The most common reason people don't build the thing they have been thinking about is not that they lack the ability.
It is that they're waiting to be ready, and ready isn't a state that arrives.
What people think they are waiting for
Usually one of four things.
More skill. Once I understand this properly, I will start. But understanding comes from building. You cannot read your way to competence in something practical; the reading only makes sense once you have hit the problem it describes.
A better idea. The one I have is too small. Small is the correct size for a first thing. Small means you finish it, and finishing is the skill.
Someone's approval. This is the big one and the most invisible. People wait for a manager, a mentor, an institution, some authority to say yes, you may. In most of the situations I am describing, there's no such person. Nobody's job is to authorize you.
Confidence. Waiting to feel confident before starting inverts the actual order. Confidence is downstream of having done it. You get it after, as payment, not before as a grant.
What you actually need
Two things, and neither is credentials.
A real problem. Not a market opportunity. A specific thing that's specifically annoying to specific people, ideally including you. If you can't name the person who has this problem, you don't have a problem, you have a concept.
A willingness to serve someone with it. This is the part I think gets left out of most advice about building things, and I think it's doing more work than any of the rest.
If you are building to be impressive, you will stop the first time it is boring, which is roughly week two. If you are building because a specific person is struggling and you can help, you'll finish, because they're still struggling and that fact does not care about your motivation.
A real problem and a servant's heart. That is genuinely the whole list.
The gap has closed
What makes this more true now than it was is that the distance between an idea and a working thing has collapsed.
I teach students who build functioning applications in a semester with no computer science degree. Not because they're exceptional, though some are. Because the barrier that used to filter people out is mostly gone.
That means the excuse has gotten thinner. Ten years ago "I wouldn't know how" was a real answer. Now it is a description of a week's work.
What starting badly actually looks like
I want to be specific here, because "just start" is useless advice on its own.
Your first version will be worse than you imagined it. It will handle one case and break on the second. It will look wrong. You will show it to someone and they will find a problem in eleven seconds that you didn't consider.
That isn't failure, that's the process working. Every one of those is information you couldn't have obtained by planning longer.
The people who succeed at this aren't the ones who plan well. They are the ones with a high tolerance for their own early work being bad, and who keep going anyway because someone is waiting on the other end.
The thing underneath
I think a lot of waiting-to-be-ready is actually fear wearing a respectable costume.
Preparing feels responsible. It looks like diligence. And it's infinitely extendable, which is its real appeal, because as long as you're preparing you can't yet have failed.
Starting exposes you. Your thing might not work. People might see that you didn't know what you were doing, because you didn't.
The uncomfortable truth is that everyone who has built anything went through exactly that, and the only difference between them and the people still preparing is that they were willing to be visibly not-ready for a while.
Where to actually begin
Pick the smallest real problem you know of. Preferably one that belongs to someone you can talk to.
Tell them you want to try to fix it and that your first attempt will probably be bad. This is a surprisingly effective sentence. People are generous with someone who is honest about where they are.
Then build the worst version that helps at all. Show them. Fix what they tell you is wrong.
That is it. That is the loop, and it does not require anyone's permission, which is fortunate because none is coming.
Whatever you have been circling for the last year: what would the worst useful version of it look like, and could you have it by Friday?