The AI One-Person Company Community Is a False Premise: What an OPC Needs Isn't Another OPC, It's Small and Medium-Sized Businesses

The AI OPC community is a false premise. At the invitation of Chongli’s tourism department, I organized this in-person session. All nine people there were one-person companies who signed up through my group — freelancers with mature, stable businesses. Before the AI era they were already one-person companies; the business itself made money, and using AI just made it more efficient, that’s all. We meet every three months, and across this year’s three meetups, everyone went from trying AI, to taking on projects entirely with AI, to starting to pick their projects — in just half a year.

The OPCs that came after AI aren’t all unable to support themselves; the problem is a narrow model. One path is self-media and ad revenue. The top few AI bloggers all have teams, yet ads are their only income. They write articles with AI and iterate on traffic feedback, and what comes out is content that panders to traffic — one moment this one is amazing, the next the sky is falling — with no information value. The other path is training. Too few people can deliver AI implementation for businesses, and even where there’s traffic and demand, there aren’t reliable people to scale up how much work can be taken on.

Getting business is hard because AI OPCs have no division of labor, which means everyone is competing with everyone. Complex jobs almost nobody can do; ordinary jobs everyone can do, and so can the client. On an image-generation gig at fifty yuan per image, a few hundred images, the client thought the price was too high, because they can generate images with Codex themselves. For things like image generation and animated comic dramas, where anyone delivers an eighty-out-of-a-hundred result, every gig becomes a race to the lowest price: a hundred, fifty, thirty, twenty, and finally for free.

OPT — one-person companies forming a team to take on business — is an even bigger false premise. It assumes each person fills in one piece of the puzzle, but one-person companies are possible now precisely because AI has filled in those pieces, and there are very few areas where people are much better than AI. Put together, the team ends up with overlapping, redundant capabilities and communication overhead, the profit isn’t enough to split, and it isn’t much stronger than a single OPC.

Nor is there a channel partner with no AI implementation ability who can keep bringing in business; even with the business in hand, they wouldn’t trust an OPC they don’t know to deliver it. The people who can deliver well are few, they aren’t short of work, and they’re the ones passing work out.

Take me. I like doing projects, not teaching; my group pushed for a month before I scheduled one small class. With three projects running in parallel I can’t take on more, so I usually pass work to relatively reliable longtime group members I know inside out, and only hand a project over once I have a deep understanding of what they’ve done before. If I’m not comfortable with it, I don’t hand it over.

I don’t pass work along casually, because that isn’t responsible to either side. Clients who haven’t figured out what they want, or who won’t pay in advance, are clients capable OPCs won’t consider; and jobs where the client is reluctant to pay aren’t worth recommending to anyone.

An OPC actually doesn’t need another OPC; what an OPC needs is a partner that keeps bringing work — small and medium-sized businesses. These businesses need to cut costs and raise efficiency, but their existing teams can’t master doing the work with AI that quickly, and that’s exactly the gap OPCs fill: OPCs flexibly take on mature work, and the businesses reduce staffing costs; OPCs, in turn, sometimes need a backstop. A healthier model is many-to-many — many small and medium-sized businesses working with a group of OPCs, rather than OPCs forming some kind of community of their own.

EP0110 · 2026-09-30 · 04:11. 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 AI OPC community is a false premise.

This event was in Chongli. We were invited by Chongli’s tourism department to organize an in-person sharing session. All nine people there were one-person companies, and they all signed up through my group — really, they’re freelancers with mature, stable businesses. And these nine people have one thing in common: before the AI era, they were already one-person companies. The one-person company itself was already making money, the thing itself already worked, and using AI just made it more efficient, that’s all. We basically meet up in person every three months, and across this year’s three meetups, everyone’s business has grown really fast. Going from first trying to use AI to power their business, to now taking on projects entirely with AI, and even starting to pick and choose projects, took only half a year.

