Why Seeing the Opportunity Didn't Make Money: How Much You Earn Is Decided by the Last Two Steps

We rely far too much on one path: solving problems with ability. A lot of the time, we’ve never studied the underlying logic of making money. By “we” I mean people who know technology and live by their skills, myself included. I’d already built a cloud version of Claude Code in February; Claude only officially opened its own this week. Yet I never thought about how to turn that insight into money.

Making money has four stages: discovering value, creating value, replicating value, and monetizing value. Take entertainment short videos. The emotional value of a laugh you can scroll to anytime is the discovery; filming it is the creation. Plenty of people can do this much — the gap lies in the last two steps. Getting the same video in front of ten million people is replication. With tips, someone tips a hundred yuan, the platform takes half, and he gets fifty; with live-stream selling, if a product is a bit on the shady side, the margin can reach sixty or seventy percent. That’s monetization.

When we were making comic dramas, we had an information gap: model capability was limited, and our methodology and prompts were better. But we chose the monetization path poorly. Taking on production work was fine — the client pays the production fee up front, and the return is guaranteed. With a full-workflow collaboration system for comic dramas, the bulk of the money went to the AI models; a client spends ten thousand yuan and I make one thousand. The industry was also in cutthroat competition, burning cash on subsidies, and without outside investment you couldn’t make money. Later the models swallowed our advantage, and our way of working was copied. Some people did training instead, charging while the method was still worth something, and the money came faster.

Pinqieqie is a tool I built that cuts live-stream recordings into short videos. My original plan was three yuan a minute; ten MCNs paying a thousand yuan a day each would already be great. But next month a new model might just do the job outright. People who do business would use their technical lead to collect money early: let people use the system at cost, with a prepaid membership fee of five thousand yuan a year. Or sell a service directly, charge by delivery, and be an AI-native service company — if the tool gets replaced, the cost of delivery actually drops. Change how you charge, and the result changes.

Back to me. I get bored of old things and love new ones, and I love taking on ten at once. I’m good at discovering value and capable of creating it, but I haven’t thought about how to replicate it or monetize it more efficiently.

A friend told me Doubao Work sells one seat per person, around two thousand yuan a year, and you get a small commission for each one you sell — just a few hundred yuan. My first reaction was to look down on it; I figured what I teach is Claude Code, the most powerful productivity tool. But the people around me who’ve made so-called big money over these years all did it by replicating small money. I could use CC to build a tutorial that sells along with every seat — make it once, and it replicates. Sell a hundred seats a month, and that’s tens of thousands of yuan.

Then there’s responsibility and the sense of being worthy: professionals always feel the product isn’t polished enough, feel embarrassed to charge, and aren’t good at selling — and selling is exactly those two steps, replication and monetization. So we’ve done so many, many things, but we don’t make money. People like us have never lacked ability. How much you actually earn is decided by those last two steps.

EP0109 · 2026-09-29 · 05:04. 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.

We rely far too much on one path: solving problems with ability. A lot of the time, we’ve never really studied the underlying logic of making money. When I say “we,” I mean people who know technology, know their industry, and live by their skills — myself included. Why do I say that? Look, back in February I’d already built a cloud version of Claude Code. It runs on a server in the cloud, you open a browser on your computer, phone, or tablet and pick up where you left off, it never has to be shut down, twenty-four hours a day, and I use it every day. Claude only this week officially opened its own cloud version to all subscribers. I saw the trend, I built the thing, but I never once thought about how to turn that insight into money.

To make it concrete, making money can be broken into four stages. The first is discovering value. The second is creating value. The third is replicating value. And the fourth is monetizing value.

Let me just give an example — say, someone making entertainment short videos. People need to unwind, but offline entertainment is expensive, takes time, and isn’t always there when you want it. He realizes that something you can scroll to online anytime and get a laugh out of gives you emotional value — that’s discovering value. He films funny content — that’s creating value. Up to this point, a lot of people can actually do it. What really sets people apart is the next two steps. How do you replicate? You shoot one video once, grow the account, even pay for promotion, right, and get it in front of ten million people — now there are ten million portions of happiness. And monetizing? Go live and take tips: someone tips a hundred yuan, half gets taken by the platform, and he walks away with fifty at most. That funnel is narrow. What about live-stream selling? You trust him, so you buy things from him. If a product is a bit on the shady side, his margin can hit sixty or seventy percent. Someone spends a hundred yuan, and he makes sixty or seventy of it.

