A spatial computing framework centered on gesture and context, with large language models inside the environment rather than beside it.
Question
Most LLM interfaces are rectangles. Spatial computing keeps putting language in a panel next to the world. The question is whether the model can live in the room — answering through gesture, context, and proximity rather than a chat stack.
Constraint
Input is the body in a place, not a prompt box. The system should be context-aware enough that the user does not have to narrate the scene it can already see.
Method
A unified WebXR framework for gesture-driven, context-aware human–AI interaction, investigating the integration of advanced large language models (Gemini) inside WebXR environments so the model can participate in an immersive scene rather than caption it.
Artifact
A comprehensive research proposal and experimental environment for spatial human–AI interaction.
Authored as independent research toward a spatial computing agenda.