Install, open a template, run it, edit it, ship it.
Step 1 — Install
Requirements
| Component | Minimum | Recommended |
|---|---|---|
| RAM | 8 GB | 16 GB+ |
| Disk | 10 GB free | 50 GB+ for models |
| GPU | None | 8 GB+ VRAM for local inference |
| OS | macOS 13+, Windows 10+, Ubuntu 22+ | Latest |
No GPU? Use cloud providers with your keys. See hardware notes.
Install
- Download from nodetool.ai.
- Run the installer.
- Launch. No setup wizard.
Python and inference engines install on first use. Installation guide for platform notes.
Local models (optional)
- Open Models in the sidebar.
- Pick:
- Flux or Qwen Image for images (8–12 GB VRAM)
- Llama or Qwen for text
- Wait for downloads (~20 GB).
Cloud-only: open Settings → Providers and paste a key from OpenAI, Anthropic, or Google. Skip the download.
Step 2 — Run a workflow
Open a template from the Dashboard.
Movie Posters
- Dashboard → Templates → Movie Posters.
- The graph opens: inputs left, agent middle, image generator right, preview last.
- Fill the inputs:
- Title: Ocean Depths
- Genre: Sci-Fi Thriller
- Audience: Adults who love mystery
- Press Ctrl/⌘ + Enter.
Creative Story Ideas
- Dashboard → Templates → Creative Story Ideas.
- Set inputs:
- Genre: Cyberpunk
- Character: Rogue AI detective
- Setting: Neon-lit underwater city
- Run.
Step 3 — Edit
- Save with Ctrl/⌘ + S.
- Change inputs, re-run.
- Click a node to inspect it. Hover an edge to see the data flowing through. Press
Space, search “Preview”, drop one anywhere on the canvas.
Step 4 — Ship as a Mini-App
A Mini-App hides the graph and exposes inputs and outputs only.
- Open the workflow.
- Click Mini-App in the toolbar.
Custom UI? See VibeCoding.
Next
| Goal | Page |
|---|---|
| How workflows work | Key Concepts |
| Full interface | User Interface, Workflow Editor |
| More examples | Gallery, Cookbook |
| Models and providers | Models & Providers |
| Deploy | Deployment |
| Stuck | Troubleshooting, Debugging |