Since the second half of 2025, coding agents have spread through the engineering population at a pace nothing else in this industry has matched, and they are still gaining capability on a monthly cadence. Watching them work produces two reactions at once. The first is admiration. The second is dread, because the agent is smarter than me, knows more and knows it deeper, learns faster, and ships faster, which leaves very little I can claim to be uniquely good at and a shortening runway before the severance package arrives.
I’ve had this conversation with a lot of colleagues, and it converges on the same plan almost every time. Someone has an idea that will obviously find users. No meetings, no boss, no approval chain: build a small tool or a SaaS, wrap some AI around it, put it online, and collect money while sleeping.
The plan sounds good until you look at the numbers. In independent developer surveys, more than 80% of solo products earn under ¥1,000 (roughly $140) in their first year online, and about 60% of them go completely silent within three months of launch. What I want to argue here runs against every instinct a programmer has. If your plan is to vibe-code a product aimed at some niche’s pain point, hoping to make a little money or get lucky enough to stop working, you’ve misread what kind of era this is. This is an era of surplus production and scarce demand.
Nobody Is Short on People Who Can Write Code
The app, the mini-program, the AI tool, the SaaS: structurally, each one is the same business as a bubble tea shop, a flower stall, or a hotpot restaurant. All of them sell a product into a market that already has too many. Doing production in an age of overproduction is a reliable way to work hard for nothing. Scarcity sets price, and that hasn’t changed since the beginning of trade.
Open the app store and search for the thing you want to build. In all likelihood several people already built it, most of them are dead, and one is barely breathing. Search GitHub and you’ll find a pile of projects that will always have more stars than yours. Yet 90% of programmers haven’t adjusted their thinking. They spend six months writing code, design a clean architecture, polish the UI, and assume that once the thing is good, users will arrive on their own. Good wine still needs a sign outside the alley, and the alley is longer than it has ever been.
Economics has a name for this: oversupply on the supply side. When the number of producers in a field far exceeds demand, margins collapse. Lower barriers to programming and the spread of AI tooling accelerate the inflow further. The idea you had this morning is being had by a thousand other people on the planet, and a hundred of them have started typing. Nothing about that arithmetic favors you.
For an ordinary person trying to build something in this industry, the first step isn’t typing hello world in an IDE. It’s saying out loud that you are going to do marketing and sales. That sounds like motivational-poster filler, but the logic underneath is brutal. If you can actually sell, a crowd of people will build the product for you. The market is already full of finished products waiting to be picked. You can walk over, take the fruit someone spent two years growing, and put it in your own pocket. You might earn more than they do, because they are still stuck on the production side, still debating tech stacks and polishing details, while you are studying who buys, where they buy, and why they’d buy from you.
A Method That Costs Nothing
Concretely, here is a process that costs zero yuan. Step one: take a sheet of paper and pretend you’re quitting tomorrow to work on this full time. Then write down everything that would have to be settled by then. Who the user actually is, how they solve this problem today, whether they’d pay, who the competitors are, how long the build takes, what servers and API calls cost per month, where the first hundred users come from, which traps you have to avoid. When the list is done, read it to an AI and let it fill the gaps. You now own something valuable: a product spec that exists before the product does. Armchair planning costs nothing and embarrasses no one, and it’s the cheapest way to find out you’re wrong.
One detail matters here. Most people, writing this list, slip into a flattering picture of their own product. Add a question to the list and answer it honestly: name the reason a user would abandon their current tool for yours. If the answer is “mine is easier to use” or “mine is cheaper,” the thinking isn’t finished. Switching costs are higher than you think. Users have to relearn a workflow, rebuild habits, and take a risk. You need a reason strong enough to make them give up all of that.
Step two: take the list and use every competing product in the category. Not skim their homepages. Register an account, run the whole flow, pay where payment is required, take the free trial where it isn’t. Score each one against your list as you go, and write down why the healthy ones are healthy and why the dead ones died.
A common mistake at this stage is stopping after the top three. Widen the search to the top twenty, and include the ones that are neither alive nor dead. The half-dead products carry the most information of all. Something kept them from dying and something else kept them from growing, and working out what limits a product tells you more about a market than any success story does.
Step three: publish what you found. An article, a video, a post, about the products you tested and the traps you hit. Video travels fastest right now. Once you’ve done these three things, none of which cost money, you land in one of three very different places.
Three Ways It Ends
The first ending: nine of the ten products you wanted to compete with are dead. Congratulations, you just saved six months and whatever savings you had. That is money earned. Everyone understands that not spending is the same as earning, and almost nobody acts on it. Most people need to hit the wall before they turn around, and they insist on spending six months proving themselves wrong. Three months of research that avoids someone else’s six-month sunk cost is a good trade at any price.
