Co-Design-Canvas Project Main Image

Studio-In-Flux

Project Type Artificial Intelligence
Role Team - Manas Bhatia, Vaibhav Jain
Tools Used Python, LLM, Hugging Face, Agent-based Modeling
Course A6956-1 / Spring 2025 / Spatial AI
Course Instructors William Martin, Sebastian Schloesser (TA)
This project investigates the potential of 3D and physical AI to facilitate insights, decisions, and predictions for problems involving higher-level spatial reasoning.

By leveraging open source AI models, we developed a chain-of-inference system that can analyze spatial patterns, simulate agent behaviour, and recommend optimized spatial configurations.
About

Our team explored how semantic models, computational structures, and artificial intelligence can together form a reasoning pipeline that reads space not just as geometry, but as behavior, constraint, and opportunity. Grounded in real-world contexts but abstracted through spatial theory, we built a prototype AI system that reflects and tests our spatial hypothesis through simulation and inference. This work is both a technical experiment and a conceptual proposition for how AI might help designers observe, interpret, and intervene in complex spatial environments.