Playco reports 50% fewer manual fixes in game prototypes built with GPT-6 Astra

Playco used GPT-6 Astra to create three themed game prototypes from a single grey box foundation, a simplified version built from basic geometric elements.

Most prototypes worked on the first attempt. Playco said only the cyberpunk version required a performance fix; the others came together without another iteration.

Three worlds from one foundation

The team first built an unthemed prototype, then made several gameplay and creative adjustments. From that base, it developed three playable worlds that the company said GPT-6 Astra produced in one go.

Fewer manual fixes than before

With the previous model, Playco’s initial grey box was less polished. Repeated prompts to correct it eventually became counterproductive, forcing engineers to make manual changes. Playco reported 50% fewer manual fixes with GPT-6 Astra.

The company also saw improvements in spatial reasoning, reference-image recreation, responsive interfaces in Unity and game feel. Playco said the model found bugs more easily and identified opportunities to improve the player experience because Playbot can run a game and validate its own changes.

GPT-6 Astra is being used in Playbot, an AI-powered IDE for professional game developers. The tool connects directly to Unity and Godot, allowing models to edit scenes, run and test games, validate changes and work in parallel inside the engines.

Joao Vieira, Playco’s lead product engineer, said the approach makes it possible to turn more ideas into playable prototypes that developers can compare rather than only imagine.