Member-only story
From Sketch to Unity Prototype: Comparing Claude, Codex and Bezi
This is the continuation of my previous test. This time, I’m not doing a “build this game” shit test with AI and letting it figure out the rest. This time, I’m actually giving more context and more things to work with.
What I’m trying to achieve here is to find a good workflow so my teammates at Roshka Studios, especially the non-developers, can turn their ideas on paper into a working and playable prototype.
If you can’t read the article because of paywall then please click here
This would give the whole team a better tool to pick up ideas and discard the ones that, in the end, turn out to be “not that fun”. For me, greenlighting these projects is very expensive, because the team can lose a lot of time trying to make something fun when it simply isn’t, or when it only looked interesting on paper.
It would also help a lot when pitching to the publisher, because it makes it easier for them to understand what we are trying to build if they can actually play it and see it in action.
One more thing: right now I don’t care that much about the quality of the generated code. The main goal is to have a playable version…