Now, on the flip side, the OPCs that came after AI — it’s not that none of them can support themselves, it’s that the model is pretty narrow. It’s doing self-media and collecting ad money from traffic. The top few AI bloggers basically all monetize through ads. They all have teams, but ads are their only revenue stream. What they do is use AI to write articles, then iterate based on traffic feedback. The result is content that panders to traffic. Reading their articles is like watching a weathervane — one moment this one is amazing, the next moment the sky is falling. There’s no information value at all. Some don’t even test things themselves; they just rewrite other people’s articles. Traffic peddlers like this are poison for the whole industry.

Then there’s the training business. People who can actually do AI implementation for businesses — the ones who’ve succeeded are far too few. And even when there’s traffic and there’s demand, there aren’t reliable people to scale up how much of that business can be taken on.

So why is getting business hard? Because AI OPCs have no division of labor. That means everyone is competing with everyone. If it’s a really complex job, almost nobody can do it. If it’s an ordinary job, everybody can do it. Even the client thinks, I can do this too. Today there was an image-generation gig, fifty yuan per image, a few hundred images, and the client thought the quote was too expensive, right? Why? Because they can generate images with Codex themselves, the compute cost is actually tiny, the images aren’t that hard, so naturally they don’t want to pay more.

Things like image generation and animated comic dramas, where the models are really strong and whoever does it delivers an eighty-out-of-a-hundred result — when a gig comes in, everyone can do it, everyone is a competitor, there’s no cooperation, so it turns into a price war. You’ll take it for a hundred, I’ll take it for fifty, I’ll take it for thirty, I’ll take it for twenty, I’ll do this one for free and you pay me on the next one.

Recently there’s also a new concept called OPT — one-person companies forming a team to take on business. But that’s an even bigger false premise. Its assumption is that everyone has their own area of strength, right, so together they can complete the puzzle and take on the work better. But think about it the other way: the reason one-person companies are possible now is precisely that AI has filled in those areas. The places where people can do much better than AI are very few. So an OPT put together like this ends up with overlapping, redundant capabilities, plus communication overhead, and the profit a project brings in actually isn’t enough to split. And the OPT isn’t much more capable than each individual OPC.

Take it a step further: there’s no such thing as a channel partner who has no AI implementation ability but can keep bringing in business nonstop. Even if they could get that kind of business, they wouldn’t feel comfortable handing it to an OPC they don’t know well. The people who can really deliver a project well are far too few. And those people basically aren’t short of work — they’re basically the ones passing work out to others.

Take me, for example. I really like doing projects, I don’t like teaching. My group pushed me for a month before I finally scheduled one small class. And no matter how efficient I am on projects, once I’m running three in parallel, it’s hard to take on any more. So what I usually do is recommend the work to partners I think are relatively reliable — that’s the only kind of collaboration I do. And they have to be people I know inside out and trust, basically longtime members of my community I’m very familiar with, and I need a deep understanding of the projects they’ve done before I’ll hand a project over. If I wouldn’t feel comfortable handing it over, I won’t give it.

Why not just pass it along to anyone? Because that wouldn’t be responsible to either side. If the client hasn’t even figured out what they want and I just pass it on, it might not be a solid opportunity in the first place. Clients who haven’t figured things out themselves, or who won’t pay the collaboration fee in advance — OPCs who are actually capable won’t consider them. Then delivery is painful for them, and painful for me too. Some clients are also reluctant when it comes to paying, and there’s no point recommending those jobs to anyone else.

An OPC actually doesn’t need another OPC, right? So what do they need? A partner who can keep giving them work — that’s small and medium-sized businesses. Why small and medium-sized businesses? They need to cut costs and raise efficiency, but their existing teams can’t get trained and master doing the work with AI that quickly. That’s where OPCs can fill the gap. So these OPCs should work closely with small and medium-sized businesses. On one hand, OPCs can take on mature, established work more flexibly; on the other hand, the businesses can cut some of their staffing costs. And sometimes OPCs still need a backstop — for example, when they can’t solve something technically, someone needs to step in and help. A healthier model is many-to-many — many small and medium-sized businesses working with a group of OPCs, rather than OPCs forming some kind of community of their own. That’s it for today. See you tomorrow, bye.