Idea to game, without the part in between

Everyone who has ever had a game idea has run into the same shape of wall. The idea is clear. What it would feel like is clear. And then between that and anything playable sits a body of work — a language, an engine, a build pipeline, a collision bug at eleven at night — that has no relationship to the idea at all.

It is not a filter for good ideas. It filters for people who will tolerate learning a language, which is a completely different trait. A lot of good games do not exist because the person who thought of them was not that person.

That is the gap Flockbay works on. Not the having of the idea, and not the making of it good. The mechanical stretch in the middle that turns most ideas into a note in a file somewhere.

More on this: AI survival game maker with real terrain and weather

An idea is not a specification, and being honest about that is important. "A game about a lighthouse keeper who cannot leave" is a feeling, not a set of rules. Somewhere between that and a playable thing, decisions have to get made about what the player does with their hands for the next hour.

The useful move is to make the smallest concrete version of the feeling and go and stand in it. Build the lighthouse and the rock it sits on. Put the sea round it and turn the weather up. Walk to the edge. Ten minutes in a real space tells you more about whether the idea works than an hour of writing about it.

In Flockbay that first step is done by hand, in the builder, and what it consists of depends on the question a new project opens with: 2D or 3D, two pictures, no default. A 3D idea gets terrain, water, weather, buildings and characters, placed directly. A 2D idea is standing up before you have done anything at all — sky, a band of scenery along the horizon, ground with a ledge cut into it, a platform off to one side, and a person who walks and falls and jumps under the keys you are already pressing — and you take it somewhere by painting more ground and arranging sprites on it.

Neither is something you ask for in a sentence, and that is deliberate. The shape of a place is a taste decision and you are the one with the taste.

More on this: The creator in AI game creator is you

Once the place exists, say what happens in it. The light has to be relit every twenty minutes. The storm gets worse over the hour. Something in the water reacts to the light. Flockbay writes the GDScript that makes those true, inside your Godot project.

Press play. It runs locally on the copy of Godot in the app. And the version in your hands will be different from the version in your head, always, because ideas do not survive contact with input and timing. Twenty minutes is too long. The storm arrives too early. The thing in the water is invisible and might as well not exist.

You say so, and it changes, and you play again. That is the whole process. Not writing the code does not mean not making the decisions — the decisions are all that is left, and they are the ones that determine whether the idea survives.

More on this: An AI game generator from text, and what text is actually good at

It will not turn a sentence into a finished game while you watch. Nothing does that, and anything that appears to is handing you a template with your nouns swapped in, which you will recognise as not being your idea within a minute.

It also does not have taste. It will write exactly what you asked for, and quite often what you asked for is not what you wanted, which you only discover by playing it. That discovery is the job and it is not delegable.

The practical facts: desktop app for Windows, Godot bundled inside. Building needs an internet connection; playing what you built does not. Flockbay is free on your own AI: your Claude or ChatGPT plan, your own agent, or an API key. Flockbay builds from a library of ready-made objects, or with Godot's own shapes.

More on this: What one sentence can generate, and where it runs out