Indie Developer Builds GTA-Style Sandbox From Public Map Data
A solo developer running an AI agent overnight is now attempting a global-scale open world, which puts the cost of map-scale sandbox experiments far below what the format used to demand.
Reporting from 1 source: Game Spark.
Charlie Chapman posted on X that he is developing a GTA-style game built from real, public-domain map data, aiming for a global-scale sandbox. The world has a weather system with rain, wind and storms, procedurally generated building interiors, and working vehicles including skateboards, snowboards, airplanes and rockets. A photo mode is implemented. Much of the work is done by AI: Chapman checks the progress of Claude Opus 5.5 on three Macs at home before bed and again the next day. The project has not been released.
The workflow is the part worth watching. Chapman reviews what Claude Opus 5.5 produced, playtests it, and leaves instructions for the next task before going to bed; the following day he checks the results. That loop runs on three Macs at home.
What comes out is a sandbox built on public-domain map data, with weather, generated building interiors, and travel by skateboard, snowboard, airplane or rocket. There is a photo mode. Nothing has been released to the public, and updates go out on his X account as they come.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.
Sources
- Game Spark 地球規模のGTA風ゲーム開発中。実際の地図データでオープンワールドを構築