A friend wanted to build project management software for freelancers. A month of research turned up more than thirty competitors, twenty-five of which had stopped shipping, with the remaining five under a thousand monthly active users. He ran the numbers and saw that even executed perfectly, he’d be fighting for scraps with acquisition costs high enough to eat the entire margin. He dropped the idea and went back to what freelancers actually struggle with, which turned out not to be project management at all but keeping a steady flow of clients. He built a job board for freelancers instead. Six months later it was doing over ¥20,000 (about $2,800) a month.
The second ending: the research changes how well you understand the problem, and the code you write afterward is far better aimed. You know which part users will pay for and which part only you enjoy building. Your odds improve because you’re building the right thing, not because you’re building it well. That kind of understanding doesn’t come from reading competitor analyses. It comes from registering, paying, and feeling the purchase decision from the inside.
A note-taking app is a good illustration. A developer set out to build a note tool for programmers and went through every mainstream product on the market. He noticed that all of them sold organization and search, while the real friction for programmers was speed of capture. Nobody wants to leave flow state to file a thought. They need something that opens in a second and records in three. He shipped a minimal clipboard utility instead, and it picked up more than 2,000 paying users in its first month.
The third ending is the interesting one. Your write-ups attract people: engineers between jobs, people trying to switch fields, small business owners hunting for a solution. You become the person who knows this niche. At that point you can build your own product or not. If you’d rather not, you can distribute the products you reviewed, run the channel, own the traffic. The people who make products end up asking you to push theirs. Most people who get this far stop wanting to write code at all.
Promotion Is Trust
The whole path rests on understanding what promotion actually is. It isn’t persuasion and it isn’t hype. It’s building trust and matching supply to demand. Reviews, analyses, videos: each one adds to your standing as someone who knows a specific niche well. When users believe you understand the field, they come to you for recommendations. When vendors see that you reach a precise audience, they come to you for partnerships. You become the bridge between supply and demand, and bridges collect tolls.
Influencers all end up selling products, and not because selling is easy. Every industry is short of people who can sell. Whatever is scarce is what people pay for, and the scarcity is not on the production side.
From a buyer’s side, the obstacle to any purchase is uncertainty. Users don’t know whether the product solves their problem, whether the price is fair, whether support will exist next year. A trusted recommender removes a large share of that uncertainty, which is why conversion on a creator’s recommendation beats conversion on an ad by a wide margin. The user isn’t buying a product so much as buying insurance against a bad decision.
This also explains how some people build substantial influence out of reviews alone. They aren’t selling products. They’re lowering the cost of deciding, and that has real value that people will pay for.
The Blind Spot
Programmers share a specific blind spot: we solve everything with technical thinking, and there’s a certain arrogance baked into it. The first response to any problem is a script, a tool, a system. Technology is one means of solving a problem and often not the fastest one. A spreadsheet, a group chat, or a single document settles a great many problems that someone will otherwise spend a month building software for.
Starting a company works the same way. Not every business needs code. Validating demand, building a channel, and accumulating an audience require none of it. Do that work first, and the code you eventually write has commercial value instead of being a toy you enjoyed making.
I’ve watched too many programmers spend half a year on a product they consider perfect, then launch into a market where nobody will pay. The reason is almost always the same: they never worked out who would use it, why they would use it, and why they would pay. Those questions don’t get answered in an editor. They get answered by market research, user interviews, and competitor teardowns.
The deeper reason is temperament. Programmers love technology and resist selling. We enjoy flow state and dislike talking to strangers, being rejected, and discussing money. That preference pushes us toward what we’re good at and comfortable with, and away from what actually matters. It’s the comfort-zone trap in its purest form.
Getting out of it requires accepting one fact. The commercial world does not care how elegant your code is. It cares what problem you solve. Users buy solutions, not source code. A rough product that sells beats a polished one that doesn’t, and it isn’t close.
Why This Is the Best Bet
Making money without writing code comes down to moving yourself from the production side to the sales side. In an economy with too much production, distribution is what’s scarce. If you can sell, someone will produce for you. If you can find users, someone will build for them. If you own a channel, everyone with a product will come asking.
This isn’t an argument that technology doesn’t matter. It’s an observation about which skill is scarcer right now. A programmer who can also sell has a far higher success rate as a founder than one who can only write code. If you can learn only one thing next, learn sales.
The return on investment is also better. Technical skill demands continuous reinvestment and decays fast; half of what you learn this year is obsolete next year. Sales is a durable skill. It pays when you sell a product, when you sell a service, and when you sell yourself to an employer.
The path looks roundabout and is in fact the steadier one. For ordinary people, the highest-value position in this economy isn’t production. It’s sales. Every industry works the same way: go where the shortage is. Right now the shortage is salespeople, so go be one.