🤖
Talk to it. Watch it build.
An AI agent that operates the whole app.
▸
Describe what you want in plain language and the agent constructs the node graph in front of you: wiring nodes, choosing engines, running the graph, and suggesting the next move. It's part helpdesk (new here? just ask "how do I…?") and part super-user (it does anything you can). Not a chatbot bolted on; it drives the real app through the same 55 tools you use.
And the learning goes both ways: you learn the craft by watching it work; it learns your preferences over time.
🎨
Generate anything, any engine.
20+ models across every modality.
▸
Image, video, voice, music, 3D, and talking-head avatars: each node picks the best engine for the job, and you can mix providers in one graph. Swap models as the field moves; you're never locked to one.
- Video: Sora, Luma Ray, Runway, and more; extend, modify, upscale, reframe, interpolate
- Image: Seedream, FLUX, GPT Image, Gemini, plus editing & grading
- Audio: voiceover in 80+ languages, music, and auto-captions
🎬
Direct like a filmmaker.
Shots, not prompts.
▸
Compose shots with real cinematic controls instead of wrestling a text box.
- Shot language: size, movement, lens, and angle as simple selectors
- Color & format: grade, LUTs, and framing from 2.39:1 to vertical
- Character lock: hold one identity across every shot
- Script breakdown: parse a screenplay into a shot sequence
👥
Made for teams.
Multiplayer, review, and dailies built in.
▸
Collaboration is first-class, not an afterthought, for a two-person remote team or a full facility.
- Live multiplayer: shared editing, presence, and remote cursors
- Review & dailies: scrub frame-by-frame and annotate together
- Comments: pin feedback right where the work lives
🔒
Provenance you can trust.
Every take reopens the workflow that made it.
▸
Save a Take and it captures the exact graph and parameters that produced it, so any take can reopen its workflow back onto the canvas months later. Promote the best version into a Studio Library that survives archive, offline, and delete. A single source of truth in a landscape that shifts weekly.
🧭
One place, start to finish.
An orchestration layer, not another tool to learn.
▸
Idea → takes → editor → delivery, without leaving the canvas. It doesn't replace your DCCs; your artists keep Maya, Houdini, and Nuke. It sits above them, coordinating the generative layer, and is designed to fit alongside your existing pipeline.
And each node is a creative building block (a Shot, a Character, a Grade, a Take), not low-level model machinery. You direct the result; the app runs the models underneath.