Why an AI Venture Shouldn't Be Love at First Sight: Build Only What Can Be Productized and Iterated On Indefinitely

Pretty much everyone around me who has made their first pot of money in recent years spotted a clear money-making loop and then figured out how to replicate it, doing that one thing. They know that doing this kind of work and finding this kind of client brings in this kind of steady income, so they hire more people and go find more clients. Almost all of them work this way. The loop can keep being replicated because they can do the same thing a hundred times without getting bored. People like us lose interest by the second time.

That got me thinking. Those of us who are supposedly smart love novelty and tire of the old, just like a player; we don’t like doing the same thing over and over. So we have to find the thing we can do for a long time without tiring of it. As in love, love at first sight is not as good as never tiring of each other, and work is the same. Plenty of projects have already been paid for up front, yet we just don’t want to do them, because we got them working early on and there’s no fun of a challenge left. Still, I think almost everyone can probably find one thing they can keep doing and keep iterating on without ever getting tired of it.

I may have found mine recently: Claude Code in the cloud. I built it in February, and it runs on a server in the cloud; open a browser on a computer, phone, or tablet and pick up right where you left off. It officially opened to the public at the end of July. In the seventy-some days since, it has had 1,450 updates, roughly twenty updates a day.

When something suddenly feels awkward to me, or I wonder whether it could work another way, I add a new feature right away. A user gripes about something on the co-creation board, the AI builds it, and I click once to ship it. Every time I open it, it’s a different it. It always feels fresh, so I keep doing it and never get bored. Pinqieqie, my live-stream auto-editing tool, is the same.

There’s another thing about me: once other people can do something, I no longer want to do it. I like taking on ten at once, and doing what others can’t is what really excites me. We did an AI short drama project before, taking commissions to make dramas, and we also built a system where teams could collaborate on AI short dramas. Now every company has launched a platform for teams to collaborate on AI short dramas, the models are that capable, and the output from each platform isn’t really any different. So I don’t want to do the AI short drama project anymore. I have no motivation and no interest left.

In fact, back when we were doing it, we had already concluded that AI short drama wasn’t something I could do every day and keep doing for the long haul.

So this year we have one standard for what we take on: only build things that can be productized and keep being iterated on.

EP0112 · 2026-10-02 · 02:10. The picture uses a digital human, and the voice uses my cloned voice. This is an English translation of the approved Chinese script used to create the narration.

The people around me who’ve made their first pot of money over the past few years, pretty much all of them spotted a clear money-making loop and then figured out how to replicate it. They just do that one thing. They know: hey, if I do this kind of work and find this kind of client, I get this kind of steady income, so then I need to hire more people and go find more clients. Basically every single one of them works this way. And that loop can keep being replicated because they can do the same thing a hundred times without getting bored. But people like us? By the second time, we’ve lost interest.

So here’s the thought this sparked for me. Those of us who are supposedly smart love novelty and get tired of the old, just like a player, right? We don’t like doing the same thing. So we really do have to find that one thing, the thing I can keep doing without getting tired of it. It’s like dating: love at first sight is not as good as never tiring of each other, and work is the same. There are lots of projects where the money’s already been paid up front, and we just don’t want to do them anymore. Why? Because we already got them working early on, and there’s no fun of a challenge left. But I think pretty much everyone can probably find one thing they can keep doing, keep iterating on, and never get tired of.

Take me, for example. I think I may have found mine recently: Claude Code in the cloud. I built it in February, and it runs on a server in the cloud. Open a browser on your computer, phone, or tablet and you can pick right up where you left off. It officially opened to the public at the end of July. So now, seventy-some days in, it’s had 1,450 updates, roughly twenty updates a day. I’ll suddenly think, hmm, this part feels so awkward, could it work like this instead? And I add a new feature right away. A user gripes about something on the co-creation board, the AI just builds it, and I click once to ship it. Every time you open it, it’s a different it. It always feels fresh, so I keep doing it and never get bored. Pinqieqie, my live-stream auto-editing tool, is the same.

And there’s something kind of twisted about me: once other people can do a thing, I don’t want to do it anymore. I like taking on ten at once; doing what other people can’t is what really gets me going. Take the AI short drama project. We did that before, taking commissions to make dramas, and we also built a system where teams could collaborate on AI short dramas. And now ByteDance, Tencent, Baidu, iQIYI, Bilibili are all doing AI short drama. Every company has launched a platform for teams to collaborate on AI short dramas, the models are so capable, and the output from each platform isn’t really any different. So I don’t want to do the AI short drama project anymore. Zero motivation, zero interest. Actually, back when we were doing it, we’d already concluded on our own that AI short drama wasn’t the kind of thing I could do every day and keep doing for the long haul.

So this year we have just one standard for what we do: only build things that can be productized and keep being iterated on. That’s it for today. See you tomorrow, bye.