automate
Node Agent
Build complete AI workflows from a single sentence. Describe what you want, and the agent builds the nodes, wires the connections, picks the models, and explains what it made.
Key benefits
Builds full workflows in seconds
Type one instruction and watch the agent place every node, connect every wire, and configure every parameter on the canvas in real time.
Edits existing workflows on prompt
Open any workflow you already have and ask the agent to change it. Swap a model, add an upscale step, rewire a section, or insert an If/Else branch - all without touching a node manually.
Wires in your reference images
Drop images onto the canvas before prompting. Label each one by role - character reference, style reference, conditioning image - and the agent connects them to the right nodes automatically.
Picks from 500+ models
The agent has access to every model in the Scenario catalog: image, video, 3D, audio, and your custom-trained models. Name a model to use it; say nothing and the agent chooses a sensible default.
Reads your workspace
The agent sees the workflows you have already built, the models you use most, and the settings you prefer. That context shapes every graph it produces.
Explains every graph before you run
After building, the agent writes a plain-text report naming every node, the model behind each generator, the resolution it chose, and how data flows between steps.
How it works
- 1
Open Workflows
The agent's chat panel opens alongside the canvas.
- 2
Describe your workflow
Type what you want. Be specific about inputs, the steps in between, and the output format. Drop reference images on the canvas if style or composition matters.
- 3
Watch it build
The agent places nodes, connects them, and configures parameters on the canvas in real time. When it finishes, it writes a short explanation of every node and how they connect. Refine in conversation Each follow-up message edits the same workflow. Swap a model, add a branch, change the resolution - iterate in plain text instead of dragging nodes manually.
Use cases
Game asset production
Generate game icons, then ask for sprite sheets, then add rarity variants - all in the same workflow, all from follow-up prompts. Route asset types to different style LoRAs just by describing the condition.
Product photography
Start with image generation, background removal, then add an upscale pass, then wire in a conditional branch for smaller images - each change is one sentence. The workflow grows with the conversation.
Character design
Drop a reference image, generate a character, then swap the style reference, then change the aspect ratio, then add a post-processing step. Every edit is a follow-up, not a rebuild.
Concept art to 3D
Turn a 2D image into a textured 3D model and a turntable render, then go back and adjust the render resolution, add a post-processing pass, or swap the generation model - all from the same chat thread.