An agent that builds a Minecraft town from a vague description. It draws a concept image, works out zoning and building plans, then converts the image in two steps, putting the structure into language before turning it into coordinates, which keeps the result coherent. The frame goes up in bulk; a bot adds decoration on site.
Judges liked the choice of Minecraft building and the original techniques behind the 3D generation. They noted that repeated trial and error had brought it to a practical level, and that it connects children's creativity to skills for an age of AI.
The heart of it is probably the detour. Instead of placing blocks straight from an image, the system writes the structure out in language and only then converts to coordinates. A generated picture carries no dimensions or internal consistency, so that intermediate representation is what keeps the build from falling apart. It is also easy to explain, which tends to play well in judging. Separating bulk framing from on-site detailing by a bot matches what each job needs.
See all winners of 第4回 Agentic AI Hackathon with Google Cloud →
Summaries are written by this site. Project pages include demo videos, screenshots and the team's own write-up (videos may autoplay).