So when I have an information gap, choosing the whole path is critical. We had exactly that kind of gap when we were making comic dramas. Back then model capability was limited — if you just had AI do it directly, the results weren’t good. We had a methodology and prompts that could produce something better. But we didn’t pick the monetization path very well. We took on production work, and we also developed a comic-drama system where teams could collaborate across the entire workflow. The production work was fine — the client pays the production fee up front, so the return is guaranteed. Building the system was less rational. When clients generate videos, the bulk of what they spend goes to the AI models, and I only earn the difference. A client spends ten thousand yuan, and I make one thousand. And it’s not just that I earned little — the industry was in cutthroat competition, burning cash on subsidies, running at a loss, and under normal operations you simply couldn’t make money without outside investment. Low monetization efficiency, and a long cycle on top of that. Later, as the models got more capable, the advantage of our methodology and prompts was swallowed up by the models. As for the team-collaboration model, other products copied it too, so that competitive edge was gone as well. Yet also in comic dramas, some people went into training — teaching the method to others and charging for it while it was still worth something. That monetizes more efficiently, and the income comes in faster. You see, discovering value and creating value — we did both. The problem was the monetization step.

Take Pinqieqie as another example, a tool I built that automatically cuts long live-stream recordings into individual short videos. My original idea was to charge by duration — three yuan a minute, people use it, and I just collect the money. The best case would be ten MCNs all using it. If each of them paid me a thousand yuan a day, I’d already be very happy. But isn’t it possible that next month a new model comes out and just does this outright, right? What people who do business think about is: while we’re technically ahead right now, can we get the money back in advance? For example, let people use the system at cost, but have them prepay a membership fee — five thousand yuan a year. Thinking further ahead, Pinqieqie could also just sell a service. I offer a service with the best possible value for money, charge a service fee based on delivery, and become an AI-native service company. When a new model comes out, the tool may get replaced, but my cost of delivery actually goes down. That’s also part of thinking about monetization: the same thing, charged for in a different way, gets a different result.

Back to me. I’m someone who gets bored of old things and loves new ones, who likes a challenge. Once something repeats a second time, I lose interest. I love being the one who takes on ten — doing what others can’t gives me a real thrill. But that comes with a limitation. I’m especially good at discovering value, and I’m capable of creating it, but I’ve never really given much thought to how to replicate that value, or how to monetize it more efficiently.

So why did I start reflecting on this? A friend told me that Doubao Work sells accounts per head — one seat per person, around two thousand yuan a year. Sell one and you get a small commission, just a few hundred yuan per seat. My first reaction was, huh, what’s the point of earning that little? My habitual thinking never even went in the direction of replication and monetization.

But think about it carefully: the people around me who’ve made so-called big money over these years all did it by replicating small money. Selling seats basically means teaching companies how to use Doubao. Once the accounts are bought, if the employees can’t use them, it’s money wasted — someone has to teach them. Me, I couldn’t be bothered to teach people how to use Doubao, right? I figured what I teach is Claude Code, the most powerful productivity tool there is. But there’s no need for me to teach people Doubao myself — I can use CC to teach people Doubao. What I mean is, I use it to build a Doubao Work tutorial. A company buys seats, the employees learn by following the tutorial, and I don’t have to spend time going company by company to teach. Make the tutorial once, and it sells along with every seat — that’s replication. Do the math: sell a hundred seats a month, and that’s tens of thousands of yuan in revenue.

It’s precisely because I looked down on it that I might have missed this opportunity. There’s one more thing: a question of responsibility and of feeling worthy. Professionals like us always feel the product isn’t polished enough, feel embarrassed to charge for it, and aren’t good at selling. But selling is exactly those two steps — replication and monetization. We never spent time thinking about them, and we never did them. So we’ve done so many, many things, but we don’t make money. People like us have never lacked ability. How much you actually earn is decided by those last two steps. That’s it for today. See you tomorrow, bye.