Pick an idea worth the months

The fastest way to waste a month of building is to build the wrong thing well. Now that the build is fast, this is the dominant risk, not a footnote. The constraint used to be can-I-build-it, which quietly filtered out bad ideas because the cost of building was high enough to make you think twice. Strip that cost away and nothing stops you racing in the wrong direction for weeks. Before any tooling, any stack, any prompt, you owe yourself an honest pass at the idea. Three tests, all of which have to hold.

First, it has to be personal. You should feel the pain yourself, because that is the only way you will know whether a feature is right without running a survey. When you are the user, you have a working compass. When you are guessing at someone else's problem, every decision is a coin flip and you will not know you got it wrong until much later.

Second, it has to be a genuinely good idea, not a clever-sounding one. A surprising number of apps are built on premises that fall apart the moment you say them out loud to someone who would have to pay. Pressure-test the core before you fall in love with it. The cost of a bad idea is not the build, it is the months you spend polishing something nobody wanted.

Third, it has to have a market. Personal pain plus a good concept still fails if you are the only person on earth who has the problem, or if the people who have it will never pay. You are not looking for proof here, just a credible reason to believe a paying audience exists. A short pass through customer research before you open an editor is the cheapest insurance on the whole list, and pairing it with a clear go-to-market strategy keeps you honest about whether anyone will actually buy.

This is lesson zero, and it is easy to skip because it feels less productive than opening an editor. Resist that. An afternoon spent killing a weak idea is the cheapest afternoon you will ever spend, and it is the one decision the AI cannot make for you.