Hanzo Studio - Visual AI Workflow Engine

Hanzo Studio is a visual node-based AI workflow engine for building, testing, and deploying AI pipelines.

Overview

Hanzo Studio is a visual node-based AI workflow engine for building, testing, and deploying AI pipelines. Fork of ComfyUI with Hanzo branding, custom nodes, and cloud deployment. Live at studio.hanzo.ai.

Why Hanzo Studio?

Visual workflows: Drag-and-drop AI pipeline builder

Custom nodes: Extend with Python — any model, any tool

API mode: Run workflows programmatically via REST API

Self-hostable: Docker, K8s, or bare metal

White-label: Full branding customization

OSS Base

Fork of ComfyUI (comfyanonymous/ComfyUI). Repo: hanzoai/studio.

When to use

Building visual AI workflows (image gen, text processing, pipelines)

Creating custom AI node integrations

Running ComfyUI workflows with Hanzo branding

Deploying visual AI tools for non-technical users

White-labeling a visual AI platform

Hard requirements

Python 3.10+ with pip

Docker for containerized deployment

Port 8188 available

Quick reference

Item	Value
UI	`https://studio.hanzo.ai`
Port	8188
Image	`ghcr.io/hanzoai/studio:latest`
Repo	`github.com/hanzoai/studio`
Branch	`main`
Upstream	`comfyanonymous/ComfyUI`

One-file quickstart

Docker

docker run -d --name hanzo-studio \
  -p 8188:8188 \
  --cpus=1 --memory=2g \
  ghcr.io/hanzoai/studio:latest \
  --cpu --listen 0.0.0.0

API mode (run workflow)

curl -X POST http://localhost:8188/prompt \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": {
      "1": {
        "class_type": "KSampler",
        "inputs": {
          "seed": 42,
          "steps": 20,
          "cfg": 7.0,
          "sampler_name": "euler",
          "scheduler": "normal"
        }
      }
    }
  }'

Core Concepts

Branding Approach

CRITICAL: Never do blanket sed 's/ComfyUI/Hanzo Studio/g' on minified JS — it breaks class definitions, property assignments, and dynamic imports.

Use branding/patch_frontend.py (Python, context-aware):

Only touches display strings, URLs, and quoted contexts

Logo SVGs replaced directly (comfy-logo → hanzo-logo.svg)

Favicon generated as ICO with H mark bitmap

Custom Node Development

# custom_nodes/hanzo_inference.py
class HanzoInference:
    @classmethod
    def INPUT_TYPES(cls):
        return {
            "required": {
                "prompt": ("STRING", {"multiline": True}),
                "model": (["zen-70b", "zen-32b", "zen-14b"],),
                "temperature": ("FLOAT", {"default": 0.7, "min": 0.0, "max": 2.0}),
            }
        }

    RETURN_TYPES = ("STRING",)
    FUNCTION = "inference"
    CATEGORY = "Hanzo AI"

    def inference(self, prompt, model, temperature):
        import requests
        resp = requests.post("https://api.hanzo.ai/v1/chat/completions",
            headers={"Authorization": f"Bearer {os.environ['HANZO_API_KEY']}"},
            json={"model": model, "messages": [{"role": "user", "content": prompt}],
                  "temperature": temperature})
        return (resp.json()["choices"][0]["message"]["content"],)

NODE_CLASS_MAPPINGS = {"HanzoInference": HanzoInference}
NODE_DISPLAY_NAME_MAPPINGS = {"HanzoInference": "Hanzo AI Inference"}

Kubernetes Deployment

apiVersion: apps/v1
kind: Deployment
metadata:
  name: hanzo-studio
spec:
  replicas: 1
  selector:
    matchLabels:
      app: hanzo-studio
  template:
    spec:
      containers:
      - name: studio
        image: ghcr.io/hanzoai/studio:latest
        args: ["--cpu", "--listen", "0.0.0.0"]
        ports:
        - containerPort: 8188
        resources:
          requests:
            cpu: 250m
            memory: 512Mi
          limits:
            cpu: "1"
            memory: 2Gi

White-Label

Fork hanzoai/studio

Edit branding/patch_frontend.py with your logo/colors

Run python branding/patch_frontend.py during Docker build

Deploy with your domain

Troubleshooting

Issue	Cause	Solution
Broken class names in UI	Used sed on minified JS	Use patch_frontend.py only
OOM on large workflows	Insufficient memory	Increase K8s memory limit
Custom nodes not loading	Wrong directory	Place in `custom_nodes/`

hanzo/hanzo-engine.md - Rust inference engine for backends

hanzo/hanzo-flow.md - Alternative workflow builder

hanzo/hanzo-chat.md - LLM API for custom nodes

Hanzo Studio - Visual AI Workflow Engine

On this page