Looking for collaborators — PhysicClaw-VEA
We’re building an interactive 3D app featuring a virtual entity
that reacts in real time through shaders, emotions, and voice — powered by real AI.
Stack: React 19 · Three.js WebGPU · Supabase · TypeScript · Vite
What we’re working on:
Multi-user auth + persistent 3D cloud scenes (v2.0)
We need help with:
• Backend / PostgreSQL / RLS (Supabase)
• 3D Frontend (React Three Fiber, custom shaders)
• AI / LLM integration (OpenClaw · Gemini)
Repo: GitHub - yomero243/PhysicClaw-VEA: PHYSIC-CLAW-VEA is a real-time, interactive visualization environment designed to give a digital "body" to the Soul of an OpenClaw agent. The project transcends the conventional chat interface, transforming the agent's activity logs, karma, and identity into a dynamic 3D entity that resides on the web. · GitHub
Interested? Reply here or DM 
This is a pretty ambitious concept, giving an AI agent a “body” instead of just a chat interface is a big shift.
Turning logs, state, or “emotion” into a visual, reactive 3D entity is where things start to feel more alive and less like a tool. If done right, it could make interactions way more intuitive, especially if users can see changes instead of just reading them.
The stack also makes sense for what you’re aiming at. React Three Fiber + WebGPU opens the door for more complex visuals and effects, and Supabase for persistence + multi-user scenes is a solid foundation if you’re thinking shared environments.
The real challenge here is going to be cohesion. You’ve got a lot of heavy pieces:
real-time rendering
AI/LLM behavior
voice + emotion mapping
multi-user state
Getting all of those to feel unified instead of disconnected systems is where the project will either shine or fall apart.
Also really interesting angle with “karma, identity, and activity logs” driving the visuals. Curious how literal that mapping is. Like, are emotions shader-driven (color, distortion, motion), or are you going deeper into animation and form changes?
From a collaboration standpoint, it might help to show a short clip or live demo of the current state. Projects like this are easier to rally people around when they can see the entity reacting in real time.
Overall though, this is pushing into that space where AI + real-time graphics actually start to merge into a new kind of interface. Definitely not a small project, but a very interesting direction.
1 Like
Thanks a lot for the feedback! Right now, I’m hitting a bit of a technical snag because my background isn’t in backend architecture. But I’m still totally committed to making the system work, and I’m working on a plan to get around these infrastructure issues.
This project started as a custom version For my personal OpenClaw entity aiming to give the entity a unique physical shape using rigging and 3D meshes. Initially, I planned to use a third-party API like Mixamo for animations and skeleton management, so users could have more options. But since we can’t do direct external integration, I’ve had to switch gears and set up our own S3 bucket to handle our assets instead.
Now, the goal is to build a persistent scene that serves as the main hub for “Claw.” I’m not just looking for a model viewer—
I want an architecture that lets multiple entities (like adding MOLTBOOK) live together in a 3D environment, all working smoothly. We’re moving from just a tool to a full-on living environment.
Here are a few brief clips illustrating how the entity may select its own assets based on its soul.md.