This is an interactive demo. Click "Run Demo" on any workflow to see the VRAM orchestration in action.

Skills — ComfyUI Workflows

Manage your generative media workflows — the "Big 7" classification system

VRAM Master — Intelligent GPU Orchestration

SpiritBox's VRAM Master automatically manages your GPU memory. When you generate an image or video, it unloads the LLM to free VRAM for ComfyUI, then reloads it when generation completes. No manual management required — click "Run Demo" below to see it in action.

P1 — Logic
LLM Brain (swappable)
P2 — Action
ComfyUI (takes priority)
P0 — Ambient
Voice (always resident)
Execution Log
Waiting for workflow execution...
How It Works
Upload Workflows
Drop any ComfyUI workflow JSON into SpiritBox. The Workflow Architect automatically maps inputs and classifies it.
AI-Controlled Nodes
Mark which nodes your AI can control — prompts, seeds, steps, dimensions. The AI fills in the rest.
Chat Integration
Just ask your Spirit Guide to "make me an image of..." and it automatically selects and runs the right workflow.
100% Local
Every pixel is rendered on your GPU. No cloud API. No per-image fees. No usage limits. Your hardware, your creations.
Back to Overview