2026 has truly been the year AI exploded. Pretty much every month, some model provider ships an even more powerful model, and I've gone from full-on FOMO at the start to simply lying flat. From the beginning of this year until now, I haven't launched a single new indie project. Not for lack of ideas — it's that every time I do some research, I find the same thing: whatever I can think of, someone else has already thought of; whatever I want to build, someone has already shipped, feature-complete and with a great experience.
AI Moves Too Fast — How to Deal with FOMO
Honestly, I had plenty of FOMO at first: always wanting to try the newest models and features, following the latest tricks in the community, studying the best practices and workflows shared by big accounts on Twitter, plus the endless stream of Skills, creative websites, apps, videos, and so on. Every time I finished scrolling Twitter, I felt like I'd fallen even further behind — as if I'd be left behind by the times if I didn't immediately adopt the newest tool.
But eventually I noticed: after all that chasing, I hadn't shipped a single extra product. The only thing that grew was my "look into this later" bookmark folder.
I don't chase any of it anymore. At this stage, large models — at least for coding, design, and image generation — can already handle 90% of the problems in day-to-day development. For my needs, an even stronger model is like using a cannon to kill a mosquito, and unless the price drops low enough, it just doesn't appeal to me. Of course, model providers keep releasing stronger models to survive in a brutally competitive market, and that's completely understandable. It's just that not all of us need to take part in that race.
My approach to FOMO these days is simple: • Good enough tools are good enough: Pick one or two models and an editor you're comfortable with, and learn them inside out. That's far more productive than switching to a new tool every week • Start from problems, not tools: Look for solutions when you hit a concrete problem, instead of seeing a new tool and wondering "what could I build with this?" • Control your information intake: Ninety percent of what's on Twitter is noise. Changes that actually matter will reach you within a few days, even if you miss them at first
Tools are the means; the product is the goal. The lower the barrier to using tools, the more valuable it becomes to know what to build.
AI Can Boost Productivity and Assist, but It Can't Give You Ideas
I often see incredibly clever ideas on Twitter from people all over the world — the kind of thing AI would never come up with on its own. For example: • Stamp-frame camera: Collecting and recording the little joys of everyday life in the form of postage stamps • Receipt printer: Showing small daily moments or purchases as an animation of a printer spitting out a receipt. The printer blends perfectly with the Dynamic Island — beautiful, fun, and full of a sense of ritual • Bubble tea log: A very niche logging scenario — record every bubble tea you drink, then generate a beautiful calendar • As for creative images and videos, there are even more — all kinds of wild ideas that AI can help you bring to life
But what happened next? I saw countless stamp-style camera apps, countless apps with receipt-printing animations, countless apps for logging bubble tea, coffee, and meals, and countless images and videos that are identical except for the subject. They're beautifully designed, with smooth and elegant interactions — one look and you want to download them. But to me, just like PixBlur, which I built, they're pretty, useful, but unprofitable vases.
This made me realize a harsh reality: in the age of AI, ideas themselves have no moat.
It used to take months for a good idea to go from going viral to being cloned, because others had to find a designer, write code, and get through app review. Now, when an idea blows up on Twitter, a dozen "lookalikes" pop up within