise made me anxious. With four or five days left, I stopped. I wasn’t going to commit to a startup in two weeks, and without that commitment I wasn’t going to break through anything. I took a day off. The next day The next morning I opened some of the software I use every day, the ones I use a lot but have never been happy with. On a whim I thought: why don’t I finally do all those improvements I’ve been sitting on for years? Start with reading. I read far more than I write, and I’d long wanted a Markdown reader that matched my taste and my reading habits. It didn’t need every feature a commercial product has. It needed the features I wanted, a UI I like looking at, and keyboard shortcuts that combine the way I think. I opened a coding agent. We spent about an hour pinning down the feature list and the look, broke it into an architecture and a task list, and I handed the tasks off. Two hours later I had a Markdown reader that was entirely mine. I spent a few more hours fixing the bugs the agent had introduced and adding things I noticed I wanted while using it, including a small agent plugin. Seven or eight hours in, it was perfect for me. It has the texture of a newspaper page. It has just enough agent in it: I can ask questions about what I’m reading, take notes, and it keeps track of my thoughts and a bit of memory. File navigation feels like Neovim and the shortcuts are a simplified Vim flow. It was everything I’d wanted, and I was more excited than I’d been all month. I’d been thinking about this reader for years, mostly about how to strip out everything I didn’t need. Writing it by hand would have been weeks of work plus debugging, so every time, I’d decided to put up with what I had. Then the rest of them After that I went through the other tools I live in. My terminal news reader is a newsboat-based RSS reader for working through subscriptions without leaving the shell. I reworked its features and keybindings and added some smarter behavior. I’d never liked its rendering pipeline, so I swapped it for a Textual-based one, then added translation and export so it can hand things off to my other tools. About a day later it did what I wanted. My careers system: I cut what I never used, kept what worked, and reshaped it around how I operate. It had always been slightly awkward. Now it fits. And I wrote a research assistant for a family member, designed entirely around how she does her research, so she can get through it faster. What the $100 bought I never got the money back. The demos are sitting in a folder, and I may come back to them if I have more thoughts. But I read in a reader I like. I check the news in a terminal that behaves. My job tool stays out of my way. Someone in my family finishes her research a little faster, and a few more tools are next in line. I touch these things every day, and every one of them had been slightly wrong for years because fixing them by hand cost more than the annoyance was worth. That math changed, and I hadn’t noticed. AI is a tool. It can make you money, and it can also make your days better. ai side-project tools essay