Playco Reduces Manual Fixes by 50% in Game Prototyping with GPT-6 Astra

Game development startup Playco significantly improved efficiency in the game prototyping process by leveraging GPT-6 Astra. Playco is developing Playbot, an AI-based IDE for professional game developers, which connects directly to engines like Unity and Godot to allow AI to edit scenes, test games, and verify changes.

Compared to previous models, GPT-6 Astra features enhanced spatial reasoning and vision capabilities, handling in-game element placement and UI responsiveness more accurately. The Playco team created a grey box-based circular prototype and, based on it, generated 3 game prototypes with different themes in one go. Most prototypes worked immediately on the first attempt, with only the cyberpunk version requiring performance fixes.

Notably, manual fixes decreased by 50%. In previous models, the low completeness of initial prototypes made correction via prompts inefficient, but GPT-6 Astra autonomously identifies bugs and areas for improving player experience, significantly reducing manual developer work. This enabled developers to build an environment where they could actually play and compare all 10 ideas.