Arturo Nunez demonstrates a browser based game engine in which a user asks an assistant to add characters, buildings, rain, animation, movement, and a following camera. The creator works in the language of game design while the engine handles assets and implementation details.
Arturo Nunez argues that conventional engines expose repeated boilerplate and specialist concepts that make small game projects difficult. Nereu instead describes assets with reusable tags, then lets systems query those tags to apply movement, physics, rendering, and game rules across many scenes.
Arturo Nunez explains that the assistant receives scene context, game intent, available tools, and relevant nearby assets. A level of detail inspired context strategy limits what reaches the language model, while vision models help tag thousands of assets. His goal is to help people learn game design by creating and iterating, rather than asking AI to produce complete games in one step.
Watch on YouTube
