You've got 42 prompts
The old startup doctrine was simple: build fast, grow fast, raise capital, sell the company. There is a second option, and it has never been this reachable. Build to print. Not shares, not headlines, not imaginary future wealth. Money. For you and your team. What remains is the question no machine will answer for you.
Valuation doesn't matter if you're printing
A valuation is not a number. It is a claim about the future that somebody else has to agree with. It exists only as long as the next buyer carries it. When sentiment drops, the number drops, and you are holding nothing that feels independent of that.
Revenue is the opposite. It is either there this month or it isn't. It needs no agreement, no round, no single moment where everything is decided. It is boring, and that is precisely its property.
The difference is not merely accounting. It decides who you are building for.
Build to sell and you optimise for the buyer's taste: growth curves, metrics that look good in a deck, a story that fits a fund's thesis. Build to print and you optimise for the customer's taste. Those are two different products, and it shows.
Software that generates real revenue, serves real customers and funds the people building it is a machine. It has an efficiency, it has wear, it can be tuned. You can understand it.
A lottery ticket has exactly one property: it wins or it doesn't. In between there is nothing to do but wait and maintain the story.
The difference shows on the day something breaks. A machine can be repaired. A ticket can only be bought again.
The machine's second property is unglamorous and therefore easy to miss: it pays the people building it while it is being built. That sounds like bookkeeping and is in fact the whole of the freedom. If you pay yourself, you do not need to convince anyone before trying the next thing. You do not have to wait for a quarter to close, and you never have to drop an idea because it is hard to pitch.
That is the actual return. Not the amount, but the decoupling from somebody else's approval.
The bottleneck was never quality
For decades, average software was entirely sufficient. Not because people liked bad software. Because the better one did not exist.
Building was slow and expensive. That kept the number of vendors in any niche small, and the customer had three options: take the mediocre one, take nothing, or build it themselves, which they could not afford. They took the mediocre one.
What they tolerated is familiar to everyone: interfaces that need to be explained. Workflows shaped around somebody else's company instead of their own. Missing features they waited for. And the sentence that this is simply how it is.
The benchmark was never “good”. The benchmark was “less painful than the alternative”, and the alternative was usually nothing at all.
When almost everyone can build, building is no longer the difference
That is the uncomfortable half. Anyone whose position rests on being able to produce software is losing it right now. Not in ten years, but in the time it takes a competitor to learn the same tool.
The good half follows immediately: when building is no longer the difference, the result becomes the difference. Software that is faster, sharper, simpler, more useful and more enjoyable than the sea of average products around it.
For the first time in a long while, that is good news about software. The industry is entering a race where being genuinely great pays. For forty years that was a luxury for companies who could afford it. Now it is the entry ticket.
It also shifts which skill is scarce. Not typing, not architecture, not knowing a framework. What becomes scarce is judgement: seeing what is missing, deciding what can go, and noticing when something is done.
The bottleneck moves from production to decision. Defending the old scarcity means defending an empty position.
| Scarce before | Scarce now | Why |
|---|---|---|
| Someone who can build it | Someone who knows what it should be | Production got cheap. Selection did not. |
| Developer time | Attention to detail | Where everyone can ship, execution decides. |
| Feature count | The nerve to leave things out | Building more costs nothing now. Building less does. |
| Access to tools | Access to customers | Everyone has the tool. Not everyone has distribution. |
| Speed | Direction | Fast in the wrong direction is now very fast. |
Software building has become a real-time strategy game
Look at what you actually do all day. You allocate limited resources. You explore new technologies. You identify markets. You recruit people and agents. You construct systems, defend your position and react to competitors in real time.
It is Civilization, except the map is the economy. It is an RTS, except the units are developers, AI agents, distribution channels, products and prompts.
The analogy is not just pretty, it predicts something. Because anyone who actually plays these games knows how you lose them, and it is always the same three ways.
You lose by expanding without an economy. You lose by teching without defending. And you lose by holding an opening far too long on a map where it does not work.
The game is accelerating. The cost of experimentation is collapsing, the number of possible moves is exploding, and a small team can now attempt things that previously required an entire company.
That sounds like pure upside and isn't. When moves get cheap, choosing the moves becomes the entire craft. A player who makes every available move is not playing well. They are just playing loudly.
Where selling and capital are still right
A thesis with no limit is an ideology. So here is the limit, and it follows a single question: does the path to the first paying customer cost more than the first paying customer can bring in?
Where the answer is yes, capital is not a detour but the only fitting instrument. That holds for anything with a long lead time and a high stake: hardware, medicine, infrastructure, regulated markets with approval cycles, research that pays out in years.
It also holds where the market ultimately carries only one player. If speed of capture genuinely decides everything, calm profitability is the wrong weapon, and choosing it means losing with healthy numbers.
And there is the hardest limit, which has nothing to do with capital: printing requires somebody who pays. If you have no distribution, building more does not fix it. It only means missing faster.
What are you going to build?
Suppose you had 42 prompts left today. Not infinite, not zero. A number large enough to build something serious, and small enough that wasting it hurts.
This is not a frugality exercise. It is an accurate description of your situation. The tools are here. The limit sits somewhere else.
Forty-two is the most famous answer in literary history, and it is famous for exactly one reason: nobody knew the question. A computer ran for seven and a half million years, produced a perfectly correct answer, and it was useless because the question was missing.
That is precisely where you stand. Prompts are answers. They have become cheap, they will keep getting cheaper, and without the question they are worth nothing. Scarcity has moved from the answer to the question, and that is the only shift in this entire piece that counts.
The winners will not be the companies with the largest teams, the most funding or the highest valuations. They will be the ones who know what they are aiming at.
This piece is the sequel to “Build it yourself”. That one is about whether you can build, because the cost calculation against off-the-shelf software has flipped. This one is about what you do with it.
This piece is a position, not a data analysis. It deliberately carries no market figures, because the thesis needs none: it can be checked against your own working week. The reference in section 07 is to Douglas Adams, The Hitchhiker's Guide to the Galaxy (1979).
