Simple
Garments

Fixed media couldn't convey the living nature of space.

But expression alone cannot mediate architecture–communication. Designers speak, but spaces don't respond

Creating a tool rather than static representation will enable AI-mediated communication.

Simple Garment integrates live motion tracking, speech recognition, and generative AI into a unified real-time web-based 3D system.

WebSocket and WebSpeech APIs capture gestures and voice, which are translated by multi-stage language models into spatial instructions.

Python pipelines then generate 3D geometry in real time, turning architecture into an interactive interface for human-AI collaboration.

User first roam around the space, and then start to speak with the LLM to find what they want for the space.

The spatial instructions will be sent to the backend server, and the server will generate the 3D geometry based on the instructions.

The 3D geometry will be re-rendered in the frontend, and the user can see the changes in real time.

Various spaces can be generated by the user's movement and speech.

Such as Korean Traditional Park with diagonal path, or SF-like space with trees and buildings.

Asking for a open, and safe place also creates a space corresponding to the user's request.

From object to interface: architecture as a partner, enabling AI-mediated collaboration rather than one-way authorship

Bridging HCI patterns with spatial reasoning, it integrates gesture, language, to generate Client-centered way of designing.